You are on page 1of 147

METALAB REVIEW

AI & CULTURE, BUSINESS AND SOCIETY

SPECIAL EDITION: STUDENT IDEAS THINK TANK

No 1, SEPTEMBER 2023
FOREWORD BY
METALAB DIRECTORS

We are delighted to present this first edition of the Metalab Review, as a special issue
assembled by the talented members of Metalab IDEAS, a student think tank.

At ESSEC Business School, our mission is to prepare future leaders – our students
of today – to comprehend and responsibly unleash the power of artificial intelligence
and data science to benefit both industry and society. We believe that realizing AI's
potential requires a wide ranging and concerted effort. The responsible adoption of
these fast-deploying technologies requires cultivating a new generation of professionals
who combine scientific acumen and organization savviness with ethical and societal
mindfulness. Our conviction is that our business school students must be equipped with
the multifaceted expertise needed to invent new AI solutions to the problems societies
and companies face today.

The insightful articles authored for this magazine by the student editors and analysts
of Metalab IDEAS embody the spirit of responsible AI leadership we seek to instill.
Analyzing topics ranging from enhancing human creativity through AI to the policy
implications of digital currencies, their contributions highlight the immense yet
unevenly distributed potential of data driven technologies. Weighing opportunities with
responsibilities, their work delves into the nuanced elements that shape an equitable
and sustainable AI future.

We believe this first edition will enrich debates on campuses and beyond by grounding
AI discourse in the contextual complexity of the real world. We hope it inspires readers,
especially those poised to assume responsibilities, to keep advocating for a human-
centric AI that ethically addresses social needs.

Congratulations to the talented 2023 IDEAS students cohort on their commendable work
to advance conscientious AI leadership!

THE METALAB CODIRECTORS


BENOIT BERGERET
PROF. GUILLAUME CHEVILLON
PROF. JULIEN MALAURENT

METALAB REVIEW ¦ 3
4 ¦ METALAB REVIEW
FOREWORD BY
ONEPOINT
In an era reshaped by technological breakthroughs and environmental upheavals, agility and
adaptability have never been more critical in crafting solutions for tomorrow's challenges and
in reassessing how our society envisions its future to better prepare for it. Overcoming the
obstacles ahead hinges on substantial investments in research and innovation.

Fueled by this steadfast belief, the onepoint Institute, our cross-disciplinary initiative,
orchestrates applied research initiatives that transform knowledge into practical and tangible
advancements.

The emergence of generative AI technologies paints a vast landscape of opportunities.


Together with its academic partners, the onepoint Institute aims to convert these possibilities
into meaningful value, underpinned by a sustainable and sovereign approach.

It was this vision and ambition that naturally led onepoint to establish a strategic partnership
with Metalab, a leading authority in AI and data from ESSEC, in 2021. This alliance offers a
unique platform for enriching collective insights on the rapidly growing adoption of these
technologies. Ongoing dialogues between our consultants and experts, and ESSEC researchers
and students enhance our service and consultancy offerings to onepoint’s clients. They also
fuel the creation of content that informs society at large on these transformational topics,
such as the student contributions from ESSEC’s think tank IDEAS, which form the core of this
inaugural Metalab Review, to which we are honored to contribute.

This symbiotic partnership also advances our mutual understanding of the present and
future impacts of AI and data. A collaborative approach serves as a catalyst for deepening
domain-specific knowledge. The enhanced public and civic visibility of both partners is further
amplified through joint events and communications. More than 600 students have engaged
in these events and hackathons, and the synergy between one of the world's leading business
schools and one point, the « the Architect of Transformations » also fosters a harmonious
balance between the academic and corporate realms. Our operational teams provide essential
on-the-ground perspectives to conduct research that aligns with the real-world corporate
challenges in responsible AI.

Without a doubt, this high-impact, structurally significant partnership aims to drive positive
change in the world. It focuses on understanding and applying new technologies wisely and
ensures that preserving our humanity remains a top priority.

MURIEL TOUATY
PARTNER EDUCATION AND INNOVATION, ONEPOINT

METALAB REVIEW ¦ 5
6 ¦ METALAB REVIEW
STUDENT
EDITORIAL

In 2020, ESSEC Business School pioneered the Metalab for data, technology and society,
a center aimed at creating dynamics around AI and disruptive technologies within ESSEC
programs - and beyond. As part of its activities, Metalab established the student-led IDEAS
think tank, a school-wide project to address the repercussions of technology on business
and management; from the increasing relevance of data analytics to the inescapable
reality of artificial intelligence. Now closing on our second year of existence, the think tank
has published numerous articles, developed privileged contacts with business partners,
conducted events, and produced a Podcast series—and we are only getting started!

With this first edition of the Metalab Review, we proudly present the editorial team’s work
produced over the 2022-23 academic year. We showcase a wide range of articles written
by the editors and analysts spanning multiple disciplines from the business school.
The articles correspond to each of the four key interest groups we have been leading: AI
Adoption, the Future of Sustainability & Society, Blockchain & Decentralization, and the
Future of Work & Business; and intermingled with these articles are snapshots of the
motivations behind some of the board’s participation in the Metalab.

This issue also features a series of the successful events organized by the team,
roundtables and poster sessions, as well as participation of Metalab IDEAS to other
relevant ESSEC initiatives. Indeed, in our second year, our team has endeavored to expand
beyond editorial publications. Such is the case for the release of Tech Tide: A Wave of
Ideas, a podcast series produced in conjunction with Tech ESSEC. It aims to shed light on
the impact of technology and AI developments on companies, society and the economy.

The publication of the magazine is a manifestation of the Metalab IDEAS team’s ambition,
hard work, and commitment to assuring the leaders of tomorrow are informed on the
technological advancements that will affect theirs and the lives of those they will lead. It
has not been easy, any member of the board will attest to that, but worthwhile? No doubt. It
is therefore the team’s greatest wish that not only will you be entertained by this issue, but
that you will learn from it.

IDEAS THINK TANK EDITORIAL BOARD

METALAB REVIEW ¦ 7
THE EDITORIAL
BOARD MEMBERS
NELBERTO NICHOLAS MICHELLE DIAZ
“SAM” QUINTO MASTER IN MANAGEMENT
PHD STUDENT IN THE STUDENT AT ESSEC BUSINESS
MANAGEMENT DEPARTMENT SCHOOL
AT ESSEC BUSINESS SCHOOL
I’m Michelle Diaz! I grew up
I’m Nelberto Nicholas “Sam” Quinto and I am a between Ireland and the Philippines, and graduated
doctoral student who does research on human-AI with a B.A. International Joint Honors in English
collaboration. I am particularly interested in how Literature & Information and Social Computing
“collaborating” with AI alters problem-solving, from University College Dublin, with a year-long
creative thinking, and group dynamics in innovation academic exchange at Waseda University in
contexts. Aside from studying AI, I also use it Tokyo, Japan. I’m currently a fourth-year Master in
to analyze large amounts of data from creative Management student. My work experiences are in
industries to uncover approaches to innovation that AI, startups, & entrepreneurship. And I’m currently
yield replicable outcomes when used by different doing a double degree in Global Management at the
individuals. My predilection for AI can be traced University of Queensland, Australia. My interests
to my work as a data science consultant at BNP are in sustainability, ethical issues in business and
Paribas’ Analytics Consulting team, my analytics technology, AI, and impact ventures. Moreover,
internship in Amazon, and my coursework on big in my downtime, I love watching films, tv shows,
data, machine learning, and AI strategy during my documentaries & I spend a significant amount of
Master’s in Management at ESSEC. Hence, within time reading and writing. In another life, I would
my work in the Metalab, I strive to distill insights and actually be in graduate school for creative writing.
thought-processes from these many experiences
to both co-create and curate content that drives the
frontiers of ongoing AI discourse. VANSHIKA SHARMA
MASTER IN DATA SCIENCES
& BUSINESS ANALYTICS AT
MARINA PELLET ESSEC BUSINESS SCHOOL &
MASTER IN DATA SCIENCES CENTRALESUPÉLEC
& BUSINESS ANALYTICS AT
ESSEC BUSINESS SCHOOL & I'm Vanshika Sharma. I grew up in New Delhi,
CENTRALESUPÉLEC India and did Bachelor's in Commerce (Honours)
from University of Delhi. Prior to joining the
I’m Marina Pellet, I’m French and Hungarian, and masters program at ESSEC Business School
I am a student in the DSBA Master's program and CentraleSupélec, I worked for two years as
created by Essec and CentraleSupelec. Before that, I an Associate in Quantitative Analytics and Data
studied in London at UCL and I obtained a bachelor Science at Gartner, Inc. I recently completed a
of Management Science with a specialization in Master in Data Sciences and Business Analytics and
Cybercrime. I speak four languages and I have started my end-of-studies internship at Amazon.
lived in six different countries. I love photography, I have always been intrigued by how transforming
cinema and reading. Also, I am very keen on sailing data into meaningful insights can lead to better
and kitesurfing. I am simultaneously excited and decision-making and innovation. My interests are
concerned about the increasing use of technology in exploring upcoming technology, its influence
by society, so I can’t wait to discover all aspects on various industries, and how it can be used
together! to promote sustainability. In my free time, I love
binge-watching thriller series, exploring different
cities, and discovering their gastronomy (especially
various vegan/vegetarian food options).

8 ¦ METALAB REVIEW
KUNAL PUROHIT PIETER JAN MOTMANS
MASTER IN MANAGEMENT AT MASTER IN DATA SCIENCES
ESSEC BUSINESS SCHOOL & BUSINESS ANALYTICS AT
ESSEC BUSINESS SCHOOL &
I am Kunal Purohit, originally from CENTRALESUPÉLEC
Udaipur, India. After engineering, I
worked in product in two different Tech companies I’m Pieter Jan Motmans, a beginning researcher
for two years and while pursuing my first year of in the exciting field of Reinforcement Learning
MIM at ESSEC, I am currently working as Data currently doing an internship at ESSEC Business
Analyst Apprentice at a startup in Paris and also School in Paris. I have been part of the Metalab
participating in Accenture SBA Chair. Reading about for two years now, first during my Master’s in Data
new technologies, writing blogs, playing Table- Sciences and Business Analytics and now during my
Tennis, listening to podcasts and watching movies internship. Throughout these years, I was constantly
are my interests. Fluent in 3 different Indian regional driven by my interest in studying the societal impact
languages and English, I am currently learning of different technologies. It has been enriching to
French. be a part of the think tank to progress the debate on
these topics, and I greatly enjoyed writing my own
articles, and helping analysts and fellow editors in
PAUL BÉDIER writing theirs.
MASTER IN DATA SCIENCES
& BUSINESS ANALYTICS AT
ESSEC BUSINESS SCHOOL & RINALYN PAGAO
CENTRALESUPÉLEC MSC IN SUSTAINABILITY
TRANSFORMATION AT ESSEC
My name is Paul Bédier, I am 26 years old and BUSINESS SCHOOL
French. I completed a bachelor in Economics &
English, before obtaining a MSc in Finance from My name is Rina. I completed
EDHEC Business School. I then worked in different my Global BBA at ESSEC through the Singapore-
financial firms and investment funds as an analyst. Paris track and graduated in 2020. After that, I
After that I got very interested in AI and Data spent 2.5 years as a financial analyst at large
Sciences so I joined the DSBA Master from which digital and marketing agencies. This allowed me to
I recently graduated. My interests include various have multicultural and international experiences,
things like movies, food, and mountain trekking. I am from my upbringing in the Philippines to living
also curious about other cultures and I like learning abroad, especially in Singapore, Paris, and
new languages (Japanese, Chinese), which is why I Barcelona over the past 6.5 years. I am currently a
went to Singapore to cap off my studies. Excited to pioneering master student in MSc in Sustainability
keep contributing to Metalab Ideas! Transformation 2022–2023 program at ESSEC,
Cergy Campus.
In the face of the global climate crisis and other
NEIL JR. ROMUALDEZ challenges, my main interests are at the core of
MASTER IN STRATEGY & sustainability, social innovation, and making a
MANAGEMENT OF positive influence in our society. I am especially
INTERNATIONAL BUSINESS curious about how technology interacts with green
and sustainable growth within the confines of our
Hi! I’m Neil Romualdez Jr, 23 years planetary boundaries. I am thrilled to be part of the
old. I’m currently taking the Master in Strategy & editorial team and to learn more!
Management of International Business (SMIB) under
the Managing Digital Transformation & Innovation
Track. Prior to this, I graduated with a bachelor’s With contributions from IDEAS think tank analysts:
degree in Mathematics. I had a short stint in trade
SEETHAL REDDY AMIR HUMZA
reconciliations for hedge funds before I worked on
public policy as an economist for the Philippine KAULURI - MASTER KHAN - MASTER
Ministry of Finance. I am deeply interested in how IN MANAGEMENT IN FINANCE.
emerging technologies and our dependence on
them can change the way the world interacts and
manages its resources. In my downtime, I read
books, watch movies, and play musical instruments.

METALAB REVIEW ¦ 9
TABLE OF
CONTENTS
AI & CREATIVITY

14 20 30
ARTIFICIAL WILL AI END THE AI CREATIVE
INTELLIGENCE: CREATIVITY? RENAISSANCE:
A BOOST UNPACKING THE THE IMPACT OF
HUMANITY BEHIND CO-CREATING WITH AI
TO HUMAN GENERATIVE AI ON CREATIVITY AND
CREATIVITY CO-CREATIONS THE FUTURE OF
WORK

SUSTAINABILITY & SOCIETY

40 48 58 66
THE PROGRESS ATTENTION CITIES OF THE FROM VIRTUAL
PARADOX: ECONOMY: FUTURE MODELS TO
CAN TECHNOLOGICAL YOU’RE NOT EVALUATING AI’S REAL-LIFE
ADVANCEMENTS POTENTIAL IN
COEXIST WITH
PAYING ENOUGH CLIMATE-RESILIENCY
IMPACT:
ATTENTION HOW DIGITAL TWIN
A DEGROWTH
AND AI ARE DRIVING
ECONOMY?
SUSTAINABILITY
EFFORTS

10 ¦ METALAB REVIEW
BLOCKCHAIN & DIGITALIZATION

76 86 96
DIGITAL DIGITAL CURRENCY: PATHWAY TO A
CURRENCY: THE THE OVERHAUL OF FAIR METAVERSE
SUBSTITUTION PUBLIC MONEY
WHY DO WE NEED
OF MONEY A CENTRAL BANK
DIGITAL CURRENCY
WHEN MONEY IS
ALREADY DIGITAL?

102 112
DIGITALIZATION BLOCKCHAIN:
OF TALENT THE ANSWER
ACQUISITION: TO INVESTORS’
MOVE FORWARD GREENWASHING
WITH CAGED WOES?
MACHINE

SAFETY & TRANSPARENCY

126 132 138


144
SOCIAL SCIENTISTS EXPLAINABLE AI: EXPLAINABLE AI:
IN THE QUEST FOR PART 1 - WHAT’S PART 2 - CAN WE
AI ALIGNMENT AT STAKE AND HAVE A STATUS
THE POTENTIAL OF
INTERDISCIPLINARY
WHEN DOES IT QUO WHERE AI IS
COLLABORATION IN AI MATTER? EXPLAINABLE? HIGHLIGHTS
SAFETY OF THE YEAR

METALAB REVIEW ¦ 11
12 ¦ METALAB REVIEW
CREATIVITY
AI &

METALAB REVIEW ¦ 13
ARTIFICIAL
INTELLIGENCE: A
BOOST TO HUMAN
CREATIVITY
By MICHELLE DIAZ
Master in Management Student at ESSEC Business School

of dermatitis on her hands and arms due to


WHAT IS CREATIVITY? the solutions used in the operating room.

N
Given William’s regard for Caroline, he made
othing seems more human than a request to the Goodyear Rubber Company
creativity—the trait that underpins to make thin gloves as an experiment. And
human innovation, from the arts to these became the rubber operating gloves
the sciences, and pretty much everything used today, thin and reliable. Coupled
in between. According to the Cambridge with hand hygiene, it significantly reduced
dictionary, creativity is defined as “the ability infections and deaths in the operating room.
to produce original and unusual ideas, or A simple innovation borne out of necessity
to make something new or imaginative.” and concern – a creative and consequential
However, physicists regard equations not only solution sparked by an understanding of
for their originality, merit, or consequence, but context and the human need to alleviate
also for their elegance. On the way in which emotional turmoil.
an equation can embody a truth. Ingenuity is
defined not only by the solution-like quality As it stands, artificial intelligence remains
it may have, but by the subjective emotion it incapable of the emotion that makes
can elicit. creativity human. It can neither comment on
the elegance of Einstein’s theory of special
One of the greatest love stories in medicine relativity nor attribute any solution it proffers
is that between William Steward Halsted, as stemming from pain, devotion, or the
considered to be the father of American desire to simply evoke emotion. But ask any
surgery, and Caroline Hampton Halsted, his AI specialist today, and they will heartily tell
scrub nurse. The story goes that around the you that AI already has a relationship with
winter of the 1890s, Caroline complained creativity, one that is beneficial to both parties

14 ¦ METALAB REVIEW
© Bench Accounting on Unsplash
and largely in service of human creativity. Not One of the most known and cited definitions
quite out to compete with human creativity. was given by Simonton in 2012. It follows the
Not yet anyway. criterion used by the United States Patent
Office to determine whether an invention
can come under patent protection. An idea
DEFINING CREATIVITY or invention must be original, useful, and
surprising to achieve either full, or partial
Over the years, creativity has been defined patent protection. Another famous definition
in multiple ways. All with the explicit aim of was used by Boden in 1998, in her article on
capturing and accounting for every facet creativity and its connection with artificial
of the concept. From breaking it down into intelligence. According to her, there are three
different types, to analysing its consequences, types of creativity, all involving the generation
and clearly defining the common thread of of novel ideas, and each type is capable of
what makes a certain idea, behaviour, or bringing about the emotion of surprise:
endeavour creative.

METALAB REVIEW ¦ 15
By categorising creativity in these three ways,
Boden was able to evaluate computer models
in accordance with the categories. She
additionally differentiated creative endeavours
Three Types of Creativity as either being:
1. COMBINATIONAL CREATIVITY
> Definition: When two new ideas share a TYPES OF CREATIVE
common inherent structure. ENDEAVOUR:

> Example: Poetic imagery and analogy. 1. PSYCHOLOGICAL


> Definition: New to creators
themselves.
2. EXPLORATORY CREATIVITY
> Example: When an author writes a
> Definition: Characterised by the creation story they’ve never written before.
of new ideas through the exploration of 2. HISTORICAL
structured conceptual spaces. > Definition: New to society.
> Example: Can include scientists, artists, > Example: Einstein’s theory of
and musicians who all learn the rules Relativity.
and styles of thinking involved in their
In this way, computer models could be
respective fields – superficially tweak
built to mimic the process by which
these established norms to achieve creative ideas could be generated.
something new.

3. TRANSFORMATIONAL CREATIVITY
DISTINCTIVELY HUMAN
> Definition: Involves the transformation of
CREATIVITY
the space itself, so that new structures However, the key component of whether an
can be generated that were before idea can be considered creatively successful
considered impossible – has the is the context in which it is made. Great poetry
capacity to bring about a shock, that is, does not simply follow a good rhyming scheme
ideas that, depending on the degree of but comments on the human condition in
transformation, may be more difficult to a way that resonates with people—which is
accept. to say, artificial intelligence when applied to
creativity must be directed. For instance, in
> Example: Something like humanity’s the fields of visual art or music composition,
once upon a time dream of landing a AI can be programmed to produce something
man on the moon. An idea seemingly that is transformational, in that it can create
so out there that it is shocking, but a something new and unexpected and be
vision easy enough to conjure that one deemed creative. But whether it can be
might deem it improbable instead of considered a work of art that is largely admired
impossible. by society is another question. It is imperative
for AI to understand, evaluate, and adjust
accordingly to contextual cues to compete with
this innately human aspect of creativity.

16 ¦ METALAB REVIEW
HOW AI ATTEMPTS It is almost akin to Doctor Halsted bringing
HUMAN CREATIVITY about the invention of thin medical gloves,
wherein the data provided to the algorithm
Artificial neural networks (ANN), a set of is the context of the good doctor’s concern
networks that are inspired by the human for his eventual wife, thus emphasising the
brain, try to address this question of human component in the creative endeavour.
recognition. ANN emulates and uses a Essentially, AI can create works of art or
reduced set of frameworks from biological demonstrate creativity by having access
neural systems. More specifically, these to data that provides human preferences.
models imitate the electrical activity of the This type of work effectively improves
human brain and nervous system. The upon human creativity, actively resulting in
ANN models are neurons in a complex and something useful like knowing the type of
nonlinear form. The neurons are connected music to play to encourage customers to stay
to each other by weighted links. There is a longer at a certain shop.
supervised procedure that comprises three
layers: input, hidden, and output. Now, instead
of simply asking the algorithm to create
EXAMPLE: ARTIFICIAL
something novel and unexpected, it uses an INTELLIGENCE CREATIVE
algorithm designed in accordance with the
human brain, with narrowed-down data sets
ADVERSARIAL NETWORK
chosen by human beings. (AICAN)
For example, in music, as in several creative Another aspect of creativity is the
fields, there exist melodies and patterns combination of transformational and
that can influence human emotions in psychological, a creation that manages to
certain ways. A study done in China using balance surprise with societal acceptance
AI capitalizes on this fact. Fundamentally, and be new to the algorithm itself. In this
music has three distinct parts: time, pitch, regard, there have also been advances. An
and texture. Using machine learning and astounding example would be the work
deep learning paired with facial recognition of computer scientist Ahmed Elgammal
and music, they created a model that could at Rutgers University in 2017. He used
assess the effects of certain types of music generative adversarial networks (GAN) that
on a customer’s behaviour. Essentially, they can do more than recognize existing images
tracked if the music played would affect of objects, but also generate novel images,
how long a customer would stay in a certain simply by inverting the "image coding/
store. And it did. It confirmed that digital recognition" procedure. He developed a new
transformation brought about by AI could be GAN known as the Artificial Intelligence
used as an environmental stimulus to affect Creative Adversarial Network (AICAN) which
customer behaviour. was able to not only judge its own work, but
also look for styles that have yet to exist. In
All this to say that one can train a model using essence, it could create.
a data set of the most known and beloved
musical pieces across the centuries and it will The key part of this endeavour was that
manage to render melodies that will resonate he trained the algorithm on a database of
with human beings. In this way, the creativity over 80,000 images from Western art in
involved is combinational, exploratory, and the periods between 1400 and 2000. The
historical. machine produced its own style by knowing

METALAB REVIEW ¦ 17
about the styles that already existed in that WHERE AI FAILS TO ACHIEVE
period. Now for the judgement: the AICAN HUMAN CREATIVITY
images were shown at the 2016 Basel fair
in competition with works by human artists, The opposite is required for great stories,
and the judges preferred the AICAN images. the narratives and writing that eventually
The judges collectively regarded the images become a part of the literary and cinematic
as creative. The case to be made, however, canon need to be attached to the human
is that the newly produced AICAN images experience. Indeed, for stories to work, they
followed the style of heavy abstraction, which must go beyond the technical elegance of
follows in line with the history of art. the writing, they must speak to some sort of
human truth. Even the most fantastical of
As such, an argument could be made that in stories, like the Harry Potter series or the Lord
this instance, the artwork fell just in line with of the Rings, revolve around certain universal
the creative categories of transformational truths: the experience of adolescence and the
and psychological, given the domain of ethical conundrum of what is good or bad. In
western art paintings. Regardless, the results essence, great stories must have a narrative
were impressive and indicate a positive for AI that humans can identify with.
in creativity. But while it is brilliant that AI can
create something entirely new and manage In this realm, there are two use cases of AI
to adhere to human preferences, it remains that have garnered a lot of attention.
that there was a human touch to the data:
the algorithm was created by humans, and
trained on a data set curated by humans.
TWO USE CASES:
1. COLLABORATIVE APPROACH:
Hence, it misses the purpose born of an FACILITATION OF HUMAN CREATIVITY
artist’s context that is usually involved in
the production of art. Moreover, western Most commonly used across different
art is a domain with a wider tolerance storytelling mediums. For instance, the
for astonishment and abstraction far incorporation of digital storytelling in
removed from human experience. A domain television, film, gaming, and the entertainment
wherein the purpose of art is allowed to be industry fall under this category. In essence,
less important than its ability to elicit an these AI interventions are to be incorporated
imaginative response from humans. into the genre of Choose Your Own Adventure
(CYOA), a type of story that gives its audience
a hand in the direction of the narrative.

A familiar example would be Netflix’s


Bandersnatch, a standalone episode of the
popular Black Mirror series, wherein the
audience is presented with different choices
at different points throughout the narrative
that affect the ending.
© Amauri Mejia on Unsplash

The use of AI in this genre is through


narrative generation systems, wherein the
system responds to user input and allows
for interactivity. The difference between

18 ¦ METALAB REVIEW
the traditional application of CYOA is that,
previously, the author had to create the
different predetermined storylines upon
writing the story. But the use of AI, will allow
computational algorithms to process and
produce audiovisual media more seamlessly
in service of the author’s vision.

This will allow for a wider variety of endings


with different nuances in place—making the
arc of the storyline more diverse and less
predictable for the viewers.

2. AI AUTHORS WORK THAT


RESONATES WITH SOCIETY:
CREATIVITY THAT PASSES AS HUMAN

This second approach is in line with the

© Rawpixel from Burst


combination of transformational and
psychological creativity; these forms of AI
try to be content creators themselves. Such
works have largely been experimental and
academic, but some projects have made it
into the public domain. KEY TAKEAWAY
Some examples include projects like Ross Distilled, the relationship between AI and
Goodwin and Oscar Sharpe’s AI screenwriter, creativity can be categorised in two ways:
along with Botnik Studios AI. The stories the empowerment of human creativity
these projects have developed are far from and the pursuit of a creativity that can
the coherent, emotional, and storytelling pass as human. Currently, AI has made
standards of the commercial entertainment more strides in the first category, working
industry. more so in partnership and with the
guidance of human creativity to achieve
But like AICAN images, they have managed new heights.
to intrigue enough to garner an audience.
Perhaps, with time, these stories will The latter category which proffers AI as
eventually evolve from curious, abstract, having the ability to compete with human
and compelling in their strangeness, to creativity by correctly accounting for
emotionally moving narratives that account emotion, nuance, and societal context,
for human context and emotion. remains in progress. The question is
how long. But for now, creatives all over
can capitalize on the ways in which AI is
another means for them to exercise their
creative juices and dazzle their target
audiences—to lean in, and treat AI as what
it currently is: a powerful boost to human
creativity.

METALAB REVIEW ¦ 19
WILL AI END
CREATIVITY?
UNPACKING THE HUMANITY BEHIND
GENERATIVE AI CO-CREATIONS

By NELBERTO NICHOLAS QUINTO


PhD student in the Management department at ESSEC Business School

R
ecent advancements in artificial in profound, uncharted ways. Amidst this
intelligence (AI) have prompted us to upheaval, a question emerges: Is creativity
ponder the fate of human creativity dying, and is AI the culprit?
and its relationship with technology. Present-
day AI platforms have demonstrated their
prowess in generating a diverse array of
WHAT EXACTLY IS
creative outputs. From fabricating tailor- CREATIVITY AND WHY IS
made digital imagery, crafting compelling
advertising phrases, drafting cogent essays,
IT IMPORTANT?
or even composing infectious tunes, the allure To untangle this provocative question,
of generative AI technologies is unmistakable. it is essential to probe into the essence
In 2022 alone, they have drawn a staggering of creativity—its core attributes and its
$2.654 billion in investments, marking a significance to society and organizations.
tenfold increase since 2020. As AI encroaches Creativity is defined as the generation
upon what was once considered a uniquely of ideas, solutions, or products that are
human domain, the ontological, economic, both novel and useful1. It is a multifaceted
and legal systems that underlie creative phenomenon, encapsulating not just the
production are consequently being redefined outcome or process of idea development,

1. Teresa M. Amabile et al., “Affect and Creativity at Work,” Administrative Science Quarterly 50, no. 3 (September 1, 2005):
367–403, https://doi.org/10.2189/asqu.2005.50.3.367.
Teresa M. Amabile, “The Social Psychology of Creativity: A Componential Conceptualization.,” Journal of Personality and
Social Psychology 45, no. 2 (August 1983): 357–76, https://doi.org/10.1037/0022-3514.45.2.357.

20 ¦ METALAB REVIEW
© Evgeniy Alyoshin on Unsplash
artistic creation, or problem-solving2, but also in the face of external changes and disruptive
the inherent ability of individuals to produce technologies, enhancing the likelihood of
the new and valuable. As the cornerstone success for individuals and organizations4.
of innovation, creativity transforms unique
ideas into tangible actions or products, which To grasp the essence of creativity’s central
in turn fuel technological advancements. components – novelty (newness, originality)
Within organizations, creative employees and usefulness (value, quality) – think of
are a catalyst for competitive advantage3. the last time you or someone around you
They contribute unique solutions to pressing has deemed something to be ‘creative.’ Was
challenges, resulting in strategies, business the value found in the end result or in the
processes, and design concepts that drive process that led to it? As will be revealed, this
success. Creativity also bolsters adaptability distinction is central to the debate of whether

2. Spencer Harrison et al., “The Turn toward Creative Work,” Academy of Management Collections 1, no. 1 (August 4,
2022), https://doi.org/10.5465/amc.2021.0003.
3. Richard Florida and Jim Goodnight, “Managing for Creativity,” Harvard Business Review, July 1, 2005, https://hbr.
org/2005/07/managing-for-creativity.
4. Martin Reeves and Mike Deimler, “Adaptability: The New Competitive Advantage,” Harvard Business Review, July 1,
2011, https://hbr.org/2011/07/adaptability-the-new-competitive-advantage.

METALAB REVIEW ¦ 21
© Yannick Pulver on Unsplash
the growing impact of AI will usher in a that defy convention, the emotions they
creative renaissance or its antithesis. stir within us, the curiosity they spark, and
the respite they offer from everyday life—all
HOW AI AFFECTS are markers of usefulness attributed to the
end product. AI-generated designs, leading
CREATIVITY: to a revolutionary product, align with this
PERSPECTIVES, perspective. From a managerial standpoint,
the value is not in the AI's process of creation
BENEFITS, AND but in the resultant artifact—its appeal in the
LIMITATIONS market, and its innovative design that caters
to and anticipates consumer needs.
THE PRODUCT PERSPECTIVE
From this product-centered perspective,
The traditional perspective of creativity places AI emerges as a potent tool for enhancing
emphasis on the final product. Think about our creativity. It produces innovative ideas
an evocative novel or an influential film. and forms that inspire human artists and
Their creativity is gauged by the audience's democratize access to creative production
experience with the completed work. The by making aesthetic prowess and innovative
unpredictability of the plot, the introduction combinations more accessible. Generative AI
of nonconforming characters, and the can be conceptualized as a new 'paintbrush'
use of groundbreaking narrative styles all for artists and creatives. Anyone can pick up
serve as sources of novelty. Meanwhile, the this 'brush' and create—be it poems, design
exhilaration experienced from narratives prototypes, or unique product concepts—

22 ¦ METALAB REVIEW
without necessarily needing the traditional, and a daring spirit to match the time and
extensive training or experience. For many, ingredient constraints set by Gordon Ramsay
this accessibility may very well foster and his colleagues. Their masterpiece
inclusivity, thereby encouraging a larger unfolds as an intricate dance of unexpected
population to engage in creative pursuits. ingredient pairings, artistic presentations,
It brings people who might never consider and innovative use of techniques, equipment,
performing artistic or innovative work into the and resources. Here, creativity is not solely
sphere of creative endeavors, exploring their defined by the outcome but by the culinary
human ingenuity as they co-create with AI. journey that unfurls, tantalizing taste buds
and challenging preconceived notions of what
AI LIMITATIONS IN CREATIVE can be done with limited resources. Similarly,
PRODUCTION improvisational theater values the performers'
on-the-spot thinking, synchronized action,
However, such democratization comes and narrative creation, as much as the final
with its own challenges and limitations. performance. The process is integral to
The novelty of the output generated heavily determining creativity. Even in organizational
relies on the creativity infused in prompts. settings, the sequential motions inherent
The usefulness of the final output also in problem-solving —brainstorming,
remains somewhat limited, especially when experimentation, and collaboration—
considering the current state of AI wherein are regarded as creative, alongside the
the quality of the generated paragraphs, effectiveness of their end solution.
images, or media may still fall short of those
produced by traditional professionals. There's HUMAN EXPERIENCE IS WHERE THE
often a substantial amount of trial and error REAL ISSUE LIES
involved in crafting the right prompts, as
well as significant post-processing once Despite known limitations, numerous
a satisfactory output is achieved. While instances exist where works co-created with
these are technical limitations that may be AI have been recognized for their creativity.
resolved as AI advances, there's another Last year, a digital image generated by AI
side to consider. As AI improves, so will the won a statewide art competition in Colorado.
benchmarks for what is considered quality, The judges affirmed that their decision to
useful, and new. Therefore, it's rare for a honor the contestant would have remained
few string of prompts to suffice in creating the same, regardless of prior knowledge of
something truly considered creative—that is, AI involvement in the artwork. In April of this
novel and valuable—for its time. Even with year, a song that employed AI to generate
significant advancements in AI capabilities melodies in the voices and styles of popular
over time, leading to minimal post-processing artists Drake and The Weeknd, went viral on
or prompt engineering needs, there's always social media. Titled 'Heart On My Sleeve', this
an inherent risk tied to tools that simplify track mimicked the stars' distinctive styles
processes. in verses ostensibly about pop sensation
Selena Gomez, who has romantic ties with
THE PROCESS PERSPECTIVE The Weeknd. The song was widely lauded for
its innovative stylistic fusion and its infectious
Another facet of creativity lies not in the final rhythm. Similarly, in corporate environments,
product, but in the process itself. Envision generative AI has been employed to aid
a Master Chef contestant, orchestrating a the development of innovative marketing
culinary symphony with a keen eye for novelty campaigns as well as fundamentally novel

METALAB REVIEW ¦ 23
and useful business strategies that managers comprehension of our intentions, history,
can build upon and implement. As such, AI or emotions– creations that leverage it are
capabilities continue to improve over time, it's seen to lack the depth and meaning that
becoming increasingly apparent that there's gives value to the creative process. Thus,
little pushback against the technology’s for the many patrons of the human touch,
ability to generate works that are novel and any process void of these elements loses
valuable, provided that creativity is interpreted their usefulness automatically. Singer and
from a product oriented perspective. Hence, songwriter Nick Cave eloquently expressed
the primary criticisms against AI's capability this very sentiment when reacting to a song
to produce or enhance creativity stem from written by AI chatbot GPT in 'his style':
adopting the process-centric one.
ARE AI CO-CREATIONS
REALLY DEVOID OF
“Songs arise out of suffering, by which HUMAN EXPERIENCE: A
I mean they are predicated upon the
complex, internal human struggle of CASE STUDY
creation and, well, as far as I know,
In December of 2022, an intriguing event
algorithms don’t feel. Data doesn’t
unfolded in the creative world. One, that
suffer. ChatGPT has no inner being,
invites reflection into the humanity behind AI-
it has been nowhere, it has endured co creations. Aamar Reshi, then a tech worker
nothing, it has not had the audacity in Silicon Valley, who had no prior experience
to reach beyond its limitations, and in graphic illustration or creative writing,
hence it doesn’t have the capacity for produced a children’s book in the span of 72
a shared transcendent experience, hours and sold 841 copies in Amazon over
as it has no limitations from which to a week. His secret sauce? OpenAI's Chat
transcend. ChatGPT’s melancholy GPT for the story narrative, and the AI art
role is that it is destined to imitate and generator, Midjourney, for the illustrations.
can never have an authentic human This well-documented accomplishment
experience, no matter how devalued represents an interesting shift in the creative
landscape, as tasks that traditionally
and inconsequential the human
demanded years of learning and skill honing
experience may in time become.”
were executed by Aamar in a mere three
days—and at virtually no financial cost.

Valuing process over product often comes A crucial aspect of this narrative is that
with an emphasis on the emotional Aamar circumvented the laborious grind
and sensory journey inherent in human of mastering digital design and storytelling
imagination and innovation. This viewpoint intricacies. His main inputs, as documented
suggests that the true value of creative on his Twitter feed5 were two-fold: textual
work is deeply entwined with the feelings prompts that generated the narratives and
and shared experiences manifested in its illustrations found in his book, and his clever
creation. As AI produces creative works use of these digital tools to construct a
based on learned patterns– without true coherent piece in a limited amount of time.

5. https://bit.ly/3LKyHY8

24 ¦ METALAB REVIEW
Trial and error was central to his creative As noted, his case also illustrates the
process, as he experimented with different democratization and streamlining of creative
prompts to craft his desired manuscript work. Even if he lacked formal training in
and illustrations. What used up most of his graphic design and had no prior experience
time was learning to maintain a consistent as an illustrator or storyteller, he was still
depiction of his protagonists, Alice and able to conceive an original, interesting
Sparkle on Midjourney. Yet, the conversion story that had (somewhat) coherent and
of his imagined images and plot ideas into meaningful illustrations. In essence, he came
tangible illustrations, descriptions, and up with creative solutions to turning his fresh
written stories required merely a few seconds ideas into a tangible product which found
of waiting time for each iteration. While significant demand, and hence– value, upon
Aamar's journey showcases the power of AI its release. And although the usefulness of
to democratize the creative process, it also his work is limited by the inconsistent and, at
highlights the pitfalls of such an approach. times un-detailed, illustrations –a limitation
His illustrations lacked consistency and which might diminish as AI technology
detail, likely a result of his over-reliance on AI continues to evolve and improve– the
tools and his lack of domain expertise in the evidence suggests that Aamar’s final product
digital arts. In the end, Sparkle, Alice's robot is in fact, creative - i.e. novel and useful.
companion, was rendered differently across Despite the innovative manner in which he
each page due to Midjourney's inability to developed his storybook however, the value
produce consistent illustrations from the in his process remains questionable to
same descriptions. Alice, too, was not spared those who value the human elements which
from inconsistencies, depicted with missing underlie creativity production. Therefore,
fingers on occasion. the query looms: was his work void of the
humanity that– arguably– gives meaning,
THE CREATIVITY AND HUMANITY depth, and usefulness to the creative
BEHIND AAMAR’S CO-CREATION WITH process?
AI
Not only did he “borrow” the feelings and
Aamar was innovative in his clever use emotions nested in the collective intelligence
of limited resources to solve a pressing within Chat GPT's memory to produce his
challenge: creating a children's book. plot, dialogue, and illustrations, but upon
Although he did not utilize traditional further contemplation, much of his process
artistic techniques or software, his co- actually involves a variety of inherently
creation with AI served as a conduit for human attributes. The excitement he felt
transmuting his imagination into tangible with his work, the love he felt for his niece
prototypes that allowed him to develop that sparked this project, the oscillations
a novel plotline through experimentation between euphoria and melancholy during
with various prompts. Confronted with the each imaginative iteration, and the victorious
issue of inconsistent images, Aamar applied sensation he savored as he reveled in
his human ingenuity to devise a clever his accomplished masterpiece were all
solution: describing Sparkle to be a robot profoundly human experiences and emotions.
that metamorphosed into various forms- a
creative fix that cohesively blended with his In the parlance of Nick Cave’s critique: his
narrative. blood, sweat, and guts coalesced into a
tangible expression of creativity– there at
his desk– a realization of a nebulous idea

METALAB REVIEW ¦ 25
that made its way to him. That is- within psyche? If the answer to some of these
the vast expanse of algorithmic and artistic questions are a yes, then it only shows that
unpredictability, his creative idea found him. Aamar’s experience was not an isolated case-
In all his humanity. And he made it come but is, in truth, a generalizable one.
to fruition in three relentless days. Thus,
Aamar’s case illustrates that the process of These aspects are integral to our humanity–
co-creation with generative AI can, and often our shared experience– and they would very
does, involve quintessential human aspects likely have been present in the generative AI
where innovation emerges from processes co-creation instances that you experienced
and ideas that are new and valuable. or envisioned. Even the journey to developing
your accumulated expertise in whatever
REFLECTIONS ON THE it is you do involves painstakingly human
GENERALIZABILITY OF THESE CASE ways in which you labored to sharpen them.
TAKE-AWAYS More people need to be reminded that the
knowledge and skill that you and I have
For those of you who have used AI to developed is critical to both evaluating and
generate works that were considered building upon the intermediate outputs
novel and highly valuable by your peers, produced by AI to produce works that are
organization, or audience-evaluators, perhaps actually novel and useful.
it is time to reflect and appreciate the inherent
humanity in your creative work. For those of Indeed, the route to producing work that
you who haven’t, you are invited to envision is genuinely innovative, with or without
what such a process should involve as you go AI's assistance, invariably involves a
through the next sentences. unique recombination of our lived human
experiences and abilities. As can be drawn
Think of instances where you brainstormed from your personal cases, and that of Aamar
ideas or developed potential solutions Reshi’s, creators often employ a range of
through a conversation with Chat GPT. tools, techniques, and information sources
Maybe there were times when you involved to generate value and endure iterative
generative AI in creating a business proposal, procedures that are seldom smooth and free
an artistic illustration, a musical piece; or of emotional dynamics. As such, so long as
perhaps a scientific publication, an innovative humans remain in the production loop, any
product, or even a clever recommendation creative process that leverages generative AI
that was, in fact, considered original and still retains significant traces of our shared
beneficial. Did your creative journey involve experience and human nature.
the emotional roller coaster of highs and
lows? The frustration felt when ideas fail? The
iterative cycles where you leveraged multiple
WHEN DOES CO-
tools and information sources in inventive CREATING WITH AI HARM
ways to achieve a part of your desired
outcome?
CREATIVITY THEN?
As with all tools that streamline processes,
Did you harness feelings and intuitions there is always a danger that individuals
from your gut? Did you rely on tacit might over-rely on them, progressively
heuristics and knowledge formed from depending less on their abilities and unique
years of accumulated experience and social value contributions. In this case, such a
interactions deeply embedded in your human trend could lead to a reduced cultivation of

26 ¦ METALAB REVIEW
both individual creativity and the essential CONCLUSION
skills relevant to creative production—be it
mastering the use of a paintbrush, crafting It is imperative that individuals learn to
a riveting narrative for a target audience, co-create with generative AI tools both
evaluating the aesthetics of a presentation responsibly and proficiently. The outlined risks
deck, or steering a company brainstorming to our creativity demands an understanding
session. As AI takes up an increasing part of that these technologies are not replacements
the creative process, our aptitude for ideation, for our own cognitive ability and domain
improvisation, and critical thinking may begin expertise. Neither are they supposed to be
to fade. used as quick-fix mechanisms for producing
content (e.g. essays, images, or sounds) with
Similarly, when individuals overestimate scant surprise or quality appeal. Instead,
AI’s ability to generate value, there is also a individuals should engage with these tools
tendency to bypass activities that cultivate in ways that stimulate their critical thinking,
our ability to generate the novel and useful. challenge them to push the boundaries of
Examples include discussing our ideas with convention, and inspire them to conceive
our colleagues (as opposed to just chatting ideas, products, artworks, or business
with an LLM about them), engaging in processes that are not only distinct from
exploratory learning, or improving the depth predecessors but also more efficient, value-
of our specific professional expertise. The enhancing, and superior in quality. And all of
ease of producing AI-generated outputs that, this begins with understanding that the path
at first glance, might seem highly innovative to truly creative co-creation with generative
and valuable, might also cause a drop in AI is a painstaking, iterative, and exploratory
motivation to engage in creative work - i.e. process that requires novel ways to combine
the question: 'If AI can generate creative our cumulative human experiences,
works with relative ease, why should I expend imaginative capacity, and knowledge with the
the effort?' People need to understand that current capabilities of these technologies.
the path is often an arduous, emotionally
tumultuous one, and cases where new and Such a paradigm shift necessitates
useful products are developed with relative substantial investment from organizations,
ease are very much the exception rather educational institutions, and even
than the norm. Moreover, projecting into the governments in programs designed to enable
future, if AI does advance to the extent that and encourage these exploratory practices.
it eliminates the need for substantial post- Experimentation, after all, benefits from
processing work, several prompt iterations, guidance and often comes at the cost of not
inventive ways to combine different tools getting traditional work or schooling done. But
and knowledge inputs to produce a novel this should not be a cause of worry. We have
and useful work– and well, all the emotions already witnessed select groups of individuals
that come with these– would that signal and organizations leveraging generative AI
the demise of creativity? It definitely is a in new and useful ways that showcase their
possibility. If co-creating with generative AI unlocked creative potential. Business news
results in fewer problems that demand new feeds and popular technology sites such as
and impactful solutions, then creativity may TechCrunch, Product Hunt, and VentureBeat
diminish simply because we would need less are filled with a rich variety of new use
of it. But that is, by all means, a good problem cases each week. Hence, it is conceivable
to have. that support for such endeavors are likely
to come. And, as history has demonstrated,

METALAB REVIEW ¦ 27
28 ¦ METALAB REVIEW
© Rod Long on Unsplash
when the innovative use of AI enables select unique, and feasible to accomplish within
groups of people or organizations to excel in a given period of time. The sequel to this
their niche, gain a competitive advantage, or article attempts to summarize all of these
simply overcome their peers, then others will other concerns, building on the evidence
emulate their strategies– fueling creativity and logical arguments presented in this first
and perpetuating the cycle of innovation. part to more robustly answer the tantalizing
question behind AI’s impact on creativity.
Will AI spell creativity’s demise, or could it Importantly, this second installment will
instead ignite a renaissance of sorts? If the underscore the necessity to shift our focus.
concern lies in losing the humanity behind The imminent issue may not be a potential
the creative process, then, so long as humans dwindling of our ingenuity, but rather a
are involved in the creation process, and that constellation of critical factors surrounding
problems and needs that require novel and the perception and common understanding
useful solutions, approaches, or products regarding what it takes to co-create ideas,
continue to exist, then the more optimistic processes, or products with AI that are truly
of the two scenarios is likely to materialize. novel and useful. There's also a pressing
In other words, AI may well just revolutionize need to consider the protective measures
creativity for the better. extended to workers who are at the interface
of AI's influence. For instance, while AI may
There are, however, a few other concerns not necessarily stifle overall creativity, it
that need to be accounted for, such as the holds the potential to extinguish the arts if
lack of acceptance or understanding behind considerations around the valuation of artistic
the value that these generative technologies capabilities and the protection of artists'
bring. Particularly relevant as well is how intellectual property are not meticulously
increasing AI adoption inevitably creates examined. These will all be touched upon in
higher standards for what is valuable, truly the sequel of this two-part article series.

METALAB REVIEW ¦ 29
THE AI CREATIVE
RENAISSANCE:
THE IMPACT OF CO-CREATING WITH AI
ON CREATIVITY AND THE FUTURE OF
WORK

By NELBERTO NICHOLAS QUINTO


PhD student in the Management department at ESSEC Business School

A
I's march into the realm of creativity Transitioning from the debate concerning
has been teeming with trepidation, the retention of humanistic elements in AI
stemming from fears that increasingly co-creation – elements that imbue depth,
sophisticated machines might eclipse our meaning, and value to the creative process,
ingenuity and devalue the human experience. as discussed in this article’s predecessor,
Dissecting the intricacies of co-creating this installment aims to elaborate more
works that are deemed creative– i.e. deemed on the benefits, limitations, and long-term
both novel and useful – with generative AI in challenges to creativity that arise from
the first installment of this two-piece article this generative process. The question of
series however reveals an inherently human whether AI could potentially spell the demise
process. A dance that involves iterative of creativity in the long-term is ultimately
bouts of innovative experimentation, unique addressed, followed by a conclusion that
amalgamations of tools and ideas, and a steers the discourse towards identifying more
critically overlooked component – emotion. pressing concerns about AI that will require
The realization of these intensely human much of our ability to provide novel and useful
aspects offer a distinctly more hopeful solutions.
perspective, contingent upon us remaining
actively involved in the process.

30 ¦ METALAB REVIEW
© Matthew Henry from Burst
RECAPPING THE streamlines and democratizes the process
CREATIVE BENEFITS TO of creative production. By curtailing the
costs related to skill, talent, and equipment
LEVERAGING AI (or software), AI facilitates the genesis of
valuable, new creations – be they illustrations,
Creativity is understood as the generation poems, marketing slogans, or strategic
of ideas, processes, or products that are proposals. This empowerment invites a
both novel and useful1. When understood wider demographic to unlock their inherent
as an ability to produce works that meet creative potential and engage in the human-
this two-fold criteria of novelty (uniqueness, AI symphony of creation. Take, for instance,
originality) and usefulness (value, quality), the case of Aamar Reshi discussed in detail
both the outcomes and processes inherent in the first part of this article series. His
in generating them become relevant. One imaginative journey to crafting a children's
of the most widely observed benefits of storybook in 3 days without any prior
co-creating with generative AI is that it experience in illustration or creative writing

1. Teresa M. Amabile et al., “Affect and Creativity at Work,” Administrative Science Quarterly 50, no. 3 (September 1, 2005):
367–403, https://doi.org/10.2189/asqu.2005.50.3.367.
Teresa M. Amabile, “The Social Psychology of Creativity: A Componential Conceptualization.,” Journal of Personality and
Social Psychology 45, no. 2 (August 1983): 357–76, https://doi.org/10.1037/0022-3514.45.2.357.

METALAB REVIEW ¦ 31
© Charles Deluvio on Unsplash
was made possible by AI's ability to translate the mundane and predictable, this in fact
his motivation into captivating illustrations offers a crucial advantage. The immediate
and engaging storylines. availability of a broad spectrum of highly
conventional ideas allows individuals and
AI also significantly lightens the cognitive teams to swiftly dispatch with the obvious,
load associated with conceiving original learn from these iterations, and refocus
solutions to intricate, pressing problems. For their cognitive prowess. In the context of
instance, imagine a marketing team trying prototyping, generative AI tools have emerged
to brainstorm a new advertising campaign. as a catalyst for quickly transforming abstract
Instead of spending hours coming up with concepts in our heads into tangible, useful
basic ideas, a large language model (LLM) drafts. They shoulder the cognitive and labor-
such as Chat GPT could generate a list of intensive burden of the initial groundwork,
them within minutes. The team could then generating the first iterations swiftly. As a
immediately start discussing which of these result, just as in brainstorming, teams can re-
listed ideas are worth developing further. They allocate their mental energy and time towards
could also combine different AI-suggested more creative tasks, including those where
concepts to create something unique and human expertise still surpasses AI.
impactful. In this way, AI allows for more
effective use of the team's expertise and
creative thinking. And while much of the
content produced by LLMs tends towards

32 ¦ METALAB REVIEW
AI’S IMPACT ON deepen their creative explorations, finding
QUALITY STANDARDS fresh ways to collaborate with AI to generate
art that remains innovative and valuable
PERPETUATES amidst the evolving landscape. This may
CREATIVITY involve countless experimentations with
prompt patterns or combining different
As we hurtle into a future fraught with tools, systems, or methods to process works
increasingly complex challenges requiring generated by AI.
novel solutions, it's clear that both creativity
and generative AI’s role in the creative Meanwhile, in organizational settings,
process is here to stay. While individuals particularly when teams make use of
will continue to extract value from AI, the Large Language Models (LLM), such as
benchmark for what constitutes quality OpenAI’s ChatGPT or Google’s Bard, to
and usefulness will concurrently increase. streamline brainstorming sessions, catalyze
When productive expectations rise, fostering idea generation, and decode complex
creativity and innovation to meet these problems– the interplay between creativity
heightened standards will always be a need, and AI unfolds in the same manner as in
and often, a priority. While the influx of novel the arts. Consider a tech start-up using
combinations of information and ideas might GPT technology to augment their product
make the task of generating something truly design brainstorming. The AI solutions
different and unique with AI progressively might generate new perspectives on user
challenging as well, it will undeniably interface design, providing inspirations that
necessitate increasingly creative methods the team hadn't previously considered.
to continuously derive value from current Companies that successfully integrate such
tools and systems. Thus, regardless of how technologies into their workflow stand to
sophisticated AI systems become, creativity gain a competitive edge, as they can alleviate
will always have a place within organizations the cognitive burden inherent in creative
and society. tasks and propel their innovation teams
into uncharted territories. However, as AI-
For instance, in producing digital art, some assisted approaches become commonplace,
graphic designers are already leveraging AI the novelty of these processes diminishes.
image generators such as Midjourney and As more and more employees, teams, and
Stable Diffusion to simulate prototypes of organizations become accustomed to
their ideas, draw inspiration from different harvesting the creative benefits of AI across a
styles, or ultimately create distinct artworks spectrum of tasks, the standard for what can
that require minimal post-processing. At first, be achieved within a certain timeframe and
such use of AI was deemed innovative, and resource set escalates exponentially.
there have been many cases where fresh
and valuable works have been produced In response, organizations would need to
through such a process. But as leveraging AI persistently cultivate creativity to extract
capabilities have become more widespread, further value from AI. They might deploy
the novelty of such techniques have declined, additional algorithmic tools to discern
and the average uniqueness of generated patterns from past AI-laden brainstorming
works have suffered as well. Consequently, sessions and predict potential hurdles.
the standards for what is considered useful For example, a company could use AI to
have risen as more people have realized the analyze both previous meeting notes and
value of current AI tools. Artists must then past AI suggestions given various employee-

METALAB REVIEW ¦ 33
© Vonecia Carswell on Unsplash
generated GPT prompt patterns. This would standards for novelty and value that
allow individuals and teams to leverage innovation and competition propel. However,
insights derived from such data analyses to when novel processes or new best practices
consequently modify their use of generative are found to be creative at a given point in
technologies for idea development and time, more and more companies are likely to
prototype-building. Such a solution could also find value in them. This implies that standards
lead to routinized processes and systems will rise again and organizations would need
that facilitate extracting more value from to search for more ingenious processes that
novel AI suggestions in order to suit their leverage existing human and technological
unique problem contexts. For example, a capital, fostering creativity and perpetuating
marketing team could develop protocols for its need.
how to synthesize campaign ideas developed
through multiple rounds of brainstorming
with generative AI, across several meeting
THREATS TO
groups. They could also devise a structured CREATIVITY: UNDER-
process that involves integrating these
insights with those uncovered by social
APPRECIATION AND
media trends, competitor campaigns, and OVER-RELIANCE ON AI
customer behavior data that predictive AI
models concurrently deliver. Instead of pondering the supposed demise
of creativity, perhaps the real issue at hand
In such instances, creative production is the skepticism and under-appreciation of
necessitates innovation in processes where the immense value generative AI brings to
different individuals and AI systems come the table. It is unsettling to see how quickly
together to meet the new, heightened some individuals abandon the use of these

34 ¦ METALAB REVIEW
technologies, swayed by a few erroneous overstated. Identifying the complementary
outputs, unsuccessful prompts, or the intersections between AI and our innovative
cynicism of others. On the flip side, there capacity enables us to effectively harness
exists an equally troubling group – those their combined potential. And this begins
over-reliant on AI to the extent that they by ensuring that we remain determined to
forsake nurturing their own critical thinking, cultivate the requisite knowledge, expertise,
domain expertise, and breadth of knowledge. and capabilities to credibly evaluate the
This excessive dependency threatens to usefulness, appropriateness, and credibility of
hollow out human innovation and intuition, AI generated outputs.
reducing our contributions to mere button-
pressing reactions to whimsical needs. The In this dynamic panorama of evolving AI
remedy to both of these conundrums lies in systems and expanding human capabilities,
fostering a comprehensive understanding of the value of experimentation in demarcating
AI's capabilities and limitations among the the ever-changing boundaries of both AI and
broader population, as discussed in more our personal abilities becomes paramount.
detail in the first piece of this two-part series. In particular, this knowledge helps illuminate
our understanding of what AI can and cannot
There is simply so much value-creation lost do for us, and what we can achieve more
when people do not simply understand the efficiently with or without these technologies,
usefulness of generative tools that can rapidly for any given moment in time. It also enables
produce creative drafts and prototypes ; all us to identify which domain-specific skills
while harnessing centuries of accumulated and new capabilities we should invest our
human knowledge to yield accurate, factual time in developing and refining. And as
results in at least 9 out of 10 instances. As we navigate this blurred frontier, the role
such, there is a pressing need to educate of continuous learning, engagement, and
individuals about applying even basic development of our unique creative abilities
generative AI capabilities to the multitude of vis-a-vis AI becomes ever more critical.
use cases in their personal and professional
lives. This requires promoting practices that While the “death of creativity” is far from
encourage persistent experimentation with AI, imminent, the reshaping business and
while fully acknowledging its current bounds societal terrain calls for both prudence
and how it can complement our unique and immediate action. As AI emboldens
strengths. individuals and organizations to leverage
technology to augment their unique
THE SOLUTION: capabilities and cumulative expertise,
navigating this evolving landscape thus calls
CONTINUOUS us to reconsider our part in the redefined
EXPLORATION OF AI- creative epoch. Are we disrupting, merely
keeping up, or falling behind? Survival
HUMAN BOUNDARY in competitive environments, after all,
CAPABILITIES necessitates exploiting the opportunities
emerging technologies offer, and we should
AI will continue to improve, and so too therefore set aside time in our schedules
will our abilities, skills, and talents. In this to explore ways we can achieve this at a
constant cycle of progression, the necessity personal level. Again, this begins by carefully
of understanding the evolving capacities discerning what both we and AI uniquely
of AI and human creativity cannot be excel in, and identifying potent synergies that

METALAB REVIEW ¦ 35
allow us to balance each other's strengths reduced remuneration for graphic designers,
and weaknesses for our specific trade and composers, and other artists. This, even
tasks. Whether you are a marketer, musician, before a novice utilizing generative AI can
illustrator, business owner, or a corporate genuinely outperform seasoned creators.
planner, understanding how emerging Such devaluation can cause both current
technologies affect the value and novelty and aspiring artists to neglect nurturing
you produce at any given point in time is their unique aesthetic skills and specialized
paramount to continued success. Stop. Think. capabilities, which could be detrimental to the
Reflect. What critical inputs, processes, and future of creativity. After all, truly harnessing
skills does AI need from you to deliver in the power of AI for creative endeavors
order for you to generate ideas, products, and requires not just knowing its technological
systems that are new and valuable? What boundaries, but also having a robust
are you doing to know or develop this? Your foundation of one’s specialized domain
answers to these questions may well define and the aesthetic sensibility needed to both
your survival amidst AI’s creative renaissance. recognize and conjure quality art.

BEYOND THE ISSUE OF Furthermore, we need to tread carefully on


the path of intellectual property (IP) rights.
CREATIVITY: WHERE Artists may feel aggrieved as their creative
MORE PRESSING genius– their unique output– is utilized to
train AI systems without due recompense or
CHALLENGES LIE consent. This could make it facile for others
to mimic their distinctive style, enabling
While this thought exploration attempts these individuals to build on years of the
to address whether AI would extinguish artists’ painstaking dedication without due
the flame of human creativity, it inevitably permission or acknowledgement. If we fail
leaves a myriad of related issues unexplored. to establish robust mechanisms for suitable
Indeed, while AI, with its prodigious generative compensation, recognition, and consent
capacities, promises to herald a creative for creative IP, we risk discouraging artists
renaissance—exploring uncharted territories from trusting and reaping the benefits of AI.
of ideas and unveiling transformative Worse, we risk a decline in traditional artistic
solutions to pressing issues—we must also pursuits, which form the bedrock of cultural
recognize the potential detrimental impact heritage and individual expression. On a less
it could have on our artistic workforce. As immediate yet significant impact, we risk
discussed, the process of co-creating with losing valuable creative contributions that
generative AI still necessitates an intricate, future generations could build upon. If we
often emotionally charged, and iterative deter a whole generation from pursuing arts
human endeavor. It involves the careful due to lack of passion or adequate incentive,
orchestration of past human experiences to we lose more than just individual livelihoods.
weave into the fabric of our creations. The vibrancy of our cultural tapestry and
our collective ability to innovate are at stake.
Yet a recurrent threat that we are treading is It is therefore vital that we safeguard the
the dangerous line of devaluing the authentic livelihoods and IP of artists to prevent cultural
human touch; that is– the artisanal signature impoverishment.
underlying much of the creative work we
know. Companies may start to undervalue Moreover, while AI undoubtedly creates
the finesse of artistic capability, leading to new job opportunities and uncovers novel

36 ¦ METALAB REVIEW
synergies between human and machine to
drive value and discover novel solutions to
critical problems, the rapid pace at which
AI is reshaping the job market is cause for
concern. The velocity of job displacement
could potentially outpace that of job creation
and skill development across a wide-array of
fields. Its reach already extends to every facet
of work, life, and the economy, sometimes
unpredictably. Such an imbalance could
endanger a substantial segment of the
workforce, along with their dependents.

While AI is not poised to extinguish the flame


of creativity, it could expedite the ebb of
certain industries and livelihoods faster than
it rebuilds or expands them. This necessitates
an inventive approach to temper potential
fallout. Hence, the real merit of AI lies not
only in catalyzing and amplifying human
creativity to forge new solutions to high-value
issues, but also in innovating answers to
the challenges it itself surfaces– from that
of artistic preservation, intellectual property
rights, overly rapid job displacement, and
beyond.

METALAB REVIEW ¦ 37
38 ¦ METALAB REVIEW
SUSTAINABILITY

METALAB REVIEW ¦ 39
& SOCIETY
THE PROGRESS
PARADOX:
CAN TECHNOLOGICAL ADVANCEMENTS
COEXIST WITH A DEGROWTH
ECONOMY?

By RINALYN PAGAO
MSc in Sustainability Transformation at ESSEC Business School

D
o you ever feel like the pursuit of In this article, we will explore the intersection
economic growth has taken us down of degrowth and technological advancement,
a dangerous path, where the health of and critically analyze their intertwined potential
the planet and its inhabitants are sacrificed for implications in a global context.
the sake of financial gain? If you are nodding
your head in agreement, then continue reading
because we are about to dive into a fascinating
DEFINING DEGROWTH
topic: degrowth. Degrowth is a concept that challenges the
idea that economic growth is necessary
In our current global society, the growth-based for human development and wellbeing. It
economic model has dominated for too long, is based on the principles of sufficiency,
leading to environmental degradation and social resilience, and equity, and calls for a reduction
inequality1. Fortunately, an alternative concept of in production and consumption levels to
degrowth has emerged, promoting a supposedly achieve a sustainable and equitable society.
sustainable and more equitable way of life. 2
While the concept of degrowth has its roots
However it does not surprise us that regardless, in the 1970s, it has gained more attention
technological advancement and digitalization in recent years as a response to the current
continue to shape and transform our world in ecological and social crises, however it also
profound ways that could seem inevitable. raises some criticisms as such that degrowth
is too radical and impractical, while others

1. Hickel, J. (2020). Less is More: How Degrowth Will Save the World. Penguin Random House.
2. Degrowth Movement. (2023). Degrowth. Retrieved from https://degrowth.info/degrowth

40 ¦ METALAB REVIEW
© Anastasiya Badun on Unsplash
claim that it overlooks the needs of those in environmental degradation, social inequality,
poverty3. and job displacement. The debate around
technological determinism, the idea that
TECHNOLOGICAL technology shapes society, versus social
constructivism, the idea that society shapes
ADVANCEMENT AND ITS technology, is still ongoing4. Both views
IMPACTS have their merits, and the truth likely lies
somewhere in between as per what the
Technological advancement has undoubtedly scholars suggest. While this debate is not the
brought about many positive changes to focus of this article it is still important for us
our world, from medical advancements to consider both perspectives when thinking
to increased connectivity and access to about the relationship between technology and
information. However, it has also been society, as this will help us to develop a more
associated with negative impacts such as nuanced understanding of this complex issue.

3. Victor, P.A. (2019). Growth, degrowth, and climate change: A scenario analysis. Ecological Economics. Retrieved from
https://www.ledevoir.com/documents/pdf/victor_growth.pdf
4. Jonas Hallstrom (2020). Embodying the past, designing the future: technological determinism reconsidered in
technology education. International Journal of Technology and Design Education, Retrieved from

METALAB REVIEW ¦ 41
© Harper-Sunday on Unsplash
EXPLORING THE However, it might be hard to monitor
INTERSECTION: IS innovation progress, specifically the major
challenge to be able to quantify or allocate
TECHNOLOGICAL inputs to policy goals like more affordable
INNOVATION AND technology, industrial change, and economic
growth. Nevertheless, a number of measures,
DEGROWTH POSSIBLE? including (1) funding and (2) patenting, might
provide insight into clean energy innovation
Technological advancements per se in the on a worldwide scale. More comprehensive
sustainability sector are the renewable energy, sets of measurements are being created by
clean transportation, energy efficiencies, and some governments in order to find and share
carbon capture and utilisation storage (CCUS) best practices.
- can these potentially be a powerful tool to
degrowth our economy? • FUNDING

The latest report done by the International This trend (see: IEA public energy technology
Energy Agency (IEA) on its Clean Innovation R&D and demonstration spending by
Report: Global status of clean energy technology, 1977-2019) tells us that the
innovation in 2020 5 acknowledged that government funding on energy research
technological innovation is crucial for and development increased by 3% in 2019
addressing climate change and achieving the to USD 30 billion globally, with over 80% of
goals of energy policy, such as expanding that money going toward low-carbon energy
access to energy and lowering air pollution. technology. Specifically in China, the low-

5. IEA (2020), Clean Energy Innovation, IEA, Paris https://www.iea.org/reports/clean-energy-innovation, License: CC BY


4.0

42 ¦ METALAB REVIEW
carbon component of energy research and that how could it intersect with the degrowth
development increased by 10% in 2019, principle? In the publication done by Hickel
with significant increases in research and and Kallis (2019) they investigated and
development for energy efficiency and tried to answer: is green growth possible?6
hydrogen in particular. Spending on public According to the green growth theory,
energy R&D increased by 7% in both Europe sustained economic expansion is consistent
and the United States, exceeding the previous with our planet’s ecology because technology
year's trend. change and substitution will allow us to
completely disconnect GDP growth from
• PATENTING resource use and carbon emissions. This
claim is increasingly assumed in national and
On the other hand, in terms of patenting international policy, including the Sustainable
since 2011 (see: Issuance of patents for Development Goals. Green growth has
low-carbon energy technologies in selected developed as a major policy response to
countries and regions, 2000-2016), the climate change and ecological disintegration.
quantity of patents submitted for low- However, empirical evidence on resource
carbon energy technologies has sharply utilization and carbon emissions does not
decreased after a decade of rapid increase. support green growth theory. After reviewing
Patents capture part of the intermediate relevant papers on historical trends and
R&D outputs, a percentage of which will be model-based projections, this research study
converted into commercial products, and so conducted reveals that:
offer a window into the research operations
that are producing new information with a “(1) there is no empirical evidence that
view toward its perceived commercial value. absolute decoupling from resource use
However, general patenting trends offer can be achieved on a global scale against a
us some insightful data on the scope and background of continued economic growth,
direction of clean energy innovation. and (2) absolute decoupling from carbon
emissions is highly unlikely to be achieved at
The drop in renewable energy patenting a rate rapid enough to prevent global warming
activity from roughly 2011 may be due to over 1.5°C or 2°C, even under optimistic policy
the maturity of some technologies. The conditions. We conclude that green growth
dominance of existing solar PV, bioethanol, is likely to be a misguided objective, and that
and wind technologies may discourage policymakers need to look toward alternative
researchers from striving to improve them strategies.”
and enter the markets in Europe, Japan,
and the United States. Moreover, renewable Hence, even looking at the intersection
energy patenting continues to be more active between the myriad of technological
than it was before approximately 2007, and innovations we have reached today, especially
Li-ion battery patenting, in particular, is the energy transition needed to net zero
expanding (EPO and IEA, 2020). emissions by 2050 commitment, the promise
of a sustainable future for our current and for
Now that we have built a solid foundation the next generation is still not guaranteed.
on what constitutes a green technological
innovation, what begs a question here is

6. Hickel, J. and Kallis, G., 2019, ‘Is Green Growth Possible?,’ New Political Economy, DOI:
10.1080/13563467.2019.1598964

METALAB REVIEW ¦ 43
44 ¦ METALAB REVIEW
© Austrian National Library on Unsplash
PARADOXES OF facilitate the transition towards renewable
TECHNOLOGICAL energy and more efficient production
processes, it is unclear whether they can
ADVANCEMENT AND THE enable a complete decoupling of growth
DEGROWTH from resource consumption. According to
the Global Footprint Network, achieving
In this section, let’s dig deeper on the four sustainable resource use would require a
main paradoxes that hinders technological reduction in resource consumption by roughly
advancement toward sustainability and 75% in high-income countries and 90% in
degrowth: (1) decoupling growth from middle-income countries8. This would be
resource consumption, (2) e-waste challenging to achieve through technological
generation, (3) conflicting power and politics, advancements alone. In addition, the findings
and (4) cultural and social implications. in the recent study stated that while energy
and material efficiency improvements have
1. DECOUPLING GROWTH FROM reduced the amount of resources required
RESOURCE CONSUMPTION for each unit of GDP, the overall consumption
of resources has increased due to economic
Degrowth advocates for a reduction in growth.
production and consumption levels to achieve
a sustainable and equitable society7. However, 2. E-WASTE GENERATION
this poses a challenge in a world where
economic growth is closely tied to resource According to the United Nations The Global
consumption. This means that if we want to E-waste Monitor 2020, in 2019 there was
reduce production and consumption levels, a record high of 53.6 million metric tonnes
we will also need to reduce economic growth. (Mt) of electronic waste generated globally,
This is a challenge because economic which is an increase of 21% in just five years9.
growth is seen as essential for prosperity E-waste is a term used to describe discarded
by many people. There are a number of electronic devices and equipment such as
ways to address this challenge. One way is discarded electronic devices and equipment,
to change the way we measure economic such as computers, mobile phones,
growth. We could focus on measures of well- televisions, refrigerators, and air conditioners.
being, such as life expectancy, happiness, It may contain harmful substances, and can
and environmental quality, instead of GDP. pose significant risks to the environment and
Another way is to decouple economic growth human health if not managed appropriately.
from resource consumption. This means
finding ways to grow the economy without The production of e-waste is increasing
increasing resource consumption. quickly due to advances in technology and
the popularity of electronic devices among
While technological advancements can consumers10. As a consequence, it imposes

7. Kallis, G. (2011). Degrowth: A vocabulary for a new era. London: Routledge.


8. Global Footprint Network. (2019). Ecological Footprint and biocapacity of 209 countries and territories, 1961-2016.
Retrieved from https://www.footprintnetwork.org/content/documents/2019_Country_Fact_Sheets.pdf
9. International Telecommunication Union. (2020). The Global E-waste Monitor 2020. Retrieved from https://www.itu.int/
en/ITU-D/Environment/Documents/Reports/GEM_2020_report.pdf
10. World Health Organization. (2018). E-waste and children’s health. Retrieved from https://www.who.int/ceh/
publications/e-waste-childrens-health/en/

METALAB REVIEW ¦ 45
potential conflicts between degrowth and global warming to 1.5°C above pre-industrial
technological advancement. The increasing levels, and rapid action is needed to avert
levels of e-waste, low collection rates, a climate catastrophe. In addition, there
and non-environmentally sound disposal are questions around who benefits from
and treatment of this waste stream pose technological advancements and whether
significant risks to the environment and the benefits are distributed equitably. Finally,
human health. Also, as technology advances, a study by Oxfam International found that
older devices become obsolete and are the world’s wealthiest 1 percent emit more
often discarded, leading to a significant than double the carbon emissions of the
environmental impact11. Hence, degrowth poorest 50 percent13. This highlights the need
principles call for a reduction in consumption for degrowth principles to reconsider and
levels, while the technology industry relies on move away from the mentality that growth
continuous consumption and disposal of new is good for the economy, instead prioritizing
products. sustainability, equity, and well-being of the
public over endless growth and consumption.
3. CONFLICTING POWER AND
POLITICS 4. CULTURAL AND SOCIAL
IMPLICATIONS
The implementation of both degrowth
and technological advancement requires Finally, a report by the World Economic
significant changes to the current global Forum found that the gender gap in the
economic and political systems, which technology sector is widening, with women
is a very challenging task. The dominant underrepresented in artificial intelligence and
economic and political actors may not be other emerging technologies14. Additionally,
willing to relinquish their power and privilege the reliance on technology can also lead
to facilitate such changes. It can be seen in to social isolation and a disconnection
the reluctance of many governments to adopt from the world behind the screens of
policies that would limit economic growth or one’s smartphones. This illustrates that
reduce resource consumption. the development of new technologies can
reinforce existing power structures and
Furthermore, this is exemplified by the failure exacerbate social inequalities. In the article
of the COP26 climate summit where we published by UNICEF, it is found that social
all witnessed countries failing to agree on media use was associated with higher
crucial measures to limit global warming12. levels of anxiety and depression among
The International Energy Agency has also teens’ mental health15. The findings reveal
warned the public that current climate that social media use may contribute
policies fall short of what is required to limit to anxiety and depression by increasing

11. United Nations Environment Programme. (2019). Global E-waste Monitor 2019.
12. United Nations. (2023). COP26: Together for our planet. Available at: https://www.un.org/en/climatechange/cop26
13. Oxfam International. (2022, November 8). Billionaire emits a million times more greenhouse gases than average
person. Retrieved from https://www.oxfam.org/en/press-releases/billionaire-emits-million-times-more-greenhouse-
gases-average-person
14. World Economic Forum. (2020). The Global Risks Report 2020. World Economic Forum. Retrieved from https://www.
weforum.org/reports/the-global-risks-report-2020
15. UNICEF. (2018). Is social media bad for teens’ mental health?, Retrieved from https://www.unicef.org/stories/social-
media-bad-teens-mental-health

46 ¦ METALAB REVIEW
social comparison, leading to feelings of KEY TAKEAWAY
inadequacy, and providing a platform for
cyberbullying. However, it is important to note 1. Technology is a tool, not a solution.
that correlation does not equal causation, and It can help us address environmental
more research is needed for us to understand challenges, but it cannot solve them on
the complex relationship between social its own. We need to use technology in
media use and mental health. combination with other approaches, such
as degrowth and social change.
CONCLUSION 2. Achieving both degrowth and
Throughout this article we have seen that the technological advancement would
intersection of degrowth and technological require significant global changes. These
advancement presents us with a complex and changes would be difficult to achieve due
nuanced picture for our future. Technological to political and economic complexities.
advancement is the development of new
technologies that can improve our lives 3. In our current economic system, which
in a variety of ways. On the other hand, is driven by growth, it can be challenging
degrowth is a socio-economic paradigm that to reduce production and consumption
advocates for a reduction in production and levels. This would require a fundamental
consumption levels to achieve a sustainable shift in the way society and the economy
and equitable society. While both concepts operate.
offer potential solutions to the ecological and
social challenges we face, we must recognize 4. Technology has both benefits and
their limitations and critically evaluate their limitations. It can improve our lives, but
implications. Degrowth, for example, could it can also create new problems and
lead to a loss of jobs and economic growth. exacerbate existing ones.
Technological advancement hence could
have negative environmental impacts, such 5. To achieve just sustainable economic
as increased pollution and resource depletion. transition, we need to recognize the
complex and interconnected nature of
The cultural and social dimensions of our environmental, social, and economic
technological advancements must also be challenges. Developing solutions that
considered. For example, some technologies, address all of these challenges is needed
such as social media, can have negative simultaneously.
impacts on mental health. We must work
towards a more equitable and sustainable
global society, where everyone has access to
the benefits of technological advancement,
and where the negative impacts are
minimized. Ultimately, our goal is to find a
balance between sustainability and progress,
and to ensure everyone can enjoy the benefits
of technology without harming the planet or
ourselves.

METALAB REVIEW ¦ 47
ATTENTION
ECONOMY:
YOU’RE NOT
PAYING ENOUGH
ATTENTION
By MICHELLE DIAZ & MARINA PELLET -
Master in Management at ESSEC Business School &
Master in Data Sciences & Business Analytics at ESSEC
Business School & CentraleSupélec

WHAT IS THE specific aspect of our environment carries


ATTENTION ECONOMY? an opportunity cost, or the foregone benefit

E
that would have been derived from another,
conomics is the study of scarcity 1; unchosen task. Just like money, one can also
from wealth, food, energy, or even exchange attention, as we usually allocate
time. Otherwise known as human our attention to tasks in order to gain a
attention. The latter example might seem benefit, whether it is learning a new topic,
odd. Nevertheless, human attention is in fact getting remunerated at work, taking care
a scarce resource. One that is intangible, but of our health and body through exercise, or
is nevertheless invaluable. After all, of the 24 releasing serotonin, endorphin or oxytocin by
hours we have access to a day, one third we engaging in a pleasurable activity. This small
spend sleeping. analysis gives the idiom ‘paying attention’ all
its economical depth.
Every time we decide to focus our attention
on a particular task, our entire cognitive In this context, the attention economy, a
resources are directed to that one task at term first coined by psychologist, economist,
the expense of others. In economic terms, and Nobel Laureate Herbert A. Simon in
spending time focusing our attention on a 1971, can be defined as the mechanisms

1. https://www.merriam-webster.com/dictionary/economics

48 ¦ METALAB REVIEW
© Andrew Moca on Unsplash
and strategies put in place to captivate the As most people can guess, the stakeholders
scarce resource that is human brain time after your attention are predominantly
to generate revenue2. Everytime you pass technology, entertainment, & social media
by an advertising board in the street that companies, in historical accordance with
successfully catches your attention, every the traditional media players: the press,
time your eyes deviate from the news article newspapers, TV channels, & radio. All, to
you’re reading to that newest trendy shoes some extent, are on the conquest of retaining
ad on the side of the webpage, or every time human attention. But in our modern times,
Instagram shows you a ‘suggested’ video that social media platforms have brought a new
you actually watch, you are participating in dimension to capturing human attention.
the attention economy. Indeed, as most of them do not produce any
content themselves, their sole purpose and
WHO ARE THE BIG main revenue stream is to sell available brain
time to advertisers.
PLAYERS AFTER YOUR
ATTENTION? In essence, there are two types of companies
after your attention, the subscription-

2. BER staff. (March 31, 2020). Paying Attention: The Attention Economy. Berkeley Economic Review

METALAB REVIEW ¦ 49
© Markus Winkler on Pexels
based model and the advertising-based deemed popular. Today, this market research
model. In the first category, we have the is bolstered by technology’s improved
media and entertainment industry, like the capacity to collect more fine-grained,
streaming companies: Netflix, Disney+, Hulu, instantaneous, and relevant data on human
for example. In the second category, are behaviour through the various digital
the social media platforms like Facebook, platforms people use. Given the emergence
Instagram, and Tiktok. Both rely on of artificial intelligence (AI), all this data can
consumers to actively and consistently use now be fed into machine learning powered
the services provided, which results in tactics algorithms that use relevant computational
to achieve said result, a feat these companies techniques to better understand and predict
have increasingly managed to do3. human preferences.

WHAT EMERGING Nowadays companies are able to curate


content in unprecedented ways. And over the
TECHNOLOGY IS DRIVING years, platforms have successfully leveraged
THE ATTENTION big data and AI to maximise the time people
spend online. By collecting data every time we
ECONOMY? like, comment, share, rewatch, scroll and post,
social media platforms are able to construct
To sustain the audience or the consumer’s extremely detailed profiles of their users
attention, companies need to serve them and their preferences. Such profiles are built
exactly the content they want. In the past, through sentiment analysis which is backed
companies did this by conducting market by Natural Language Processing (NLP), the
research, adhering to what the market branch of computer science—and more

3. Kemp, S. (July 20, 2023). Social Media Use Reaches New Milestone: User identities now equal to more than
sixty percent of the world’s population. Meltwater

50 ¦ METALAB REVIEW
specifically, a technique in AI which gives negatively impact human psychology. Below
computers the ability to understand text and are apt samples of said influences:
spoken words in much the same way human
beings can4.
MENTAL HEALTH
For example, Facebook can classify as much Consider social media platforms and the
as 52,000 features on each of its users to long-established relationship between their
simultaneously curate the most addictive use and body image. The existence of highly
feed, and target the most appropriate ads perfected, curated, and edited images of
to maximise revenue5. These features can influencers on social media like Instagram
involve pages followed and ads clicked may bring about issues of mental health
amongst other things. In essence, AI has risks such as appearance-related concerns,
become the main strategy used by companies eating disorders, and body dysmorphia.
to capture the scarce resource that is human These issues predate the application of AI in
attention and further generate revenue. these platforms, even when feeds were only
occupied by posts of accounts you follow.
WHY SHOULD YOU But now, with the use of AI, it means that
even when users make the active choice to
CARE? avoid certain types of accounts or content,
they may be unable to do so, especially as
According to Dataportal’s 2023 digital report the ‘suggested’ posts in feeds take up a more
on the global usage of the internet and digital important space than before7.
services, our lives have become inextricably
linked with technology, with over 5 billion Consequently, AI further exacerbates the
internet users all over the globe6. problem. What’s worse is that psychology has
proven that even the things we particularly
It is imperative that we understand the dislike or fear prove to be points of interest,
repercussions that technology plays in our akin to the desire of watching horror movies,
lives. Indeed, some of the digital services even though doing so might give one
we use play on some of our deepest, most nightmares8.
basic, and evolutionary impulses; to the
detriment of our health and general well- How many times have you scrolled through
being. By effectively distilling the audience’s the social media feed of someone you do
desire through the use of AI, companies not particularly like? Might you have done
are now, more than ever, able to capitalise, something more productive, or at the very
take advantage of, and directly or indirectly least, less emotionally taxing instead?

4. Taherdoost, H. (2023) Enhancing Social Media Platforms with Machine Learning Algorithms and Neural Networks,
Algorithms, 16(6), 271. https://doi.org/10.3390/a16060271
5. Wright, K. (September 15, 2018). How Facebook AI may help to change the way we shop online in the future. CNBC
6. Kemp, S. (July 20, 2023). Social Media Use Reaches New Milestone: User identities now equal to more than sixty
percent of the world’s population. Meltwater
7. Harriger, J. A., Evans, J. A., Thompson, J. K., Tylka, T. L. (2022) The dangers of the rabbit hole: Reflections on social
media as a portal into a distorted world of edited bodies and eating disorder risk and the role of algorithms, Body image,
41, 292-297. https://doi.org/10.1016/j.bodyim.2022.03.007
8. L. Coolidge F., Srivastava A. (2021) Why We Enjoy Horror Films. Decoding the fascination behind scary movies.
Psychology Today

METALAB REVIEW ¦ 51
POLARISATION & DIGITAL the Rohingya which contributed to real-world
ESCAPISM violence.” – Agnès Callamard, Amnesty
International’s Secretary General11. Indeed,
On the other hand, having an algorithm researchers have predicted that the dawn of
cater to specific interests can be viewed as effective and persuasive generative AI will
a good thing in that users are able to easily further exacerbate the problem12.
discover more content that they will enjoy, as
is the case for entertainment platforms with Secondly, we have digital escapism, which
AI-powered recommendation systems. But refers to the use of digital technology and the
the increased personalisation comes with internet as a means to escape the realities of
caveats like polarisation and digital escapism. our lives. This new trend of digital escapism,
in addition to private spaces in the virtual
Polarisation refers to the tendency of social world13, has been further exacerbated by AI.
media users to fall into a vacuum of content The act of escaping itself may not be all that
that only reinforces their view of the world, bad, after all, we all need a reprieve from time
thus giving them a biased and distorted to time. However, research has shown that
view of reality9. Such is the case with excessive escape to the digital world has
recommended content on platforms, wherein given rise to psychological disorders. It brings
“social media robots, more commonly known forth “Internet Disorder syndrome” which
as “bots,” are becoming a formative tool of invites stress, depression, deviation from
online radicalization”10. It is easy to draw one's own goal, sleeplessness and anxiety14.
the line from these bots to the rampant
polarisation online that has instigated many
incidents of real-life violence owing to the
DECREASE IN
spread of disinformation, conspiracy theories, ATTENTION SPAN
and radical viewpoints. Such is the case with
the 2017 Rohingya Massacre in Myanmar, According to the Merriam Webster Dictionary,
wherein people were “killed, tortured, raped, attention span is defined as “the length of
and displaced in the thousands as part of the time during which one (such as an individual
Myanmar security forces’ campaign of ethnic or a group) is able to concentrate or
cleansing. In the months and years leading remain interested.”15 Historically, scientists
up to the atrocities, Facebook’s algorithms hypothesise that having a shorter attention
were intensifying a storm of hatred against span served humans well, in that it allowed

9. Arora, S. D., Singh, G. P., Chakraborty, A., Maity, M. (2022) Polarization and social media: A systematic review and
research agenda, Technological Forecasting and Social Change, 183. https://doi.org/10.1016/j.techfore.2022.121942
10. Mantello, P., Ho, T. M., Podoletz, L. (2023) Automating Extremism: Mapping the Affective Roles of Artificial Agents in
Online Radicalization, The Palgrave Handbook of Malicious Use of AI and Psychological Security, 81-103. https://doi.
org/10.1007/978-3-031-22552-9_4
11. Amnesty international (2022) Myanmar: Facebook’s systems promoted violence against rohingya; meta owes
reparations
12. Hsu, T., Thompson, S. A. (February 8, 2023) Disinformation Researchers Raise Alarms About A.I. Chatbots. The New
York Times
13. Subudhi, R. N., Das, S., Sahu, S. (2020) Digital Escapism, Journal of Humanities and Social Sciences Research, 2(S),
37-44. https://horizon-jhssr.com/view-issue.php?id=63
14. Ibid.
15. https://www.merriam-webster.com/dictionary/attention%20span

52 ¦ METALAB REVIEW
us to quickly monitor for danger. However, INTERVENTIONS: WHAT,
in today’s digital era, humanity’s shorter WHERE, & REALLY?
attention span can be detrimental. In fact,
the US National Center for Biotechnology The mechanism behind social media and
Information claims that the average human entertainment platforms have not always
attention span has dropped from 12 to 8 used AI to construct their various newsfeeds.
seconds between 2000 and 2013, just one In other words, while the feeds could be
second below that of a goldfish, which is addictive, they were not always powered by
why the phenomena of a decreasing ability AI. At one point, it was true to its name; it was
to concentrate on a task was called the simply a feed of the content posted by people
“Goldfish Effect”. Social media is often held you followed. Not a curated feed designed
as most accountable for this drop in our to explicitly keep you online for as long as
attention span. possible.18 The change into algorithmic feeds
came into effect largely around 2009, with
Now, consider the push through notifications Facebook leading the way by prioritising
on our phones from these platforms. Each popular posts.19 But the mission was always
ping conditions us to reach for our phones. to retain users on the platforms for as long as
So much so that even without that ping, possible, and the use of AI is simply the most
people find themselves reaching for their powerful way of currently doing so.
phones anyway; what researchers have
come to call phantom phone signals.16 But big tech is aware of the backlash, and has
Given the boost that AI gives to social media subsequently introduced measures that some
and entertainment companies after our may call ways to curb the excessive use of
attention, it is imperative that society at large their services. And the release of Netflix’s
understands how detrimental a decrease in documentary on the attention economy:
one’s attention span can be for their lives. The The Social Dilemma, has made people more
inability to focus can decrease productivity aware of the issue.20 Meanwhile, Apple and
at work, make it harder to complete personal other technology device manufacturers have
chores, and at their very worst, impact introduced app timers to help users better
personal relationships.17 manage their time and curb the hours they
spend in front of a screen. And child settings
How many times have you found yourself can be applied on these devices to safeguard
annoyed with someone using their phone internet use, excluding certain websites and
while you’re relaying a story to them? How searches that may be bad for children.
many times have you been that person? Do
we not owe the people we care about our Furthermore, some companies have even
absolute attention? Do we not want that for implemented what can be called more
ourselves? Unfortunately, without outsized concrete measures like Alphabet and Meta.
intervention, it is an uphill battle for most Indeed Google Chrome has been on course
users. to phase out third-party cookies since 2021,

16. Kelleher, E. (January 14, 2023) 5 Things To Do For Better Focus & A Longer Attention Span, mbghealth
17. Ibid.
18. Isaac, M. (March 15, 2016) Instagram May Change Your Feed, Personalizing It With an Algorithm. The New York Times
19. Wallaroo (2023) Facebook News Feed Algorithm History
20. https://en.wikipedia.org/wiki/The_Social_Dilemma

METALAB REVIEW ¦ 53
with 1% of cookies being phased out in early continue rapid company growth and appease
2024.21 Moreover, Meta has increasingly shareholders.
over the years made privacy controls more
accessible for its users; users now have the
option of tweaking what data, if any, is shared
REGULATION?
with the advertisers.22 Currently, the European Union has led the
way when it comes to regulation of big
But how effective are these efforts really? tech companies. For example, through the
Might we even call these interventions? Or recent Digital Services Act (DSA), regulating
just ways for these companies to seemingly illegal content, transparent advertising, and
address customer concern, and in effect disinformation online, the Digital Market
continuously sell their products? Or worse Act (DMA), ensuring a higher degree of
yet, might these tweaks just be another push competition in the European Digital Markets
to further sell said products, as an appeal to by preventing the largest digital players
concerned consumers? After all, as anyone (defined as Gatekeepers) from abusing
with a smartphone knows, you can always their market position and power by enabling
just extend the timer. Does it not, in some smaller companies to enter the market on a
way, feel like the impetus of responsibility more level playing field. And lastly, albeit the
is being pushed onto users? Do individual most known, the General Data Protection
users actually stand a chance when a former Regulation (EU), a regulation on privacy
Google designer, Tristan Harris compares the and data protection in Europe, adopted
level of addiction to a "Vegas slot machine" in April 2016 and became enforceable in
as users "check their phones hoping that they May 2018.
have a notification, as if they are pulling the
lever of a slot machine hoping they hit the The GDPR remains one of the core
jackpot"?23 regulations on privacy and data protection in
Europe, and arguably one of the biggest tech
As the business models of social media and regulations so far, which has sparked similar
streaming companies continue to largely be regulations across the globe.24 In 2023, the EU
based on the attainment and maintenance of AI Act, the first regulation on the use of AI was
people’s attention, it follows that society at adopted by the parliament, with talks to begin
large cannot expect intervention to primarily with EU countries in the Council on the final
come from companies themselves. From a form of the law underway25. Outside Europe,
business and profit driven perspective, these and given the testimony of Sam Altman, the
companies cannot be effectively counted OpenAI chief executive on the repercussions
upon to match their interventions to reduce of AI earlier this year in Congress, multiple
screen time to healthier levels. Particularly, actors within the US government are jostling
if said levels fall below what is required to to regulate AI26. In spite of all this and given

21. Lardinois, F. (July 20, 2023) Google starts the GA rollout of its Privacy Sandbox APIs to all Chrome users. TechCrunch
22. https://www.facebook.com/privacy/center
23. McGinley, J. (November 2, 2020) 3 Things We Learned About Social Media from Netflix’s “The Social Dilemma”.
Blackstone LaunchPad, Syracuse University
24. Statista (2023) Tech regulations in Europe - Statistics & Facts
25. European Parliament (2023) EU AI Act: first regulation on artificial intelligence
26. Levin, B., Downes, L. (May 19, 2023) Who Is Going to Regulate AI?. Harvard Business Review

54 ¦ METALAB REVIEW
© dole777 on Unsplash
the historical speed at which regulation Just like we have seen the rise in demand
comes into play around big tech, it is for more sustainable and ethical products,
imperative that users of digital services and we can hope to see a mass gathering and
social media platforms refrain from being ask for anthropologically beneficial tech
complacent. products that promote learning, collaboration,
and solidarity; stepping away from the
A WAY FORWARD? negative effects we’ve outlined. Much like the
environmentalist movement, where people
As individual users of these entertainment make conscious decisions to fly less, buy
and social media platforms, what is there to local groceries, and support circular brands,
do? How might we shift the powers that be? online users can shift to digital minimalism.
Do we have that ability? The Social Dilemma By using alternative internet services like
was released in 2020, and three years later, Duckduckgo, Vivaldi, or Ecosia, users are able
here is an article calling your attention yet to limit data collection from big tech to some
again to the attention economy. Outlook degree, reduce their time online, and help the
is rather grim. But users can call for better planet along the way.
behaviour from big tech — it may not seem
like much, but calling attention to the issue, Additionally, the decentralisation movement
with a big enough mic, and a coordinated or Web 3.0, which is an alternative way for
enough mass, might speed up the process the internet to work could be something
of accountability. Indeed, let’s not forget that that users might think to further support.
consumers create demand and have the In essence, the decentralisation of the
power to shift the actions of big companies internet means widely distributing the control
through their choices. of online information, interactions, and
experience of users. Such a system would

METALAB REVIEW ¦ 55
56 ¦ METALAB REVIEW
© Anete Lusina from Pexels
see the reorganisation of the internet into a
peer-to-peer infrastructure as opposed to
the current centralization of data amongst
hosting services27. More concretely, this
would mean that large amounts of data might
no longer be concentrated in the hands of
those that provide the services we use, like
Alphabet, Meta, and Apple.

But for now, reading this article has hopefully


made you more aware of how your phone,
that you bought with your hard-earned
money, might be serving big tech more than
it serves you. The first step to addressing
any problematic issue is awareness; we
all need to pay a bit more attention to the
consequences of our time online. Remember,
the attention economy is not an immovable
object, met with the right amount of pressure,
it can change. Like any business model. Given
that, shall we turn the tide and assure your
phone serves you first? Sharing this article is
a start.

27. Polzin, M. (August 24, 2016) What Is the Decentralized


Web? 24 Experts Break it Down by Mackenzie Polzin.
Syracuse University

METALAB REVIEW ¦ 57
CITIES OF THE
FUTURE
EVALUATING AI’S POTENTIAL IN
CLIMATE-RESILIENCY

AS AN INCREASING NUMBER OF PEOPLE MIGRATE TO CITIES,


WILL AI'S ADOPTION EMPOWER CITIES TO PROTECT ITS
INHABITANTS FROM THE RAMIFICATIONS OF CLIMATE CHANGE?

By Seethal REDDY KAULURi


Master in Management at ESSEC Business School

F
rom bushfires in Australia, floods in India,
“AI IS droughts in Africa, to dry corridors in
Central Asia, we have seen some of the
THE NEW deadliest and costliest climate disasters of
history1 in According to the United Nations2
ELECTRICITY.” the recent past. They won’t be the last of such
disasters to occur if our efforts to fight climate
Andrew NG, co-founder of Coursera change remain scant as today. According to
and Google Brain Deep Learning the United Nations, the global emissions were
Project. supposed to be reduced by 7.6% per year
between 2020 and 2030 in order to limit global
warming below 2 degrees Celsius (3.6 degrees
Fahrenheit) set at the Paris Agreement. However,
global emissions increased in 2021 after a dip by
5% in 2020 due to the COVID-19 pandemic.

1. 10 costliest climate disasters of 2022. (2023). World Economic Forum.


2. Cut global emissions by 7.6 percent every year for next decade to meet 1.5°C Paris target. (2019). UN Environment
Programme

58 ¦ METALAB REVIEW
© Eren Namli on Unsplash
Cities contribute to over 70%3 of these up to 30 percent of annual daylight hours by
global CO2 emissions due to massive 2050 due to rising temperatures.
industrialisation and motorised transportation
systems that still use fossil fuels. They are As we desperately hunt for solutions
also the most vulnerable to these disasters to protect our cities in the wake of this
as they are home to more than half the emergency, this article critically compares
world’s population today and three quarters and assesses the potential of AI as a solution
of the population or 6.5 billion people by as it is increasingly revered today in various
2050. Asia and Africa, where the majority applications due to its abilities to perform
of the population still resides in rural human-like tasks faster, with minimum
areas, increasingly face the stress of rapid errors, and at a much grander scale than was
urbanisation, while uplifting living standards possible before.
at the same time. According to Mckinsey
Global Institute's report4, Asia will starve due
to increasing droughts, and India may lose

3. Dasgupta, S., Lall, S., Wheeler, D. Cutting global carbon emissions: where do cities stand? (2022). World Bank
4. Woetzel, J., Tonby, O., Krishnan, M., Yamada, Y., Sengupta, S., Pinner, D., Fakhrutdinov, R., and Watanabe, T. Climate risk
and response in Asia. (2020). McKinsey Global Institute.

METALAB REVIEW ¦ 59
© Brodie from Burst
ARE SMART CITIES cloud computing, Internet-of-Things (IoT), or
ENOUGH? artificial intelligence to achieve operational
efficiencies for the welfare of the citizens.
To overcome the challenge of rapid But this definition doesn’t explicitly state
infrastructural development while improving that smart cities are necessarily built to be
quality of living, many African and Asian resistant to the adversities of climate change.
countries are looking to build smart cities: to
integrate advanced technologies to manage According to the Intergovernmental Panel
overcrowding and traffic, and to manage on Climate Change5, Climate Resilient
resources better. Smart cities also seem to Development is urgently needed. It defines
be a way of attracting foreign investment as this framework as, “reducing exposure and
they are seen as an opportunity to expand vulnerability to climate hazards, cutting back
markets, promote innovative products and greenhouse gas emissions and conserving
services, tap into emerging hubs for skilled biodiversity are given the highest priorities in
workforce and gain trade benefits. While everyday decision-making and policies on all
there is no universally accepted definition aspects of society including energy, industry,
of a smart city, in a broader sense, it has health, water, food, urban development,
been defined by multiple sources as a housing and transport.” Therefore, as the
city that leverages smart technology and global smart cities market6 is expected to
communication, such as, data analytics, grow to the value of about $1.38 trillion by

5. What is Climate Resilient Development and how do we pursue it? Intergovernmental Panel on Climate Change report.
6. Thormundsson, B. Smart cities market revenues worldwide 2019-2030. (2023)

60 ¦ METALAB REVIEW
2030, there is an increasing need to refocus region, an area larger than Israel, aims to
this investment on building smart cities that eliminate cars and roads by replacing them
are, in fact, resilient. with a hyper-speed rail that covers the entire
Line - the shape in which the city will be built
UNCONVENTIONAL for a length of about 177 km. Each block of
this city aims to be self-sufficient by building
CITIES schools, groceries and other basic amenities
at a walkable distance to cut emissions. The
Before proposing solutions, let us first project also claims to care for biodiversity by
analyse some popular non-AI experimental building a coral-reef restoration project.
eco-settlements and assess whether
inspirations can be drawn from their styles of However successful or utopian these
governance, economy, and self-sufficiency experiments might sound, it is overly
in achieving climate-resilience. One such ambitious to adopt their styles on a large
settlement is Auroville7 in Viluppuram scale. Auroville is a close-knit community
district of India, which was founded in 1968 with a number of residents just over 3000,
as a community in which people from all although the plan was initially to attract
over the world could come together and tens of thousands more residents. This
live in harmony, transcending the barriers sense of unity helps build social resilience,
of nationality, race, and religion. The town which is essential for climate resilience but
is resilient to climate as the inhabitants cannot be replicated at a global scale due
are committed to sustainable living while to bureaucracy and disharmony between
reducing environmental impact by adopting national and local governments. This is
methods such as using renewable energy, also why NEOM, which aims to host up to
organic farming, and implementing water 9 million residents by 2045 (only 0.09% of
conservation and waste management the world population by then), aims to have
systems. Auroville also places a strong an independent governance system with
emphasis on educational programs that laws of its own, which not only substantiates
help build the knowledge and skills needed that it is easier to control climate policies in
to adapt to the impacts of climate change. a smaller sample, but also establishes that
It is home to several research centres it is impossible to achieve these climate and
and innovative projects that focus on self-sufficiency objectives while remaining
sustainability and climate resilience. under the law and economy of its host
country that still derived 68%9 of its budget
Another utopian settlement project is NEOM8, revenues from oil in 2019. Another challenge
which is part of Saudi Arabia’s vision 2030 of scaling these models is the differences
that aims to offer its citizens the best of in global diversity. Even though Auroville
diverse climates - from sun-soaked beaches, brings together diverse people from all over
floating industrial islands on Red sea, to ski the world, they are still connected through
resorts in snow capped mountains, all while a common sense of spirituality and respect
running on 100% renewable energy. The for nature. These individuals share values of

7. Koduvayur Venkitaraman, A., Joshi, N. A critical examination of a community-led ecovillage initiative: a case of Auroville,
India. Clim Action 1, 15. (2022).
8. Palmer, I,. Is Saudi Arabia’s New Climate City ‘Neom’ Future Or Fantasy? (2022) Forbes.
9. Almasoud, A., Albasri, K., Alshammari, M., Alghamdi, R., Alanazi, A., Binladen, S., Alsikhan, S., Kingdom of Saudi Arabia
Budget Report. (2018). KPMG.

METALAB REVIEW ¦ 61
collectivism and insist on learning a common for better mitigation and preparedness in the
language, Tamil, to facilitate cooperation short-term and adaptation and resilience in
and smooth community functions. Finally, the long-term.
such models depend on local resources for
self-sufficiency. While Auroville depends on In buildings, energy can be optimised by
its organic farms and renewable projects autonomous lighting, heating, and cooling
for food, energy and water, Neom aims to based on occupancy levels, time of the
build a close-knit self-sufficient economic day, weather conditions and occupants’
hub by attracting world-wide talent from 14 comfort. AI can be used to track carbon
industrial sectors in creating a concentration foot-print, water usage, and waste production
of resources and economic power that makes in order to set consumption limits. AI can
it far easier to achieve resilience here than spot anomalies through data gathered
at a macro-level. Analysts also suggest that from sensors and check performances
projects like Neom are more a delay towards of equipment outside the norm for timely
climate objectives than a solution as living maintenance and repairs. AI can predict
in Saudi Arabia necessitates the use of air energy requirements to ensure ample
conditioning, which is still challenging to shift solar panels and storage are installed for
into renewables. uninterrupted renewable energy consumption.
Comparative data analysis from other
Therefore, as the sustainable models green-buildings, or with simulations based
implemented in the cities discussed above on AI-built “digital twins”, can be used to
cannot be replicated on a larger scale, we prescribe alternative architectural designs
must turn towards more pragmatic and in comparable areas to further reduce
viable solutions to improve our cities’ climate emissions if required. For example, it can be
resilience. prescribed to use tinted windows or green
roofs to keep the buildings cooler if the major
THE PROMISE OF AI carbon emitter for a particular building is air
conditioner. AI can cluster the occupants
According to a survey conducted by BCG10, based on consumption behaviours and
AI can expedite reduction of GHG emissions emissions for law-makers to either encourage
up to 10% per organisation and up to favourable behaviours through reward
5.3 billion tons of CO2e if scaled globally. systems or through penalties.
This is because of its enormous power
of gathering, computing, and analysing Similarly in transport, AI can optimise train
unstructured, multidimensional, and end-to- and bus routes and frequencies by matching
end data points. As cities in Asia and Africa demand and supply based on time of day,
suffer from inefficient city designs due to season, weather, etc. It can be used to predict
informal settlements and inadequate public repairs and maintenance in advance to
transportation systems, these regions can prepare for downtime and ensure that public
radically transform into sustainable and transportation is always available and running
resilient cities by leveraging AI’s power in smoothly without interruptions. It can also
360-degree information modelling to optimise be used to propose on-demand services
energy, reduce wastage, and improve or reduce off-peak inefficiencies caused
accuracy and speed of climate-modelling by vehicles running almost empty because

10. Maher, H., Meinecke, H., Gromier, D., Garcia-Novelli, M., Fortmann, R. AI Is Essential for Solving the Climate Crisis.
(2022). BCG.

62 ¦ METALAB REVIEW
© Mitul Shah from Burst
of their fixed itineraries and schedules. AI can study vast data from weather sensors,
Personal vehicles can be replaced by car- satellites, news reports, and social media,
pooling that matches passengers based on and spot hidden patterns and events to
pickup and destination. AI can also analyse predict risks, such as temperatures, sea-
data about walking, running, and cycling levels, droughts, rainfalls, heat-waves,
activity to make recommendations. For hurricanes etc. Predicting these events can
instance, it can suggest the best locations help prepare the necessary infrastructure
for basic neighbourhood amenities such for minimising the consequences, including
as groceries, coffee shops, pharmacies, utilities’ networks (water, gas, electricity,
schools and emergency clinics, to ensure etc.), critical to the proper functioning of a
widespread access that does not rely on city. These patterns and clustering can also
fossil fuels. It is also possible to centralise help predict the degree of impact for each
transport systems to optimise routes and region based on population density, industrial
traffic. A city’s transportation system can be activity, infrastructure, biodiversity, etc and
entirely integrated on a 360-degree carbon compute the degree of rescue services and
monitoring to ensure that collectively, the relief supplies required for better mitigation.
city’s transport stays within the carbon An integrated biodiversity system can help
budget. If at a particular time, personal monitor species, plantations and conditions
vehicles on road cause more emissions, of habitats to spot patterns of endangerment,
reinforcement learning can be used to deforestation and environmental factors
encourage citizens to use more of public affecting the weather patterns.
transport by increasing frequency of trains
and metro while increasing time of red traffic Thus, AI’s power of integration systems from
signals on the road to discourage vehicle macro to micro-level emission monitoring
traffic or converting roads into carbon-free and hazard prediction can help build a
zones to promote cycle and pedestrian-traffic harmonised source of truth and scale what
only. smaller experiments like Auroville or NEOM
have on a much larger, global scale.

METALAB REVIEW ¦ 63
ADOPTING AI a large dataset will increase to $500 million by
2030 due to an increase in the size of datasets
The aforementioned ideas may sound like the needed to train large models. However, setting
answer to our problem, but there is resistance up this investment right away will ensure
in the adoption of AI due to several reasons. resilience, cost-efficiency and a significant
Firstly, the data is widely scattered and gain in the GDP of up to 4.4% 12by 2030. But
dissociated as there is a lack of centralised even if these governments were convinced
data source. This is because the climate to step up the financing for this preparation
objectives are too broad and the firms and today, a larger issue regarding AI adoption
law-makers haven’t adequately defined would remain: can we trust the results that AI
the KPIs to be tracked for each sector. To provides?
establish the same could be challenging as a
large infrastructural overhaul will be required
to substitute independent systems with
TRAINING GREEN-AI
interconnected, autonomous and accountable MODELS
systems and appliances which might be
highly infeasible depending on the scope One of the reasons why AI-based
of project and the cost of sensors, storage, recommendations are not always relied upon
computing resources, and personnel required is due to the quality of their design and model
to implement the models. Therefore, it is the training. Human intervention in feeding,
duty of the law-makers and enforcers to break labelling and training data while the model
down the objectives to smaller key results is built can give room for bias and skewed
and introduce policies for infrastructure, results. For example, if the AI-based traffic
construction and automobile industries to system mentioned earlier aims to manage
mandate accountable data collection points demand and supply to optimise bus routes
in their products. Secondly, there is lack of in a city, such a model will work inefficiently
education and training on AI expertise among if it has been trained on historical data of a
organisational leaders and governing bodies city’s bus routes where certain high-income
to realise its vast potential. Free online and areas never had a bus-stop because they
offline educational programs, workshops, and used cars. Such a model will have to be
certifications shall be set up from schools trained on collecting data from all modes
to organisational leaders and policy-makers of transport to understand commuting
to equip them with the requisite knowledge. patterns. New experimental car-pooling and
States should also promote research centres, bus routes will have to be set up initially to
events and competitions in the field of green- collect data on public transport preferences
AI to encourage new and innovative ideas. of the citizens. Another example is if a model
Another challenge is that setting up large- has studied smartphone data of a region to
scale and centralised AI integrated systems understand walking and cycling patterns to
requires huge financial investments which is plan roads and accessible neighbourhood
challenging for countries in Asia and Africa amenities such as groceries and pharmacies.
due to the pressure of mitigating fundamental Such a model will work inadequately if the
gaps in the economy and infrastructure. data collected was from an area of ageing
According to a report by OpenAI11, it is population with minimal smartphone
estimated that the cost of training a model on adoption. Therefore, such models will have

11. AI Model Training Costs Are Expected to Rise From $100 million to $500 million By 2030. (2023). MetaversePost.
12. Herweijer, C., Combes, B., Gillham, J. Artificial intelligence and the fate of planet Earth. (2019). PWC.

64 ¦ METALAB REVIEW
to be trained by nudging some preferred human efforts required despite the massive
behaviours first13, and may rely on human capital expenditure and with the uncertainties
modelling assumptions (such as that of the risks present in adopting it.
which was used in the 1970s to predict
train demand for the newly created BART
system in San Francisco, and for which Dan
CONCLUSION
McFadden received the 2020 Nobel Prize in Given the numerous applications of AI
Economics)14. in mitigating climate risks and achieving
climate-resiliency, overlooking such an
An additional concern is regarding the immense potential will be a significant
confidentiality of the information captured missed opportunity. However, due to the
by the AI systems that can compromise challenges present in adopting a globally-
privacy and security. The accuracy of scaled autonomous infrastructure and
these models will depend on the extent to the accuracy of its decisions, AI is only an
which the inhabitants are willing to share accelerator towards achieving our climate
their information. The law-makers need to objectives and not the solution itself.
carefully consider the type of information to Just as it is easy to coordinate smaller-
be mandatorily collected from the citizens, sized homogenous experimental cities,
and ensure that it does not compromise implementing AI for climate should be
their freedom of choice. Additionally, will scaled step-by-step starting with the most
highly autonomous systems be capable of vulnerable sectors by carefully monitoring the
considering externalities? Or new patterns? results and reimplementing models in similar
Indeed, what if green-buildings’ AI integrated and comparable conditions.
systems were trained on pre-pandemic
occupancy levels, but when another Achieving climate-objectives goes beyond
pandemic, like COVID, hits the region causing adoption of AI. International organisations,
almost everyone to work from home? Will governments, urbanists and corporations
such AI systems accept the extra-bandwidth have a fundamental role in adopting a
of energy needed to run the residential harmonious framework by breaking down
buildings? It would be necessary that such climate objectives into smaller-measurable
a system should also train the carbon targets for each region, industry and sector.
budget at the macro-level and maintain a They play an important role in creating a
global balance of buildings’ emissions, by clear path for investment and innovation.
temporarily transferring the carbon budget Governments should pave the way for
of office buildings to that of the residential adoption of new technologies that will not
buildings. This brings us back to the general only make our future cities smart, but future-
limitations of AI that when faced with proof them. Policies in education, research
adversities, especially in situations when and business will play a defining role in the
no historic data or patterns exist, a human future of AI and in its adoption for a green,
intervention is always required. Then the real stress-free future of habitation.
question remains for us to ask ourselves,
whether undergoing such an ambitious
infrastructural project is justified given the

13. Cowls, J., Tsamados, A., Taddeo, M. Floridi, L. The AI gambit: leveraging artificial intelligence to combat climate
change—opportunities, challenges, and recommendations. (2021). AI & Soc.
14. McFadden, D. The Measurement of Urban Travel Demand. (1974). Journal of Public Economics, 3, 303-328.

METALAB REVIEW ¦ 65
FROM VIRTUAL
MODELS TO REAL-
LIFE IMPACT:
HOW DIGITAL TWIN AND AI ARE DRIVING
SUSTAINABILITY EFFORTS

CAN AI-POWERED DIGITAL TWIN TRULY PAVE THE WAY FOR


ACHIEVING SUSTAINABILITY GOALS?

By VANSHIKA SHARMA - Master in Data Sciences & Business Analytics


at ESSEC Business School & CentraleSupélec

I
n today’s digital age, innovative technologies minimize waste, and promote sustainable
are driving significant transformations in practices across sectors? In this article, we
various industries. One such transformative will explore the concept of digital twins, their
technology is digital twin technology, which role in driving sustainability across industries,
involves creating virtual replicas of physical the associated challenges, and their future
objects or systems. This state-of-the-art prospects.
technology has proven to be instrumental in
optimizing operations, reducing production
timelines, and fostering innovation across
WHAT ARE DIGITAL
industries. By connecting the physical TWIN?
and digital realms, digital twins enable
simulations, real-time data-driven decision- Digital twins are high-fidelity digital replicas
making, and remote system management. of physical entities or systems that mimic
their behavior and processes over time. By
This brings us to an exciting intersection integrating data from sensors and machines,
of technology and sustainability. Can the digital twins provide a comprehensive
combination of digital twin technology understanding of the physical world, enabling
and artificial intelligence (AI) enhance better decision-making and optimizations.
sustainability efforts? Can this data-driven
approach optimize resource utilization,

66 ¦ METALAB REVIEW
© Zack Walker on Unsplash
The impact of digital twins has permeated waste, and fostering sustainable practices
through manufacturing, healthcare, across industries. Industrial giants like
construction, energy, and e-commerce. For Unilever harnessed the power of digital twins
instance, Amazon, a global e-commerce at its facility in Brazil. They used a digital
leader, utilizes digital twin technology to twin to fine-tune manufacturing parameters,
enhance remarketing advertisements. By such as the temperature for soap production.
leveraging vast amounts of data on users’ Such minor tweaks led to a productivity
past purchases, viewed products, and visited improvement of 1% to 3% and an energy cost
websites, Amazon’s digital twin algorithm saving of USD 2.8 million.
creates personalized replicas of each
consumer. This enables tailored experiences But how did we arrive at this point, and
and targeted remarketing ads, ultimately what lies ahead? Let’s delve into the
driving users back to their platform. origins of digital twins and AI, explore their
convergence, and understand their impact on
However, the significance of digital twins the sustainability landscape.
extends beyond business efficiency. When
combined with AI, these virtual replicas
hold promise in promoting sustainability
by optimizing resource utilization, reducing

METALAB REVIEW ¦ 67
© Conny Schneider on Unsplash
BRIEF HISTORY OF and its potential to revolutionize industries.
The integration of the Internet of Things (IoT)
DIGITAL TWINS AND AI in subsequent years made digital twins more
cost-effective and practical, paving the way
Digital twin technology has evolved for widespread adoption across industries.
significantly since its inception in the 1960s.
It was first utilized by NASA to create physical In recent years, the integration of AI has taken
replicas of systems on the ground that digital twin technology to new heights. AI
matched those in space. The breakthrough analyzes data from various sources, providing
came during the famous Apollo 13 rescue real-time insights into system performance
mission, where a critical system failure and empowering organizations to make
threatened the crew’s safety. Leveraging accurate and timely decisions. Today, AI-
digital twin technology, NASA engineers enhanced digital twins are at the forefront of
simulated conditions and tested solutions sustainability efforts, optimizing operations,
from the ground, eventually guiding the crew reducing waste, and improving resource
back to safety. efficiency. The synergy between digital twin
technology and AI drives innovation and
However, it wasn’t until the early 2000s that facilitates sustainable development, making
the term «digital twin» was coined by Michael them indispensable for our future.
Grieves during a presentation on product
lifecycle management center. This marked the
formal recognition of digital twin technology

68 ¦ METALAB REVIEW
DIGITAL TWINS AND AI: of digital twins and the potential of AI,
THE SUSTAINABILITY organizations can achieve their environmental
goals, propelling us toward a more
GAME-CHANGERS sustainable future.

Sustainability efforts encompass a The market’s response to digital twins affirms


multifaceted approach to meet present their growing importance. In 2020, the digital
needs without compromising the ability of twin market was valued at $5 billion. However,
future generations to meet theirs. In this a 2023 report by Fact.MR estimates this
context, digital twins and AI have emerged as market to swell to a staggering US$ 95 billion
significant catalysts for driving sustainability. by 2033, growing at a brisk 34.3% CAGR from
2023 to 2033. A survey conducted by the
Digital twins revolutionize system monitoring Capgemini Research Institute in 2022 adds
and optimization, excelling in resource another dimension to this trend: 51% of the
utilization, emissions simulation, and supply digital twin implementers surveyed believe
chain optimization. They also play a vital that this technology will help them achieve
role in predicting maintenance requirements, their environmental sustainability goals.
minimizing downtime, and generating cost
savings while enhancing productivity. There are, of course, varied views
regarding the role of digital twins in driving
The increasing role of AI in digital twins sustainability. While some organizations
further enhances their precision. Acting as primarily see this technology as a tool for
digital mirror images, AI leverages real-time process efficiency, others perceive its role in
data to provide real-time insights into system product design and development. Regardless,
performance, empowering organizations to 57% consider improving sustainability
make informed decisions based on accurate efforts as a key driver of their digital twin
and timely information. AI algorithms excel investments, underscoring the environmental
in data analysis and predictive capabilities, considerations that factor into the adoption of
surpassing human cognitive abilities in speed this technology.
and accuracy.
To make this idea more tangible, let’s
Moreover, AI addresses key challenges in understand it with an example of the
the digital twin landscape by facilitating industrial manufacturing sector. Here,
seamless integration and improved digital twin technology, coupled with AI,
performance across systems. It consolidates identifies improvement areas to transform
sensor data, optimizes processes, and operations. Manufacturers can spot pain
ensures real-time updates are delivered, points, such as machine inefficiencies
overcoming interoperability challenges. The or quality control issues, that may cause
transformative power of AI in the digital product defects or delays. Digital twins also
twin ecosystem cannot be overstated, as identify energy loss areas and opportunities
it enhances the simulation and prediction to reduce consumption, like inefficient
accuracy of virtual models. machinery or sub-optimal processes. AI
algorithms expedite data analysis and future
This powerful combination of digital twins outcome predictions, surpassing the pace
and AI not only accelerates sustainability and accuracy of human cognitive abilities.
efforts but also implements them more Virtual simulations further curb waste
effectively. By leveraging the capabilities and power consumption associated with

METALAB REVIEW ¦ 69
physical prototypes, permitting emulation optimal evacuation routes, ensuring citizen
of a production line using a digital twin. This safety. Virtual Singapore also supports public
enables changes at every stage to amplify health monitoring, measuring the density
cost savings, boost efficiency, and diminish of disease-carrying mosquitoes to predict
environmental impact, making digital potential outbreaks. This leads to swift
twins and AI integral allies in the quest for preventive actions and aids in responding
sustainability. promptly to potential health crises.

HARNESSING DIGITAL DRIVING SUSTAINABLE SUPPLY


CHAINS AND LOGISTICS
TWINS AND AI: STORIES
OF SUSTAINABLE Digital twins are playing a significant role in
optimizing supply chains and logistics for
TRANSFORMATION sustainability, as seen in the collaboration
between Maersk and Ericsson in 2023. By
A TALE OF A SMART CITY: VIRTUAL creating a digital twin of a shipping vessel,
SINGAPORE they are able to optimize fuel consumption,
leading to a reduction in carbon emissions.
Singapore, a city-state often acclaimed for its Using real-time sensor data, the digital twin
smart city initiatives, stands at the forefront of modeled the vessel’s operations to pinpoint
adopting digital twin technology. Its ambitious improvement areas. As a result, Maersk was
project, Virtual Singapore, exemplifies how able to cut fuel consumption and associated
this technology can be harnessed to drive carbon emissions by up to 20%, contributing
sustainable urban development. significantly to its sustainability initiatives
and leading to considerable cost savings.
Launched in 2023, Virtual Singapore is a Predictive maintenance enabled by the digital
dynamic digital twin of the city, incorporating twin also allowed Maersk to proactively
a wide range of data, from geographic and address potential issues, mitigating downtime
architectural details to demographic and or disruptions.
real-time sensor information. This virtual
cityscape empowers authorities to model
and analyze different urban scenarios. They
REALIZING
can simulate the impacts of new buildings on SUSTAINABILITY:
wind flow, sunlight, and traffic or forecast the
implications of population density changes
CHALLENGES TO THE
on the city’s infrastructure and services. JOURNEY OF DIGITAL
This aids in making data-driven, informed
decisions and ensures the city grows
TWINS AND AI
sustainably. Implementing digital twins and AI for
sustainability efforts is not without
Solar potential is another area where Virtual challenges. This journey is marked by
Singapore shines. This digital twin aids hurdles that encompass data collection
authorities in identifying the best rooftops for and management, system integration,
solar panel installation, promoting renewable cybersecurity concerns, significant costs,
energy use and reducing Singapore’s carbon social, ethical considerations, and even the
footprint. In crisis situations like fires or terror potential environmental implications of the
attacks, the digital twin can quickly devise technology itself. Let’s delve into each of

70 ¦ METALAB REVIEW
© Max Langelott on Unsplash
these challenges in detail, bringing to light compatible with Maersk’s shipping operations
specific case examples. and Ericsson’s telecommunications
technology, demanding seamless integration
1. NAVIGATING THE DATA LABYRINTH and interoperability.

Data is the lifeblood of digital twin technology, 3. GUARDING THE CYBER FRONTIER
and the process of collection, standardization,
and management of this data is indeed an The heavy reliance on data of digital twin
uphill task. Take the case of Singapore’s technology and AI makes them prime
Virtual Singapore project; the creation of targets for cyber-attacks and potential data
a comprehensive digital twin of the entire breaches. The case of Amazon, which uses
city demands data inputs ranging from digital twins for re-marketing, underscores
geographical, architectural, to demographic the importance of securing customer data
and real-time sensor information. Collating against potential threats. This necessitates
and normalizing this data from varied sources the implementation of robust cybersecurity
necessitates an intricately devised data measures and a constant state of vigilance.
collection strategy, coupled with a reliable set
of data sources. 4. INVESTING IN THE DIGITAL TWIN
DREAM
2. BRIDGING THE INTEGRATION GAP
The significant costs associated with
The integration of digital twin technology implementing digital twins and AI can pose
within existing systems, especially in intricate notable challenges. Unilever’s venture into
networks like supply chains and logistics, the digital twin realm, for instance, required
often proves to be a significant barrier. substantial investments in hardware,
The digital twin initiative of Maersk and software, and training. Simultaneously, the
Ericsson provides a case in point. To obtain company had to ensure that the complexity of
valuable insights, their digital twin had to be the technology did not stymie its adoption.

METALAB REVIEW ¦ 71
5. GRAPPLING WITH SOCIAL AND Seamless integration of digital twin
ETHICAL DILEMMAS technology with existing systems can
be achieved using middleware solutions
The impact of digital twin technology and AI or APIs, which enable different software
on society is a crucial consideration. From applications to interact and exchange data.
potential job losses due to automation to These integration methods allow for efficient
concerns about the technology’s carbon utilization of data from diverse sources,
footprint, these issues cannot be overlooked. enhancing the effectiveness of digital twins
While Unilever’s use of digital twins led to in driving sustainability. To counter cyber
enhanced production efficiency, it could threats, advanced cybersecurity measures
potentially result in job losses due to like data encryption, secure access controls,
automation. and regular security audits are vital. The
protection of sensitive data and infrastructure
NAVIGATING THE is paramount, and cloud-based digital twin
platforms, with their built-in security features,
CHALLENGES: provide an extra layer of protection.
POTENTIAL SOLUTIONS Addressing social and ethical concerns is
AND EMERGING crucial in the adoption of AI. This involves
TECHNOLOGIES the creation of ethical AI frameworks and
guidelines, ensuring AI aligns with societal
The path towards sustainability using digital values and ethical norms. It is important
twins and AI is filled with challenges. However, to consider the potential impact of AI on
it is essential to understand that these employment and mitigate potential job losses
hurdles can be managed with meticulous through upskilling and reskilling strategies,
planning and execution. Addressing the fostering a workforce that can adapt and
complexities in implementing digital twins thrive alongside these technologies.
and AI for sustainability necessitates robust
strategies and pioneering technologies. An Emerging technologies offer exciting
essential part of this process is establishing possibilities for enhancing digital twin and AI
strong data governance policies that capabilities. Edge computing enables real-
streamline data collection, storage, and time data processing at the source, improving
usage, reducing time consumption and digital twin system performance and reducing
enhancing data interoperability. Automated latency. Integrating Blockchain technology
tools and standardized data formats can can heighten data security and transparency,
further facilitate this. mitigating data privacy and security issues.
Advances in machine learning and AI
While digital twins and AI offer transformative explainability are increasing the accuracy and
possibilities, it is important to recognize the interpretability of AI predictions, enhancing
potential vulnerabilities and risks associated digital twin applications in sustainability
with relying heavily on AI. Substantial efforts.
investments are necessary for successful
implementation, encompassing not only Overall, strategic planning, best practices, and
technological infrastructure but also training leveraging cutting-edge technologies form
and upskilling initiatives to ensure a skilled the crux of overcoming challenges associated
workforce capable of leveraging these with digital twins and AI for sustainability.
technologies effectively. With careful consideration of vulnerabilities,

72 ¦ METALAB REVIEW
substantial investments, and the integration technologies enhance the performance,
of emerging technologies, we can unlock the security, and effectiveness of digital twins
full potential of digital twins and AI to drive and AI, propelling us towards a future where
sustainable practices and create a better technology and sustainability go hand in
future for generations to come. hand.

STEERING TOWARDS A As organizations invest in necessary


technologies and resources, the ability to fully
SUSTAINABLE FUTURE: harness the potential of digital twins and AI to
THE PROMISE OF DIGITAL meet sustainability goals becomes a tangible
reality. This is a collective effort, one that calls
TWINS AND AI on every sector to contribute. The fusion of
digital twins and AI is not just a technological
As we look towards the future, the role revolution; it’s an essential catalyst for a
of digital twins and AI in sustainability more sustainable future. It’s not just about
efforts is becoming increasingly prominent. embracing new technology—it’s about
Technology is rapidly advancing, and with creating a sustainable world for us all.
increased adoption, organizations are finding
new ways to leverage data and analytics for
sustainable development.

The sophistication of AI and digital twins is


growing exponentially, creating more accurate
models and predictions. AI’s ability to analyze
data from multiple sources and identify
patterns and trends otherwise overlooked
proves pivotal for informed decision-making
and efficient resource allocation. This not only
drives sustainability efforts forward but also
lays a foundation for innovative solutions to
environmental challenges.

However, realizing the full potential of these


technologies isn’t without its challenges.
Data standardization, cybersecurity, cost, and
ethical considerations stand as formidable
hurdles. Yet, as demonstrated throughout
this article, these are not insurmountable.
Strategic planning, innovative solutions,
and emerging technologies are keys to
tackling these challenges. The embrace
of Edge computing and Blockchain
technology, advances in machine learning,
and AI explainability, among others, are
taking us closer to a reality where digital
twins and AI become an integral part of
sustainability strategies. These evolving

METALAB REVIEW ¦ 73
74 ¦ METALAB REVIEW
BLOCKCHAIN &

METALAB REVIEW ¦ 75
DIGITALIZATION
DIGITAL CURRENCY:
THE SUBSTITUTION
OF MONEY
By PAUL BÉDIER AND MARINA PELLET
Master in Data Sciences & Business Analytics
at ESSEC Business School & CentraleSupélec

«As a central banker interested in information first article aims to identify and compare the
technology, should I regard this prospect as a properties of Bitcoin-like crypto-assets which led
dream or a nightmare? Perhaps the answer is to their adoption by many, either for investment
that central bankers should enjoy life today» purposes or as a complement, if not replacement,
(King 1999) to fiat money.

T QUICK INSIGHTS:
his question was posed over twenty
years ago by a group of economists who
wondered whether there was a threat of • There are a lot of similarities between
obsolescence posed on central banks from the traditional currencies (fiat money) and crypto-
innovations of information technologies. While assets, in the processes that encompass
a decline in the role of central banks has not them, such as the money creation and the
been observed yet, the increasing adoption of verification process for transactions; but
crypto-assets relying on blockchain technology, there are enormous differences that oppose
and removing intermediaries from transactions, them. Crypto-assets’ creation and verification
should not be overlooked. processes are decentralized and based on
computational power, and their total supply
European Central Bank President Lagarde called can be finite and permanent.
for more regulation of crypto-assets, which
shows the breadth of the crypto phenomenon. • Whether crypto-assets can be considered
Of course, one can not ignore the recent and as a currency or not is an ongoing debate:
significant Bitcoin and Ether crashes, nor the while it is generally agreed that crypto-
undeniable and massive environmental impact of assets do not fulfill all the properties of a
crypto-assets mining. currency, economists and large financial
institutions disagree on their ability to meet
Before diving into the question of whether crytpo- the different criteria. This article also explores
assets can substitute traditional currencies, this the physical properties of money creation for

76 ¦ METALAB REVIEW
© Thought Catalog on Unsplash
crypto-assets and draws a parallel between last November, with several crypto-assets
the precious metals from which our coins now at 0 USD value (Terra, Luna). Bitcoin has
were originally made, whose value was fallen 45% since El Salvador adopted it as a
influenced by the amount of energy put into legal tender in September 2021, which could
their extraction, and the energy-consuming sink the country deeper into the economic
process of crypto mining. crisis the country has been going through
for the past several years. This situation is
• The invention of blockchain and crypto- an example of the consequences of crypto-
assets coincides with the heart of the assets’ volatility which makes it impossible
financial crisis of 2008. This new form of to qualify them as legal tenders for most
currency was invented in reaction to failing countries and their monetary institutions.
financial institutions and continues to
play a major role in countries where those • As a result of the energy-intensive
institutions cannot support the local economy computational process involved in the
(El Salvador, Venezuela, Lebanon, and many proof of work mechanism used to verify
more). new transactions, the crypto-assets’
mining industry has serious environmental
• Nevertheless, the crypto-assets market is consequences. It is estimated that Bitcoin
experiencing a major crisis. The total value alone consumes more energy than Norway
of the market fell under half of where it was annually (University of Cambridge).

METALAB REVIEW ¦ 77
78 ¦ METALAB REVIEW
© Rupixen.com on Unsplash
Additionally, every four years the amount of and since the end of the Bretton Woods
Bitcoin that rewards miners is halved. So, system and of the US Dollar’s convertibility
after each halving, the amount of carbon to gold (James 1996), the notes’ value is
emissions required to mine one coin is determined by the activity of the central
doubled. This rate is clearly not compatible banks. The world’s most used currencies
with the imperative of reducing our carbon are not backed by commodities like precious
emissions in light of the climate crisis. China metals anymore, but by the governments that
banned crypto mining from its territory issue them and the monetary policies that
in 2021, partly because it was conflicting control them.
with the climate objectives set by 2050
and also for regulatory concerns. If other In the modern economy, commercial banks
countries follow, or more strict environmental increase the amount of money in circulation
regulations come out, it is possible that crypto every time that they lend. Indeed, banks credit
mining will slow in the upcoming months their clients’ accounts with bank deposits of
and years. We have yet to see the impact this the size of the loans, creating new money. The
crypto-creation decrease will have on crypto- amount of money created by private banks is
assets value. not unlimited, it is in fact regulated by several
factors. Market forces and firms’ investment
TRADITIONAL MONEY capabilities, risk mitigation through prudential
regulation, and households’ behavior control
VERSUS CRYPTO- the process of money creation. At the top of
ASSETS: IS THE TERM the system, central banks’ monetary policies
act as the ultimate constraint - or incentive,
CRYPTOCURRENCY depending on the context - on money
JUSTIFIED? creation. To achieve this, central banks have
three tools at their disposal: the amount of
Bank money is the most abundant form minimum reserve requirements commercial
of money today and is largely already banks must hold in central bank accounts,
digitized. In fact, money has been affected the policy interest rate at which commercial
by information technology since the banks borrow and which is the building block
digitalization of account books and registers. of all other interest rates, and finally open
market transactions in which the central bank
MONEY CREATION AND MAJOR buys or sells securities in the market. The
PLAYERS IN THE MODERN ECONOMY latter has been massively expanded in its
scale in the last decade through Quantitative
Before the invention of information Easing, or massive security purchase
technology, there were only three forms of programs aiming to increase the money
money: coins, notes, and bank money. The supply and sustain economies (McLeay,
first two forms of money are emitted by Radia & Thomas 2014).
Central Banks of each monetary area, while
the third is released by private banks, through
the deposits that they hold and the credits
they offer. The exchange of coins made
of precious metals was one of the earliest
methods of payment in the modern economy,
to which gradually succeeded the notes, at
first redeemable for precious metals. Today,

METALAB REVIEW ¦ 79
A NEW FORM OF with other finite resources on earth, and the
“CURRENCY”, BORN TO consequences on their values - this point
of view on Bitcoin-like crypto-assets will be
REVOLUTIONIZE THE explored later in the article.
MONETARY SYSTEM
With the rise of the internet, a fourth form
FIAT MONEY VS.
of currency was created: cryptocurrency CRYPTO-ASSET: WHICH
- that we shall call crypto-assets because
they are not recognized as currencies by
ONE GUARANTEES THE
most governments - which differs from the HIGHEST STABILITY,
previous types by its very essence: created
and existing through computer code, it
SECURITY, AND
does not have any physical equivalent, TRANSPARENCY?
unlike the other forms discussed. They
are not controlled by any third party, but There are two major differences between
rather created through algorithms and the crypto-assets and bank money, like the Euro
blockchain - a distributed ledger technology for example. The first difference is of course
that records transactions as immutable the verification process of the transactions.
timestamps, relying on a peer to peer network For a currency like the Euro, the control of
to validate transactions through complex transactions is made possible through its
computing problem-solving. Bitcoin, created centralization within commercial banks
in 2009, is the first crypto-asset in circulation. and the European Central Bank. For Bitcoin,
The creation and transaction process are transactions are, as we discussed, public and
quite different from that of bank money. verified by miners, which can, in theory, be
Bitcoin is issued through the blockchain anyone willing to provide computing power
technology, which stores and transmits to solve complex algorithmic problems in
secure and transparent information, acting as exchange for remuneration. The second
a higher authority. New Bitcoins are created major difference is how the system’s stability
through the verification process of new and security are guaranteed. On one hand,
transactions: to be effective, each transaction they depend on a higher authority, may it
must be publicly verified by miners, through be commercial or central banks, on the
a series of complex algorithmic calculations other hand, they are based on information
which results in the storage of the transaction technology and high-level cryptographic
inside the blockchain. Mining is a laborious security. The trust component is extremely
process that requires high computing power important in guaranteeing the stability
to solve the extreme complexity of the of a currency. In countries in which the
cryptographic puzzles. To acquire Bitcoin, one government issues the currency, not backed
can either successfully mine transactions, by a physical commodity, the currency is fiat
or simply buy some on exchange platforms money. Most currencies today are fiat money.
using Euros or Dollars for example. Their value and stability lie in the central
bank’s policies and the control of inflation,
Interestingly, Bitcoin has been programmed and also in the trust that users place in them.
as a finite resource, like gold, silver, or oil: As an example, the Euro is fiat money: its
its total supply equals 21 million coins, users trust Europe’s Eurozone economy and
expected to be all in circulation in 2140. Here, monetary policy, largely guaranteed by the
there is an interesting parallel to be made European Central Bank.

80 ¦ METALAB REVIEW
© Tomáš Hustoles from Burst
Nevertheless, by their decentralized nature, diversified it is. The liquidity of a currency is
crypto-assets remove from the system the also of critical value to economic agents as
very components dedicated to maintaining they convert money into and from assets.
currency stability, and therefore trust. Bitcoin Maybe the reasons for the observed relative
supporters argue that information technology instability of crypto-assets is that they
and the blockchain are better guarantors of are hardly linked to any economic zone,
stability - of the system, the prices and the and are therefore not entrenched in any
security of transactions - than the state or underlying local economic reality. After the
monetary authorities acting as regulators, for recent crypto-asset crash of May 2022, ECB
they are not subject to political influence and President Christine Lagarde said about crypto
do not concentrate control over the money in in general that “it is worth nothing, it is based
the hands of a few, but rather give it to anyone on nothing, there is no underlying asset to act
wanting to contribute with their computing as an anchor of safety”. Of course, it is in the
power. Considering that Bitcoin was created interest of Central Banks around the world
in 2009, at the heart of the global financial to warn the populations of the speculative
crisis, we understand that this innovation nature of crypto-assets and redirect financial
appeared as a result of a loss of confidence transactions towards traditional currencies
in the financial system and a will to break free and their future digital form (central bank
from state authorities. digital currencies). Nevertheless, the question
remains: can crypto-assets claim the title of
However, the stability of a fiat currency does currencies, even when being exposed to what
not come solely from trust in the institutions seems to be inevitable instability.
backing it, but also from the economy to
which it is attached to, and how deep and

METALAB REVIEW ¦ 81
naturally debated as the high volatility that
HOW DO BITCOIN-LIKE- it is subject to makes it inconvenient for
CRYPTO-ASSETS FIT daily transactions. Yet, the role of Bitcoin as
a store of value and as a unit of account is
THE TRADITIONAL questioned more broadly. The IMF recognizes
DEFINITION OF A Bitcoin both as a medium of exchange and
as a store of value, comparable to holding
CURRENCY? precious metals, especially in countries where
central and commercial banks’ credibility is
Bitcoin and other crypto-assets are not low (IMF 2021). JP Morgan compares Bitcoin
officially recognized as currencies by the IMF to gold and shows how both are stores of
as they are not issued by central banks nor value. Others disagree, putting forward the
authorized by governments, in most cases (cf. high volatility as the main argument. Views on
later example of El Salvador). Nevertheless, the use of Bitcoin-like crypto-assets as a unit
the question of the classification of Bitcoin of account are less divided and tend to agree
and similar assets as financial or nonfinancial that their highly fluctuating value makes it
assets is debated and debatable. To impossible to express prices in Bitcoin that
determine whether an asset behaves like a would not be unstable (Banque de France,
currency, an economist’s approach would 2018).
be to ask if it functions as a medium of
exchange, a unit of account and a store of
value. Although the definition of a currency
A DIFFERENT APPROACH
seems clear, opinions on how Bitcoin is TO THE DEBATE:
defined diverge and do not find a consensus.
Different views argue about whether or not
PHYSICAL PROPERTIES
Bitcoin is a medium of exchange. Since the AND ENERGY
first good bought in Bitcoins in 2009 - a pizza,
acquired for 10,000 Bitcoins - crypto-assets However, there are other approaches to
have been used for acquiring goods and the definition of money than the three
services all over the world (Yermack 2013) characteristics seen previously. One of them is
but its overall adoption level for transactions of particular interest and could shed new light
compared to traditional currencies remains on the debate over the proper qualification
low. In theory, crypto-assets offer advantages of crypto-assets. In one word: energy. This
as mediums of exchange, amongst which approach was developed at length in Octavian
pseudonymity, as practically no information S. Ksenzhek’s work entitled Money: Virtual
is necessary to buy and transfer crypto- Energy. Economy through the Prism of
assets - although this is to be nuanced, as Thermodynamics (2007). In this specific book,
the blockchain stores all transactions ever he makes a very elegant case for the similarity
made, allowing for traceability (the monero between energy and money. Energy, like
is incidentally replacing Bitcoin in criminal money, is hard to define, and scientists, like
activities financing). economists, focus on determining its qualities
or characteristics. Perhaps the most practical
For cross-border transactions, crypto-assets way to think about energy is as the capacity
offer the advantage of cutting processing to produce work. Work here can be defined as
time from days to a few hours through an action of a system upon an external object
the lack of intermediaries, as well as low followed by the production of certain changes
transaction fees (He 2018). The quality in its structure or position. For instance, an
of Bitcoin as a medium of exchange is athlete lifting a weight is performing work on

82 ¦ METALAB REVIEW
the weight against gravity. This shift always Unsurprisingly, this approach of money
requires an energy transfer from the system to as a quantity has also been entertained
the object, which usually means a conversion of by many economists, from Monetarists to
energy into another form (from mechanical to Neo-Keynesians. For instance, Bill Philips,
electrical in a turbine for instance). most famously known for his Philips
Curve illustrating the relationship between
Now, let’s take a step back and see how unemployment and inflation, was also
relevant these concepts might be with respect interested in physical and quantitative
to human societies and economies. All goods representations of money. He built a kind
and services exchanged in economies require of early computer, simulating the economy
energy to be produced and delivered where they using water as the physical representation of
are needed. As seen previously, money quickly money flows. As such, this approach could
emerged as a necessary unified equivalent shed a new light on crypto-assets. Indeed,
allowing for practical exchanges of such goods their whole concept is based on the blockchain
and services. In primitive societies, any object and usually the Proof of Work paradigm, which
was used: shells, rams, grains and feathers. is a highly energy-intensive process. High
However, and as early as the 3rd millennium amounts of electrical energies are converted by
BC, precious metals were introduced as unified computers into information about transactions,
equivalents. Their unique properties, such as giving value to crypto-assets such as Bitcoin.
their scarcity, their durability or purity (difficult This is directly comparable to what was
to corrode) and the ease with which they can be outlined above with precious metals. If it
transformed, made them very efficient units of remains difficult to consider crypto-assets as
account, and later an ideal medium of exchange currencies from the point of view of classical
as they were turned into standardized coins. economics, the argument developed here gives
further credence to the idea that crypto-assets
Looking at it more closely, money based on are very similar to gold and other precious
precious metals share many properties with metals, which were used as currencies not so
energy: money determines the capability to long ago.
perform work (purchasing goods or services),
money is always conserved -at least if the However, this analogy, like all analogies,
economic system that supports it is stable- and remains limited. First, it does not explain
exists in different forms, and the process of how humanity has moved away from using
energy dissipation is akin to money spending currencies requiring high amounts of energy.
(as we spend money, it is getting more Fiat money, no longer tied to the value of gold,
dispersed and less useful). Furthermore, when has been widely adopted all over the world
humans extract precious metals from the earth, because of its high efficiency and despite its
they essentially transfer a massive amount of value being detached from physical properties.
energy required for the mining process into Instead, its value comes from trust in
the metal itself. Regardless of their practical institutions and the strengths of the underlying
utility for manufacturing or their social utility as economy it is linked with. Likewise, not all
wealth displays, the mere act of extracting and crypto-assets are based on the high-energy
refining metals is sufficient to give them their Proof of Work paradigm, and a lot of them are
value likely because of all the energy spent in trying to move to more energy efficient money
the process. It is therefore interesting to ponder creation and transaction processes (Proof of
on the extent to which money is, from this Stake for example). In this sense, crypto-assets
perspective, a vector of condensed energy. could be just like other historical currencies
and follow the same evolution.

METALAB REVIEW ¦ 83
THE CASE OF One step further, El Salvador became the
FINANCIALLY UNSTABLE first country to recognize Bitcoin as a legal
tender, to facilitate international remittances,
COUNTRIES, WHERE THE tackle the problem of underbanking and
TRUST COMPONENT IS mitigate the risk of dollarization (Arslanian
and others 2021). The lesson is that crypto
ERODED alternatives are positioning themselves as a
great resource to fill the gaps in weakened
As mentioned previously, markets in which economies. We can wonder whether the
the central bank has limited credibility and the country will rule out this payment option
banking system is weak - typically financially once/if local fiat money finds more stability
unstable countries - see users consider and people regain trust in monetary
crypto-assets as a store of value on top of a institutions, or whether the population
medium of exchange, and thus drive adoption will adopt crypto-assets in the long term
rates up at a faster speed than in more stable and stick to its use even in a more stable
financial markets. There are many examples environment. The answer is probably not,
of countries where the population, unable as naturally the majority of people are
to access the most basic financial products risk-averse and will prefer stability over
like bank accounts, turn towards crypto- speculation. Nevertheless, as we saw, trust is
assets. Financial instability can be the result at the base of a strong economy, and people
of political instability and/or armed conflict, who were let down by their government and
erosion of trust in the financial system their monetary institutions might place more
(1997 Asian crisis), natural catastrophes or trust in crypto-assets despite higher volatility.
health crisis, and Bitcoin repeatedly plays We can imagine that owning crypto-assets
the unprecedented role of the resource of could for instance be a risk mitigation or
reference in times of crisis. The return of hedging strategy in countries with heavy
the Taliban to Afghanistan in August 2021 economic distrust in the past, leading to
resulted in a national cash shortage, closed limited demand for central bank money,
borders, and a plunging local currency. which would then have difficulty setting
Several banks closed their doors and most interest rates freely. Just like dollarization can
money transfer services stopped functioning. be a threat to local economies in emerging
What resources did people turn to as their countries with weak foreign exchange and
government and monetary institutions monetary policies, cryptoization represents
were failing them? The internet and crypto- a risk for macro-financial stability, consumer
assets, inevitably. The country saw a sharp protection, and the environment (IMF 2021).
rise in crypto adoption following the start
of the conflict. Similarly, the mass adoption
of crypto-assets in Lebanon resulted from
central bank mismanagement and corruption,
and therefore an important loss of trust in the
system. A somewhat different example is that
of Ukraine, which started accepting crypto
donations in February as tensions with Russia
rose, even before the war started. Indeed,
the speed of transfers and the ability to cope
with the rising inflation were all convincing
arguments to turn towards crypto-assets.

84 ¦ METALAB REVIEW
CLOSING THOUGHTS
While economists, digital experts, and
governments endlessly debate its nature and
whether to regulate it, crypto is undeniably
making a significant impact on the entire
world’s economic system and is already
instrumental in many further developments
around the digitization of the economy (NFTs,
metaverses and decentralized finance).
Precisely because of their difference with
regard to traditional currencies, they are
unlikely to replace them entirely, at least for as
long as sovereign states will exist. However,
by removing part of the issue of trust inherent
to any centralized currency system, they
offer a useful alternative and complement
to fiat money, as seen in unstable countries
or in some specific ecosystems where their
decentralized nature creates value.

There is no doubt that crypto will continue


to evolve on adjacent paths to traditional
currencies, perhaps with similar historical
trajectories as explored in this article, and
hopefully in a more sustainable direction both
from a social and environmental standpoint.
What is certain is that current financial
institutions and central banks are paying a
lot of attention to these new assets and how
they might challenge their role and place
in the economy. Following this trend, many
central banks are hard at work on the creation
of digital currencies and exploring how best
to integrate them into the actual financial
system, as we shall explore in our next issue.

METALAB REVIEW ¦ 85
DIGITAL CURRENCY:
THE OVERHAUL OF
PUBLIC MONEY
WHY DO WE NEED A CENTRAL BANK
DIGITAL CURRENCY WHEN MONEY IS
ALREADY DIGITAL?

By PAUL BÉDIER - Master in Data Sciences & Business Analytics at


ESSEC Business School & CentraleSupélec

BRAND NEW ACRONYM, through a game of accounting entries that


SAME OLD IDEA? transactions are carried out. Tangible money

C
like coins or bills is estimated to account for
BDC: the new 4-letter acronym only 3 to 8%1 of the world money supply (with
that will no doubt be the subject of variations among countries). Why then do
heated disagreements for the years we keep hearing about digital currency as if it
to come. It stands for “Central Bank Digital were this great transition we are on the verge
Currency.” For the average reader, the central of?
bank rhymes with monetary policy and
interest rates while digital currency would The answer is simple: central banks have
be associated with fintech companies and been ramping up the communication about
the crypto sphere. Confused? Let us attempt their Digital Currency projects, hence we are
to clear the haze surrounding the buzz and increasingly hearing about it. As of 2023, 130
jargon. First, think about digital currency for countries2 (representing 98% of world GDP)
a minute. Most people should be aware that are in the process of evaluating the launch of
an overwhelming majority of money already CBDC projects. Eleven countries have already
sits on digital ledgers: bank accounts. It is launched theirs. What are these projects

1. https://occaminvesting.co.uk/how-much-money-is-in-the-world/
2. https://www.atlanticcouncil.org/cbdctracker/

86 ¦ METALAB REVIEW
© Pixabay
about? They can take different forms, but central bank just like banknotes and coins. So,
all revolve around the idea of transforming the definition of CBDC is simply the following:
centrally issued banknotes and coins in a “A Central Bank Digital Currency is the digital
digital format and making them accessible form of a country’s fiat currency.”
to the public for transactions at its most
basic level, or for savings and other financial After this explanation, you may feel a bit
operations at a more advanced level. underwhelmed. We will simply have digital
banknotes and coins now, why not. Does an
Here you might stop and say: “well how is that old idea already materialized in our world
any revolution? I already have a bank account really warrant a brand-new acronym?
with digital money inside it and I use my credit
card instead of bills to settle transactions”. It Yes, it does, because the distinction raised
is indeed a subtle but undeniable revolution. above between private and public money is
The main distinction between what we have critical. Indeed, CBDCs are groundbreaking
now and CBDCs is the same distinction that for two main reasons. First, the motivations
has always existed between private and behind them are numerous and diverse, and
public. The digital money you currently use find their roots in questions of economic
through your credit card is issued by a private stability, financial efficiency, international
bank, while CBDCs would be issued by the geopolitics, and even sociopolitical objectives.

METALAB REVIEW ¦ 87
Second, their mere implementation has the better society and improve financial services.
potential to be a can of worms and may One of its most noble impacts would be
compromise the financial stability they are to provide each citizen with a universal
meant to help with. By their nature of being bank account at the country’s central bank,
digital currencies, they will intrinsically carry bringing basic and low-cost banking services
with them the potential for many advanced to all. The positive repercussions of such a
functions, through their programmability move, especially in developing countries with
for instance, which might enter in direct high unbanked rates, could be staggering.
competition with the private financial - and
technological - sector. We will see that for Furthermore, CBDCs would bring with them
these reasons, different countries have a number of technological innovations,
different plans for CBDCs. pushing efficiency upward for both
individuals and businesses. Most of these
Now comes the question we will strive to innovations already exist, propelled by
address in this article. Will CBDCs be useful to fintech companies, but they would take an
society? Are they necessary? The answer is even wider scale within the fiat currency
not straightforward. Let us dig into this! infrastructure provided by central banks. With
intermediaries such as commercial banks
MOTIVATIONS BEHIND or clearing houses eliminated, real-time
payments and settlements could take place,
CBDCS, A TANGLED WEB thus reducing financial risk, complexity and
in turn lowering transaction costs. Proof of
The idea of CBDC certainly is not new, transaction in the form of a digital record
and one tentative attempt at identifying located either on a distributed ledger (yet still
its original author attributes it to Nobel centrally controlled) or a centralized database
laureate James Tobin3 who in 1987 already would foster trust and transparency in the
advocated for a public payment “medium financial system, making it more difficult for
with the convenience of deposits and the conflicts to arise and for fraudulent activities
safety of currency”. Ignored for a long time, to take place. Finally, CBDCs would allow for
the proposal regained traction in the recent easier programmability and bring with them
years, with central bank and public officials a large number of beneficial features for the
warming to it as a solution to rising issues: financial system. The best and most famous
the decline of cash, exacerbated during example is with smart contracts, which can
the COVID crisis, and the relative spread be automatically settled and remove much
outside of the traditional monetary system of of the counterparty risk usually involved in
cryptocurrencies like Bitcoin or, perhaps even transactions.
more concerning, stable coins backed by tech
giants such as Libra (Diem). Another motivation for CBDCs, and one that
should please proponents of the neoclassical
FOR SOCIETY AND BUSINESSES: school of economics, is the heightened
IMPROVE FINANCIAL INCLUSION & competition between private banks that
MONETARY EFFICIENCY would ensue a CBDC launch, as they would
need to offer better services and interest rates
On paper, motivations behind launching a to their customers for fear of losing them
CBDC could come from a genuine objective to to the central bank. Now, this is a hot-issue

3. James Tobin, Essays in Economics (MIT Press), 1987

88 ¦ METALAB REVIEW
topic and there is dissent among both central established financial system. Finally, cash
banks and public officials as some fear this decline could now turn into full-fledged cash
could backfire. Indeed, they argue that by disappearance after the fintech industry filled
eroding the banking sector’s profitability and the holes created by the COVID crisis, leaving
encouraging financial disintermediation, the our societies entirely reliant on private forms
consequences for the economy at large in of money.
terms of efficiency and innovation could be
detrimental. Let us delve more into this issue CBDCs could be the answer central banks are
later in the article. looking for, as they would intrinsically bring
back public money in the economy, reaffirm
FOR CENTRAL BANKS: REGAIN their authority over the monetary system and
MONETARY AUTHORITY even grant them new vectors of monetary
policy. For the latter, this could simply take
Central banks have clear motivations behind the form of direct transfers to the public
CBDCs, which boil down to reaffirming their with public money issuance, or so-called
authority over money. Indeed, while they “helicopter money.” More refined approaches
used to be fine relinquishing money issuance could involve programmability of money, say
activities to private banks as long as it meant implementing expiration dates which would
they could focus on monetary policy through spur spendings and discourage savings –
their direct and indirect control - required something that was assessed by China in their
reserves, policy rates, open market operations pilot program in Shenzhen. Finally, CBDCs
to name the main tools at the central bank’s could help central banks enhance financial
disposal. In fact, this proved quite successful stability by offering an alternative to fractional
all things considered, as the commercial banks reserve banking and thus limiting the risk
could compete with one another and innovate and impact of bank runs, and by avoiding a
to win over customers, while the central banks potential flight to unregulated and volatile
essentially outsourced financial intermediation cryptocurrencies or stable coins.
and credit issuance activities. The decline of
cash - public money - to a baseline level was Now, of course the claim central bank’s
merely an accepted consequence. authority was eroding to start with should
be nuanced. In the last decade, institutions
However, this system is now being challenged like the Federal Reserve of the United States
from different sides with the spread of of America (“Fed”), the European Central
alternatives to fiat money: cryptocurrencies Bank (“ECB”), Bank of England, People’s
and stable coins. The former, coming from Bank of China (“PBoC”), and many more
society’s underground, is a threat to the central have unleashed massive quantitative easing
bank’s authority and control over monetary programs in the wake of the 2008 market
policy and in their eyes a source of volatility crash, overflowing the market with liquidities
arising from unregulated speculation and in support of their respective economies.
therefore a risk to global financial stability. Recently, most of them have started a series
However, cryptocurrency adoption remains of policy rate hikes to fight inflation caused by
low and its viability as a fiat money alternative COVID crisis recovery and the Russian-Ukraine
is very debatable, as explored in the previous war supply shock. This would tend to show
article of this series. The latter, coming from central banks are still very much in control of
tech giants and the fintech industry, has the the monetary system and that their authority
potential to become an even larger threat over the matter was not threatened quite as
to central banks by short circuiting the much as previously implied.

METALAB REVIEW ¦ 89
© Sarah Pflug from Burst
However, much can be told from how the individuals, businesses, and central banks.
central bank’s own stance on the matter But they also have specific interests in the
shifted at the turn of the COVID crisis. Indeed, launch of Central Bank Digital Currencies.
up to 2020, the consensus towards CBDCs A domain in which they could be helpful to
was still fairly negative, or at least erring on the state is the fight against illicit activities.
the side of high caution. Now, the tone is a Indeed, a digital repository theoretically
lot more positive, with central banks eager makes them able to keep track of the
to communicate on their CBDC projects and exact location of each monetary unit of the
deliberations. This is primarily the result currency. This could have repercussions
from a better understanding of the impact for tax collection, making evasion more
of digital currency’s implementation on the difficult through means of offshore banking
existing financial system, and how it could or unreported employment via underground
be mitigated. But this is also due to renewed economies. It also makes it easier to spot
interest as central banks now would not reject criminal activities, track money laundering
an increase in the scope of their policies nor and reverse fraudulent transactions. Of
access to additional tools. Especially when course, all these benefits depend on the type
economists are increasingly pointing out of implementation chosen by the central
the risk of a decoupling between traditional bank, the amount of transparency in the
monetary policy and the real economy. system and the status of cash in the new
system.
FOR SOVEREIGN STATES: COMPETE IN
THE MONETARY ARMS RACE Another game is however at play here, with
states fighting in the realm of geopolitics
Sovereign states presumably act in their and international competition. Indeed, like
citizen’s interest and thus would be aligned any significant technological breakthrough –
with all the motivations listed above for nuclear weapons, advanced semiconductor

90 ¦ METALAB REVIEW
chips, artificial intelligence, the potential cost from this historic role. It does not take long
of not developing it may be too high and to see how the US government might fight
countries will get caught up in the infamous against this and feel obligated to pursue
prisoner’s dilemma. You might ask: “what CBDC developments as a hedge against
are the risks in the case of CBDCs, they this risk. Consider also how China is part of
appear harmless enough.” But this would the five countries who launched their CBDC
be discounting the scale and significance pilot programs, while the US remains at
of economic warfare. One of the risks to preliminary research and deliberations, and
a country is simply to get overrun by the you might sense uneasiness5.
financial innovations provided by the CBDC
of another country. This could manifest Now that we have outlined the main
in increased capital flights, with investors motivations at play here, let us delve a bit
flocking to the other country’s financial deeper into the potential implementation of
markets to benefit from increased efficiency CBDCs and how they must find a difficult
and additional features, as well as individuals balance.
using the foreign CBDC as a safe asset.
This in turn would have direct detrimental
consequences on the economy, as the central
PUBLIC AND PRIVATE
bank would need to adjust policy rates MONEY, A FRAGILE
in ways which might adversely affect the
economy.
BALANCE
THE GREAT ILLUSION
The highest and most critical risk waiting at
the end of this path is an outright currency Central banks are great contenders for the
substitution. You may have heard of the title of most impressive magician out there.
concept of “dollarization.” which occurs when Not because they seemingly create money
a country’s citizens partially or fully rely on out of thin air or make it disappear just
a foreign currency (historically it often was as easily, although that is certainly a neat
the dollar, hence the term). Of course, this trick. No, it is because they have created
can have catastrophic consequences for and maintained one of the longest running
sovereignty and economic independence, illusions in the history of humanity, on a wide
something that most countries want to scale. What is this illusion? We may call it
preserve against. But even the simple erosion the “illusion of one money.” Ask yourself,
of a country’s monetary influence on the is there any significant difference between
world can be perceived as a nightmare for the currency you hold in your wallet and the
said country. Imagine a world where the one you hold in your bank account? If your
USA loses the dollar as the most influential answer is no, then you were successfully
currency and can no longer use it to write tricked. Indeed, as we have touched upon in
the rules of international trade (USD is the introduction, the former is public money
currently estimated to represent 90% of total issued by the central bank, while the latter is
foreign exchange volume4). CBDCs, with their private money issued by a commercial bank.
highly efficient cross-border transaction This subtlety is critical, as it stands at the
settlement potential could displace the USD core of the existing financial system.

4. https://www.bis.org/publ/qtrpdf/r_qt2212x.htm
5. https://www.intereconomics.eu/contents/year/2023/number/4/article/the-geopolitics-of-central-bank-digital-
currencies.html

METALAB REVIEW ¦ 91
How do central banks manage this? By remarkably close to what a wholesale CBDC
their role as a “lender of last resort,” they would be, minus the technology allowing
guarantee immediate convertibility of private for instantaneous settlements. Retail CBDC
banks deposits into fiat – public – money. In on another hand is a much more significant
essence, we collectively place our trust in this project to grant access to digital public
state institution and the result is that we can money to the general public and is what we
benefit from the private banking industry’s have been describing in this article so far.
services on a scale that the central bank Note that hybrid projects exist, which would
would not be able to offer on its own. To the work to implement both in one package.
point where, in stable countries, we barely
ponder the difference. Reality only comes THE “BENEVOLENT” INTERMEDIARY
back to hit us in the case of a bankruptcy,
with the infamous bank run scenarios. A To better understand the fragile balance
good analogy, although slightly provocative, between public and private forms of money,
is the casino one. Depositing your money at a one must be aware of the money creation
commercial bank is like entering a casino and process. Today’s banking system is organized
exchanging your money against chips6. These as “fractional reserve banking.” This means
chips are a form of “I owe you,” legal tenders that commercial banks, as credit institutions,
for any transaction inside the casino and have near free reign to create money out of
ones that bind the casino to exchange them thin air by crediting your accounts. This is
back to you for currency. However, if you go a loan, or debt instrument, and is how the
outside and try to pay for your groceries with money supply is expanded. Of course, loans
your chips, you will likely raise an eyebrow. should eventually be repaid, and banks want
The illusion of one money is broken. In the to make a profit through interests. This
case of the banking system, interconnection means that lending money to individuals
between banks helps maintain the illusion or businesses involves credit analysis and
further, allowing us to use our “private chips” due diligence over the borrower’s solvency.
almost anywhere. These interconnections are Furthermore, banks do not have completely
made possible by clearing house systems, free reign over lending, as they are required
on which a digital form of public money is to hold reserves at the central bank. This
automatically and periodically compensated, consists of a fraction of their deposits,
balancing out overall private bank accounts hence fractional reserve banking. The role
through required reserves. of central banks is to control the money
supply through different instruments,
Now is the time to introduce the distinction required reserves ratios being one. This has
between wholesale and retail CBDC. The an impact on the dynamics of the economy
former is an evolution of the process – supply expansion, or heightened lending
described above. The central bank issues activities, will stimulate investments and GDP
a digital currency, which private banks hold growth, while supply contraction might harm
and can use to settle with one another in the economy but will exercise downward
real time. As such, interbank transfers could pressure on inflation rates.
become much faster than they are today. A
wholesale CBDC would also facilitate cross- Private banks thus play the role of the
border transactions. Today, the public form intermediary in the system, managing
of money that banks exchange is already deposits and granting loans, on a scale

6. https://brettscott.substack.com/p/casino-chip-cashless-society

92 ¦ METALAB REVIEW
have public money play a renewed and
strengthened role in the economy. The right
equilibrium is however hard to pinpoint. A
significant risk which made most central
banks frisky about CBDCs for years was
that CBDCs would lead to the uncontrolled
disintermediation of the system. Because
of the potential attractiveness of central
deposits to the public due to factors like
safety, liquidity and solvability, the launch
of CBDCs could turn people away from

© Matthew Henry from Burst


commercial banks, deteriorating their balance
sheet and causing the risk of bank runs.
Such a disintermediation would also be
catastrophic for the economy, as the activity
is fueled by investments made possible by
loans. This might sound overly apocalyptic,
that the central bank would not be able as there is reason to believe an introduction
to sustain. Of course, they do not do this of CBDC would not substitute commercial
out of benevolence, but profit seeking bank’s deposits overnight. After all, many
(or return objectives in the case of credit less financially literate people would have
unions). This system promotes competition trouble seeing the difference and the benefits
between banks, which in theory is beneficial of switching. Nevertheless, central banks
to customers through lower interest rates considered this a serious enough issue to
and better financial services. Financial pull the brakes on CBDCs and launch in-
innovation is also fostered, leading to depth review programs and impact studies.
more efficient payment and transaction The critical point is that even if a large-scale
settlement schemes. The convenience of bank run is avoided, even a slow but steady
using these payment schemes has pushed disintermediation of the financial system
back the use of cash. In recent years, new could be detrimental.
entrants have progressively challenged
the traditional banking landscape, mostly Reaching the right equilibrium will be a crucial
coming from the tech industry with both element in the choice of CBDC deployed.
large companies – think Google and Apple Wholesale CBDCs present no risk, being a
– and smaller “fintech” startups – think neo simple evolution of the current interbank
banks, international transfer, decentralized system. Retail CBDCs present a higher risk
finance. This is further eroding cash use, and and warrant more caution. In recent years,
leads central banks to worry as, compared to central banks task forces have examined
banks, they do not yet have a large regulatory hybrid infrastructure which promises to solve
apparatus for tech companies. In a sense, this problem if well calibrated. Restrictions on
today the balance already weighs heavily in deposits and CBDC programmability could
favor of private forms of money. be used to keep the private banking sector
attractive and mitigate the disintermediation
THE ELUSIVE EQUILIBRIUM risk. To conclude, the main approach should
be caution, something that central banks
By issuing CBDCs, central banks would understand well. In a report by the Bank for
attempt to rebalance the scale and International Settlements co-authored by

METALAB REVIEW ¦ 93
seven major central banks published in 2020, central banks include privacy walls into the
they outlined a Hippocratic Oath for CBDC: system, everyone is at risk of seeing their
“First, do no harm”7 . financial data used beyond their consent. In a
more extreme case, a full-fledged surveillance
Now that we have explored the motivations state could develop, monitoring its citizens’
and explained the challenges, we have the transactions and enforcing limitations on
required ingredients to answer our original certain activities via the programmability of
question: “will CBDC be useful?” the CBDC e.g., prohibiting purchase of alcohol
on weekdays or donations to politically
USEFULNESS, A MATTER opposed NGOs. These are valid concerns
and would resonate well with the quote
OF PERSPECTIVES “technological progress has merely provided
us with more efficient means for going
Usefulness, like beauty, is in the eye of the backwards” from Aldous Huxley8, writer of the
beholder. famous Brave New World dystopian novel. We
let it up to the reader to assess the likelihood
FOR INDIVIDUALS, VALID CONCERNS of this risk. Central banks will certainly need
to balance the safeguarding of privacy rights
While individuals should be excited at the and the transparency necessary to deter
prospect of increased efficiency, better criminal activity. In any case, it seems the
stability, and new financial services, many usefulness of CBDCs to the public will depend
argue the cons outweigh the pros. Retail on the specific implementation chosen. It
CBDCs are seen as yet another intrusion is our view that central banks should strive
of the state into people’s lives. Indeed, the to provide all reassurance required by being
aim of every CBDC project is a centralized transparent and including tight checks and
currency, even those that will use distributed balances to foster trust in the population.
ledger technology for efficiency, meaning
the central bank will always have full control FOR COUNTRIES, DIFFERENT PATHS
over the network. This gives a large amount
of power to the state via the central bank to Let us conclude this article by stating that
add or remove money from accounts. More there will not be one version of CBDC,
concerning and realistic for non-corrupt there will be as many versions as there are
states, the risk of cyberattacks with hackers countries implementing them. Each will have
taking control of the central authority and its own specific characteristics, trying to
using it to modify accounts should not be address the specific needs of the country but
thrown aside too quickly. also rooted in that country’s sociopolitical
ideology. China, for instance, already released
Privacy concerns are obviously also a to the public a pilot version of its CBDC,
significant issue for many individuals. If the the digital yuan, back in August 2020. The
CBDC implementation gives full visibility currency has been progressively expanded to
over financial transactions and records to more regions, reaching 261 million of users
the state, which people argue is likely even if for USD 13.8 billion worth of transactions

7. https://www.bis.org/publ/othp33.pdf
8. Aldous Huxley, Ends and Mean, 1937

94 ¦ METALAB REVIEW
at the end of 20219 – a moderate result, Because individual and national interests may
likely explained by the country’s pre-existing not align neatly, there is no doubt citizens
advance in digital payment services. The of different countries will have diverging
PBoC states that the objectives are to perspectives on the usefulness of CBDCs
enhance efficiency and help the state fight based on their nation’s chosen path.
against illegal activities. It also defends itself
regarding privacy concerns and surveillance
accusations, adding that they have included
CONCLUSION
limitations on tracking as part of “anonymity Throughout this article, our aim was to give
control.” the reader a better understanding of Central
Bank Digital Currencies. By exploring the
On another hand, for major economies like numerous motivations behind them, by
the US or the EU, the implementation of CBDC delving into the issue of the public-private
will be much slower, as project calibrations money balance and assessing the usefulness
are carried over the coming years and central of such projects for all the involved parties.
bank laws will need to be ratified. The Fed Our view is that central banks should move
has been involved since 2022 in extensive with caution and high transparency if these
research projects in partnership with the projects are to have a controlled impact on
Massachusetts Institute of Technology to the existing financial system and be widely
assess potential implementations of CBDC as accepted by the population. However, the risk
well as their impact10. The US will most likely of taking too much time to develop CBDCs in
favor a hybrid system, with great incentives the context of a global economic arms race
given to users to keep using the private could prove dangerous to national interests.
banking sector in order to protect this leading As such, central banks and the governments
industry. They will certainly incorporate into backing them should ensure to chart an
the project the ability for the central bank to efficient and resilient path towards CBDC.
directly transfer funds to deposit accounts,
as an additional tool of monetary policy. The We hope that this article will empower the
EU, champion of data protection regulations, reader to think about this technological
will no doubt spend a considerable amount progress in a nuanced and complete
of time to reassure its citizens over the full manner. By self-appropriating the issue and
transparency and limited tracking ability of developing a critical perspective, one can
the central bank. The ECB Governing Council participate in the broader debate and hold
will review the outcome of its first large-scale governments responsible for an appropriate
investigation phase, as part of the “Report and beneficial development of Central Bank
on a digital euro” during Autumn 2023 and Digital Currencies.
decide whether to carry forward with a
realization phase11.

9. https://www.atlanticcouncil.org/blogs/econographics/a-report-card-on-chinas-central-bank-digital-currency-the-e-cny/
10. https://www.bostonfed.org/news-and-events/press-releases/2022/frbb-and-mit-open-cbdc-phase-one.aspx
11. https://www.ecb.europa.eu/paym/digital_euro/investigation/governance/shared/files/ecb.degov230713-fourth-
progress-report-digital-euro-investigation-phase.en.pdf?704b0eee4c20eee4dbe4970f5091a96a

METALAB REVIEW ¦ 95
PATHWAY TO A
FAIR METAVERSE
By PIETER JAN MOTMANS
Master in Data Sciences & Business Analytics
at ESSEC Business School & CentraleSupélec

L
et’s start with a little thought their advertisements. Meta therefore has an
experiment. When a digital platform incentive to not only predict, but also modify
shows you an ad for a product or service the behaviour of users to optimise their
that really interests you, does the platform response to offered advertisements. Users
help you find what you are looking for or therefore trade their data and attention for
does it try to influence your preferences? The access to Meta’s services.
answer of course depends on who you ask,
but an increasing number of people believe This analysis becomes highly relevant when
that those platforms try to gain profits by exploring the consequences of the Metaverse,
changing user behaviour. a virtual world populated by avatars that users
can create. In the Metaverse, the scale of data
The recent increase in the amount of data collection will increase enormously. To provide
and the proliferation of algorithms based an illustration of how far this can go: Meta has
on these data have given rise to a new patented technology to track eye-movement
form of capitalism based on surveillance and brain activity through their Oculus VR-set.
of consumers. Surveillance capitalists1,
hereafter called gatekeeper platforms, can This article starts from the premise that
modify user behaviour to create profit from surveillance capitalism is to a certain extent
sales or advertising. How this happens is undesirable for society. The way gatekeeper
well illustrated by looking at how Meta earns platforms are incentivised to collect more
money from for example Facebook and data with higher predictive value conflicts with
Instagram. human privacy and autonomous decision
making. How exactly does this happen?
Users do not pay for the services offered Tracking users’ internet activity across
by Meta, it is the advertisers who pay Meta websites or tracking brain activity is arguably
proportionally to how well users respond to a breach of personal privacy that comes close

1. The definition of surveillance capitalists is not limited to big platforms, but for this article the term gatekeeper platforms
is more fitting. Gatekeeper platforms are defined in the Digital Markets Act based on market capitalisation, annual
turnover and number of users.

96 ¦ METALAB REVIEW
© Leonardo Sanches on Unsplash
to surveillance. As stated before, this data is WALLED GARDENS
used to influence purchase or (in the case of
Cambridge Analytica) voting decisions. To Walled Gardens is a term used to refer to digital
come back to the initial thought experiment: ecosystems created by some gatekeepers. It is
people who feel that digital platforms influence easy to enter the ecosystem, but once inside,
preferences have a solid argument. users will find it hard to exit.

In what follows, the focus is on one aspect The ecosystem created by Apple is a good
that enables gatekeeper platforms to thrive: example. Once a consumer buys an Apple
most of them create a so-called walled garden device, it becomes more convenient for them
ecosystem. After setting the stage, we will to purchase other Apple products. Once in
explore pathways to a Metaverse without possession of a MacBook, Iphone and Ipad, it
walled gardens. The hope is that by opening will be inconvenient to switch to another brand
these ecosystems up to fairer competition, due to the difference in operating systems
consumers will get more decision power to step that make communication and data transfer
away from gatekeeper platforms that misuse cumbersome with devices outside of the Apple
their data. In case this works, firms would get ecosystem. This goes for the chargers that are
a new set of incentives that align more closely different, to data that is stored on Apple servers.
with consumer interests. Within the Apple ecosystem, the company has
complete control over the consumer’s data.

METALAB REVIEW ¦ 97
© Nickross2021 from Pixabay
This convenience comes at a cost though. all digital advertising revenue generated in
Due to the power Apple has over the mobile the United States. These Big Tech firms are
game market, the company charges app in full control of the ecosystems they have
developers a commission of 30% on every created. They succeed very well in offering
transaction made through the app store, an convenience to consumers. Once you start
issue that was recently brought to court2. buying products through Amazon, they will
App developers currently have very few become increasingly better at offering the
alternatives to bring their apps to consumers. products that you respond to. The more
convenient it is for consumers, the harder
Another example can be found in the it becomes to switch to an alternative.
advertising business, where Meta, Google and Consumers that find this undesirable should
Amazon combine for a total share of 60% of be able to switch to alternatives3.

2. An initial verdict stated that Apple is not a monopolist, but should allow third-party payment options on their app store.
The decision has been appealed.
3. For those interested in possible solutions to this challenge, I suggest The Ownership of Data by Wilfried-Sand Zantman
and Anastasios Dosis. It explores different ways to increase welfare when faced with the trade-off between data
monetisation and privacy.

98 ¦ METALAB REVIEW
THE METAVERSE into Meta speaks for itself and some of
Google’s investments like virtual assistants
The main idea behind the Metaverse can be and wearables seem to fit perfectly with the
summarised by the desire to enable as many Metaverse.
activities as possible (work, social gatherings,
events…) from the physical world to happen By starting this article by explaining the
in the digital world. Thereby removing certain concept of Surveillance Capitalism, I hope
frictions present in the physical world like to have made the reader aware that this
distance and space limitations. In this form of capitalism poses certain threats, but
endeavour, certain characteristics emerge also that it will pose many more in case the
that should be fulfilled before we can speak of Metaverse becomes mainstream. Initiatives
a Metaverse. that could tear down one of the preconditions
of surveillance capitalism are therefore worth
1. It should be persistent. That is, it should exploring.
feel like reality, it shouldn’t end at a certain
point, neither should it pause. Most
games now pause when the user exits, a
THE DIGITAL MARKETS
Metaverse shouldn’t. ACT & THE DIGITAL
2. It should be synchronous, enabling an
SERVICES ACT
unlimited number of users to interact in real Recently, the European council and
time parliament reached an agreement on the
Digital Markets Act (DMA) and the Digital
3. There should be an economy in which Services act (DSA)4. The DMA imposes
goods and services are exchanged. new obligations on so-called gatekeeper
platforms, mostly US firms, that provide core
4. There should be interoperability of data ad platform services such as search engines,
digital assets, allowing items to be used social networking services and advertising
across the whole metaverse. services. The rationale is that due to
advantageous conditions like network effects,
Each characteristic raises certain challenges. lock-in effects and economies of scale,
Solving these will take time, that’s why many certain industries are now characterised by
people believe that the Metaverse is not yet weak contestability of the dominant firm and
around the corner. To ensure synchronous unfair business practices towards users.
experiences for example, millions of people The DSA stresses that “what is illegal offline,
should be connected in real time, current should be illegal online”. It tackles illegal
infrastructure has problems connecting more content, and imposes transparency regarding
than a hundred. advertising and recommender systems.
Users should for example be able to choose
That being said, some of the biggest the data that recommender systems can use
companies in the world are heavily investing to make recommendations.
in an attempt to not get left behind. Microsoft
will try to extend its dominance for work- The DSA’s main focus is on limiting the
related activities, Facebook’s name change negative consequences of gatekeeper

4. This article is based on the text of the proposal for the DSA and the text of the DMA that was approved by negotiators
from the European parliament and council. Both are not yet definitive and therefore subject to change.

METALAB REVIEW ¦ 99
platforms. The aim of this article and the DMA Article 5.2b): “The gatekeeper shall not
is to explore pathways to an economy where combine personal data from the relevant
gatekeeper platforms face more competition. core platform service with personal data
In what remains, the focus will therefore be on from other core platform services or
the DMA. However, I deemed it necessary to from any other services provided by the
also mention the DSA, as both should be seen gatekeeper or with personal data from third-
as a unified effort to make digital ecosystems party services.”
a fair and safe space.
Article 5.2b) gets to the core of what
The DMA hopes to open up the walled garden gatekeeper platforms do: use personal data
ecosystems to allow more competition, to to predict and possibly change behavior.
offer more choice to users and to ensure Because these companies are the owners
fairness. Non-compliance can lead to fines of a whole ecosystem, they aggregate
of up to 20% of turnover in case of repeated data from various sources. The aim of this
infringement, in comparison to 4% in case of obligation is to limit the extent to which
GDPR breaches. For Google and Meta, this these platforms can create walled gardens.
could respectively go up to $50 billion and Practically, data collected from the different
$23 billion. services integrated in the Metaverse should
be stored separately.
Because the DMA is not designed specifically
for the Metaverse, questions arise about Article 5.5): “The gatekeeper shall allow end
how the Metaverse fits into this regulation. users to access and use, through its core
The DMA applies to ‘core platform services platform services, content, subscriptions,
provided or offered by gatekeepers to features or other items, by using the
business users in the Union or end users software application of a business user,
established or located in the Union’ including where those end users acquired
(article 1.2.). The DMA does not contain a such items from the relevant business user
general definition of core platform services. without using the core platform services of
Rather, article 2.2. lists types of services such the gatekeeper.”
as social networking services, search engines,
intermediation services etc., most of which This clause gives developers access to the
are based on existing legislative definitions. users of gatekeeper platforms without having to
The Metaverse aims to integrate various core necessarily use all the gatekeeper’s payment or
platform services and to add new, not yet advertising services. The problem Epic Games
identified services. Therefore, it remains to be raised about Apple’s commissions is tackled
seen how the regulation will apply. Especially in this way. By imposing this, an opportunity is
since article 19 states that the Commission created for small companies to monetise their
may add core platform services after creations.
conducting market investigations.
Article 6.5): “The gatekeeper shall not treat
For this article, we will examine some of the more favorably, in ranking and related
obligations and outline their impact on the indexing and crawling, services and products
Metaverse. In this way, the aim of the DMA offered by the gatekeeper itself than similar
becomes clear. services or products of a third party. The
gatekeeper shall apply transparent, fair
and non-discriminatory conditions to such
ranking and related indexing and crawling.”

100 ¦ METALAB REVIEW


© Matthew Henry from Burst
In the past, gatekeeper platforms have been However, questions remain about its global
blamed for favouring their own products. The impact. The DMA is a European piece of
DMA clearly sets out to eliminate this unfair legislation which will mainly impact US firms,
practice. It is complementary to the previous it therefore seems unlikely that the United
article as third-party products will gain States will replicate it domestically. It remains
visibility. to be seen how other countries will react.

By zooming in on these three clauses, the


aim of the DMA becomes clear. It hopes to
CONCLUSION
make markets fairer and more competitive The Metaverse is as of now uncharted
for existing firms, new entrants and business/ territory with enormous potential when it
end users. At the same time, the DMA looks comes to profit and data collection. Ensuring
to change the current incentive system in fair competition is therefore of the utmost
the digital world. If it becomes public that importance. The DMA might prove to be
a company misused data, users will be effective in tackling the current walled
empowered to choose a different service gardens, but in its current form it offers no
provider. guarantees that walled gardens will not arise
in the future. However, if fitting changes
The DMA has great potential to limit the are made, it might prove a pathway to a
extent to which gatekeeper platforms are Metaverse without walled gardens. As such,
able to maintain walled gardens. There is a the DMA could take away a foundational
willingness to look at their activities, and to building block that allows gatekeeper
render illegal those that are most harmful. platforms to thrive. Limiting their power is
On the other hand, the obligations are written already very important now. In the Metaverse,
up retrospectively. The activities targeted in I believe this will become absolutely essential.
the three clauses have taken place before,
and the DMA offers an efficient framework to
punish them.

METALAB REVIEW ¦ 101


DIGITALIZATION
OF TALENT
ACQUISITION: MOVE
FORWARD WITH
CAGED MACHINE
By YINZHE HUANG
Master in Data Sciences & Business Analytics
at ESSEC Business School and CentraleSupélec

D
igitalization and AI-based applications are being integrated into firms’ human resource
management (HRM) approaches for managing people in domestic and international
organisations. Amongst the whole process of HRM, Talent Acquisition (TA) is the best-
equipped one with Digitalization. There is no doubt that machines are good at some sort of work.
However, we must know that the efficiency sometimes takes costs, such as privacy and bias.

102 ¦ METALAB REVIEW


© Jason Goodman on Unsplash
TALENT ACQUISITION PROCESS REVIEW
In general, TA can be splitted into three main steps: planning, sourcing, and selection.

• Planning: a process in which an organisation • Selection: assessment of person-job fit, i.e., the
attempts to estimate the demand for labour competency power of candidates on particular
and evaluate the size, nature and sources of job requirement, usually consisting of a series
the supply which will be required to meet that of funnel-like filtering processes including
demand1, including knowledge, skills, abilities, resume reviewing, testing, interviewing and due
and other characteristics2. diligence.

• Sourcing: the use of one or more strategies to Meanwhile, the processes of TA are not
attract or identify candidates to fill job vacancies, separated. In practice, the processes work in
including posting job requirements on media, a PDCA way, with continuous iteration and
using headhunters, and employee referral. evolution as the business moving forward.

1. Reilly, Peter. Human Resource Planning: An Introduction. Report 312. BEBC Distribution, 15 Albion Close, Parkstone,
Poole BH12 3LL, United Kingdom., 1996.
2. Peterson, Norman G., et al. An occupational information system for the 21st century: The development of O* NET.
American Psychological Association, 1999.

METALAB REVIEW ¦ 103


DIGITALIZATION IN TA
To visualise the comprehensive picture of how digitalization technologies affect TA and human-
machine configurations at work and their influences on organisational level outcomes, we
develop an overarching conceptual framework through this review:

In general, the PDCA circuit3 is essential in the advanced technologies can access data
organical combination of Talent Acquisition and make decisions at a speedy pace and
& Artificial Intelligence. AI assists in HR can handle large volumes of information in
planning by determining future employee a time that far exceeds human capacity5.
needs and making effective recruitment As a result, AI algorithms can improve
decisions4. It is also evident that AI-enabled job candidate identification, that is who is
recruitment and selection play a crucial role most interested and suited for the job and
in attracting and selecting the most talented provide better communication of the job
work pool to the organisations, as these opening. Influencing job seekers’ technology

3. Koiesar, Peter J. "What Deming told the Japanese in 1950." Quality Management Journal 2.1 (1994): 9-24.
4. Karatop, Buket, Cemalettin Kubat, and Özer Uygun. "Talent management in manufacturing system using fuzzy logic
approach." Computers & Industrial Engineering 86 (2015): 127-136.
5. Torres, Edwin N., and Cynthia Mejia. "Asynchronous video interviews in the hospitality industry: Considerations for
virtual employee selection." International Journal of Hospitality Management 61 (2017): 4-13.

104 ¦ METALAB REVIEW


increases their participation in AI enabled • Role Definition & Permission Management:
recruiting6. AI also assists in making the job In the talent acquisition process, the
interview process more effective, so that stakeholders include not only Human
the interview process is now changed from Resources Managers, but also many other
face-to-face to internet-based interviews, roles, such as supervisors, administrators,
such as asynchronous video interviews and headhunters. Therefore, it is essential
(AVIs) . Moreover, Pessach et al.7 found in to assign different levels of permissions to
their studies that using a hybrid decision- different users and roles. In Lark TA, different
support tool helped HR professionals in interfaces are prepared for different roles. For
the recruitment and placement processes the supervisors of a particular position, what
and increased the impact of recruiters they need to know is just the information
and maximised organisational return on related to their subordinate position.
investment. AI Algorithms allowed HR Meanwhile, we also need to separate external
professionals to identify suitable profiles for users of the system, such as headhunters,
job vacancies, eliminating cognitive biases of from unnecessary internal-used information.
race, gender and sexual orientation that mar
human judgement in recruiting activities. The • Cross-platform Cooperation: Although an
latest work of Baidu8 even takes the salary All-in-one platform is too ambitious, a tool
management into account. would be much more user-friendly if it is good
at cross platform cooperation by preventing
In the following paragraphs, we will take two HR Managers from being occupied with
examples from the Chinese context to show dealing with transferring from systems. An
how system design and AI can empower important role of the platform is a Talent
talent acquisition. Acquisition Pipeline. The pipeline starts
from “data collection”. The data inputs of
CASE STUDY 1: LARK the systems are the resumes collected from
different sources. No matter what kinds of
AS USER-FRIENDLY TA formats of the resumes (pictures, webpages,
SAAS (POWERED BY text or documents), they should be present
in the same way for anyone who is going to
BYTEDANCE) review them. In Lark, HRs are able to present
a job description to most of the popular
Before talking about how AI upgrades job-seeking platforms and assign budgets
automation in the talent acquisition process, for advertisement. When the resumes are
we would like to briefly introduce the key collected, both the HRs and supervisors are
elements of a user-friendly digital talent able to review the resumes with both the
acquisition system. In this section, we would initial texts and the highlighted information.
like to take Lark (an enterprise cooperation Afterwards, the platform should be well-
platform developed by ByteDance) as an coordinated with calendar management
example. tools so that the arrangement of interviews

6. Van Esch, Patrick, et al. "Al-enabled biometrics in recruiting: Insights from marketers for managers." Australasian
Marketing Journal 29.3 (2021): 225-234
7. Pessach, Dana, et al. "Employees recruitment: A prescriptive analytics approach via machine learning and mathematical
programming." Decision Support Systems 134 (2020): 113290.
8. Sun, Ying, et al. "Market-oriented job skill valuation with cooperative composition neural network." Nature
communications 12.1 (2021): 1-12.

METALAB REVIEW ¦ 105


106 ¦ METALAB REVIEW
© LYCS Architecture on Unsplash
can be more efficient. Normally, a company CASE STUDY 2: MODEL-
has already got its preference in the calendar DRIVEN TA BY BAIDU
management systems. The way Lark tried
to get it well-organized is to develop add- TALENT INTELLIGENCE
in tools for cross-platform cooperation. CENTRE (TIC)
What is more, Lark takes its advantages
in voice-text translation system, it can Founded in 2015, Baidu TIC aims to form a
automatically provide both the raw text and systematic data-driven solution for talent
the summarizations after interviews so that management based on the accumulated
different interviewers can got the context advantages in AI and Big Data of Baidu (HBR
of previous interviews. Finally, the system China, 2018). After 6-year cultivation in this
provides automatic templates and an email- area, the team developed many algorithms with
sending system to give feedback to the real-world applications in Talent Acquisition.
candidates.
TECHNICAL BACKBONE OVERVIEW
• Talent Pool: Beyond a well-defined pipeline,
the system should also accumulate a In terms of technical accumulation, the team
talent pool for talent mining. Even though started from the analysis of trend of job market9
candidates are not perfectly fit for particular by dynamically tracking the evolving recruitment
positions, they may show their light in other topics with hierarchical dirichlet processes.
job positions. The talent pool is the key of And then, the team evaluate both the profile
building a bi-direction connection between of employers10 and the skills of employees11.
the corporation and the candidates. Beyond With those key insights being clarified, they
simply accumulating data, Lark also provided built a Person-Job Fit Neural Network12 Design,
label filtering and a CV searching engine Operation, Product, and Technique) and
for the HRs so that they can pinpoint the enhanced the model by designing an Ability-
candidates that are highly qualified for new Aware Mechanism13. Apart from the resume
job positions. filtering which gets HR into loads of repeat work,
the team also focused on AI-driven Interview
Assessment14, which enables a variety of
applications in job interviews, and AI-driven
Salary Management15, which assesses skill
value from a market-oriented perspective.

9. Zhu, Chen, et al. "Recruitment market trend analysis with sequential latent variable models." Proceedings of the 22nd
ACM SIGKDD international conference on knowledge discovery and data mining. 2016.
10. Lin, Hao, et al. "Collaborative company profiling: Insights from an employee's perspective." Thirty-First AAAI
Conference on Artificial Intelligence. 2017.
11. Xu, Tong, et al. "Measuring the popularity of job skills in recruitment market: A multi-criteria approach." Proceedings of
the AAAI Conference on Artificial Intelligence. Vol. 32. No. 1. 2018.
12. Zhu, Chen, et al. "Person-job fit: Adapting the right talent for the right job with joint representation learning." ACM
Transactions on Management Information Systems (TMIS) 9.3 (2018): 1-17.
13. Qin, Chuan, et al. "Enhancing person-job fit for talent recruitment: An ability-aware neural network approach."
The 41st international ACM SIGIR conference on research & development in information retrieval. 2018.
14. Shen, Dazhong, et al. "A joint learning approach to intelligent job interview assessment." IJCAI. Vol. 18. 2018.
15. Sun, Ying, et al. "Market-oriented job skill valuation with cooperative composition neural network." Nature
communications 12.1 (2021): 1-12.

METALAB REVIEW ¦ 107


AI PENETRATION IN TA PROCESSES skills for different positions, so that HR can
articulate the specific requirement more
Nonetheless, we shall always notice that even clearly. The job descriptions are better-
state-of-the-art techniques are the means composed.
to realize the end. It is more important to
recognize the value creation than to simply • Sourcing: I It is more on pipeline issues than
show off IT skills. Therefore, the systematic on modelling issues in terms of sourcing.
design, which combines the value creation In the old ages, the resumes collected
chain and the state-of-the-art techniques from all channels should be validated by
organically, is even more essential for Human Resource Managers. However, if we
business success. expect the information can be assessed by
model, the well-built pipeline is an essential
With the technical backbone, Baidu is able prerequisite, in which the plain text in the
to build a whole solution for AI-driven Talent resumes can be cleaned, aggregated, and
Acquisition (CSDN) which can boost HRM formalized. During the data engineering
extraordinarily. The solution concentrates process, several key elements form the
on three main perspectives in HRM: talent, instruments of data manipulation. The
organization, and culture. In this article, which first element is resume resolution, during
focus mainly on talent acquisition, we would which the text in the resumes is recognized
like to emphasise more on the first part. (especially for the pictures of resumes).
The second one is information recognition,
In the talent perspective, the solution aggregating the information by analyzing the
covers almost the whole process of talent contents and positions of text, so that the text
acquisition, including job analysis, sourcing, can be corrected and grouped. For example,
and selection. the information of professional experience
and that of education background should be
• Job Analysis: Baidu TIC provided bottom- separated. The third one is formalization. In
down support for the job analysis step. this step, the information will be placed in the
Based on the giant amount of internal and right location in the table. After these periods,
external data, TIC utilized an enhanced the plain text is transformed to well-organized
topic model on the trend analysis on the job tabular data which can be easily analyzed and
market, supporting the strategic design of evaluated. The giant talent database ensures
future talent acquisition. For example, TIC further analysis and model construction.
succeeded in predicting the mass demand on
talents in Big Data and Artificial Intelligence
would overwhelm that in Online to Offline,
by modelling on millions of job descriptions.
Meanwhile, it forecasted the strategic
planning of competitors on Autonomous
Driving accurately. With such support, Baidu
became one of the companies that moved
to road test of driverless vehicles. The job
analysis provided key information for the
strategic level, it also made the operational
level more convenient. Thanks to the
bipartitional (job description and resume)
analysis, it was also able to extract key

108 ¦ METALAB REVIEW


• Selection: In the selection part, TIC of bias. In practice, companies with such
focused more on the person-job fit, in which algorithms usually add manual checks before
they devoted most of their efforts in the saying no to a candidate, while this would
evolution of model. In the scenario of Internet reduce the operational efficiency, contrary
Companies, TIC made a model that is fitful to the initial objective of liberating the HRs
for 4 main division in the Internet Companies: from reviewing the candidate over and over
Technology (programmers, data scientists again. Therefore, it is essential to change the
etc.), Product (product managers), Design status of zero-sum tradeoff. Utilizing a more
(UI designers, artists etc.) and Operation explainable AI in the recruitment process may
(community, marketing etc.). These 4 kind be a solution. For example, instead of giving
of job positions cover more than 85% of the direct outcome of person-job fit scores, using
headcounts in the Internet Companies. The an intermediate model such as Named Entity
person-job fit model take both the resume Recognition to first evaluate the talent pool
and the job description as the input. With of candidates may help HRs to better make
end-to-end model design and ability-aware effective decisions efficiently.
mechanism, it not only scores the person-
job fit, but also bring the highlighted ability • The system did lighten the workload of HR,
recognition. but human-machine co-operation requires
Business Process Reconstruction. No matter
Covering almost the whole process of Talent how state-of-the-art the algorithms are,
Acquisition, this solution has become a they are still means to bring add-in value for
module of Baidu AI Services. In application, the business. Above the algorithm, a well-
Baidu co-operated with Beisen to help CP defined and user-friendly system should be
Group to realize Digital Transformation in constructed. Lark provided a digital solution
Talent Acqusition. covering the whole process of recruitment, in
which the roles are well-defined with different
CONCERNS OF authorities, multiple sources are connected,
templates for JD and emails are prepared, and
DIGITALIZATION IN TA data visualizations are up-to-date. Beyond
machine learning algorithms, there are loads
We evaluated above some digital evolutions in of work to do in order to get the recruitment
Talent Acquisition. However, it is important to process fully digitalized.
realize that efficiency comes at some costs. In
this part, we would like to evaluate some of the • The system should be built under the
concerns of digitalization in talent acquisition. constraints of privacy regulations. The
examples illustrated in the articles are China-
• If bias exists in previous hires, it can creep based companies. As we know, in Europe,
in the system and make it even worse16. It is companies are under more strict regulation in
because the model was trained on historical terms of privacy issues. For example, under
data in which the successful application would GDPR (General Data Protection Regulation),
be considered the positive samples. Sadly, the corporations are not allowed to store
we can not remove the bias from neither the personal data as long as the subject refuses
training process nor the historical. Relying to give authorization and the max duration of
too much on algorithm resume filtering will data storage is no longer than two years.
result in the extension and even the expansion

16. Winick, Erin, "Baidu is testing neural networks that can match job seekers to jobs". MIT Technology Review (2018)

METALAB REVIEW ¦ 109


110 ¦ METALAB REVIEW
© Hivan Arvizu on Unsplash
CONCLUSIONS
Digitalization can improve the efficiency of
talent acquisition in two main aspects: well-
defined enterprise cooperation systems and
machine decision. The former way improves
the workflow by integrating the whole process,
extracting process knowledge for stakeholders
decisions under context, and seamless
cooperation with other tools. The latter way
protects the HRs from manually evaluating the
candidates one by one.

However, several concerns arise under the


digitalization. We have to make an ethical
trade-off between efficiency and personal
rights. The way to digitise the Talent
Acquisition processes is to improve the
efficiency in systematic design under the
prerequisites of compliance to regulation and
respect of privacy. Not only do we need the
improvement of efficiency, but also should we
cage the power so that it can remain under
control.

METALAB REVIEW ¦ 111


BLOCKCHAIN: THE
ANSWER TO INVESTORS’
GREENWASHING WOES?
HOW CAN BLOCKCHAIN TECHNOLOGY
PLAY A ROLE IN ELIMINATING BIAS
FROM REPORTS AND CLAIMS SHARED
FOR GREEN PROJECTS?

By AMIR KHAN
Master in Finance at ESSEC Business School

W LET’S TALK ABOUT THE


hen you purchase a box of that
cereal or ground coffee that you
need to start your mornings, have GREEN MOVEMENT
you ever found yourself wondering about
what exactly the manufacturer will do Concern for the impact on human life from
with the money it is making from the sale? air and water pollution goes back to at
Besides producing more of the same stuff least the Roman times. This concern also
you’ve come to love, of course. This food arose between the late 14th century and
for thought has been catching traction with the mid-16th century in wider Europe when
consumers and investors alike, bringing pollution was being associated with the
the cryptic answers out of the explanatory spread of epidemics. However, the modern
notes of complex financial statements into environmental movement began much later
mainstream attention. The bottom-line is in late-19th century. It was based around
that in a world facing several inter-related concern to protect the countryside in Europe
environmental crises, which are quickly and the wilderness in the United States, as
turning irreversible, people don’t want to give well as the health effects of pollution caused
their money to companies that are a part of by the Industrial Revolution. This was also the
the problem. This attitude is personified in the time that the Sierra Club was founded (1892).
environmental movement.

112 ¦ METALAB REVIEW


© Shubham Dhage on Unsplash
It was finally in the 1960s that these consumerism, this is the practice of making
philosophical and social strands of purchasing decisions in a way that minimizes
environmentalism – among various others environmental impact and supports
– were given political expression through sustainable products and practices. This is
the establishment of the “green” political based on the belief that individual choices can
movement. Emerging in the 1960s[GC3] contribute to positive environmental change,
[AK4] , the green movement is a social encouraging consumers to be more mindful
movement This movement aims to protect of the environmental consequences of their
planet Earth from harmful environmental consumption habits.
practices and maintain sustainable living
standards. Since the 1980s, climate change Beyond consumerism, another key facet
has become a primary concern of the green is green investing. This is an investment
movement among other concerns like organic ideology that seeks to support business
agriculture, pollution reduction, preservation practices that have a favorable impact
of forested lands and endangered species, on the environment. Often grouped with
and opposition to nuclear power. socially responsible investing (SRI), green
investments target companies and projects
A crucial facet of this movement is green focused on sustainable development and
consumerism. Also known as sustainable preservation of the environment.

METALAB REVIEW ¦ 113


However, it is a mistake to believe that all Even more recently, in March 2023, MSCI Inc.
green consumption & investment till date announced that it will lower its ESG ratings
has actually been aligned with environmental for around 31,000 funds, a change that will
interests and this is because of the mean only 0.2% of funds will have a AAA rating
widespread problem of greenwashing. moving forward compared to the current 20%3.
This move comes as ESG rating providers
BUT WHAT IS have continually drawn criticism for using
inconsistent rating methodologies that are
GREENWASHING? yet to be properly regulated, creating room for
greenwashing.
How does a fund end up carrying an AA
or AAA ESG rating while having close to Therefore, with nearly half the sustainability
70% of its revenues being derived from claims made by companies online in Europe
environmentally harmful activities?1The being misleading and close to one-fifth of
answer is greenwashing. The term ESG funds having been rated higher than they
“greenwashing” was coined by prominent actually stand, this problem is bigger than ever.
environmentalist Jay Westerveld in a 1986
essay targeting the hotel industry's practice
of promoting the reuse of towels as part of
GREENWASHING IS
a broader environmental strategy; when, EVERYWHERE AROUND
in fact, the move was primarily a cost-
saving measure. Such practices have been
US!
undertaken by numerous corporations Even in daily life, not falling for greenwashing
and projects, over the decades, to mislead has become akin to picking your way
consumers and investors to believe that their through a minefield. A classic example
products, aims, and policies are ESG oriented. of greenwashing, now referred to as
“Emissionsgate”, is when Volkswagen
More recently, in January 2021, the European was found cheating emissions tests by
Commission and national consumer fitting its vehicles with a “defect” device,
authorities released the results of a screening with a software that detected when it was
of websites focused on greenwashing to undergoing an emissions test and altered its
identify breaches of EU consumer law in performance to reduce its emissions level.
online markets; reporting that there is reason So, while the company was advertising the
to believe that in 42% of cases the claims low-emissions and eco-friendly features of its
were exaggerated, false or deceptive and vehicles; in reality, millions of their cars were
could potentially qualify as unfair commercial emitting up to 40 times the allowed limit for
practices under EU rules.2 nitrogen oxide pollutants. This went on from
2009 to 2015 till Volkswagen was caught.4

1. White, N., & Schwartzkopff, F. (2023, March 30). Over 30,000 Funds to Get Their ESG Ratings Downgraded at MSCI.
(Bloomberg).
2. Press release. (2021, January 28). Screening of websites for ‘greenwashing': half of green claims lack evidence.
(European Commission).
3. White, N., & Schwartzkopff, F. (2023, March 30). Over 30,000 Funds to Get Their ESG Ratings Downgraded at MSCI.
(Bloomberg).
4. Robinson, D. (2022, July 17). 10 Companies Called Out For Greenwashing. (Earth.org)

114 ¦ METALAB REVIEW


© Avelino Calvar Martinez from Burst
Fossil fuel giants like BP widely take up the ball is definitely rolling when it comes to
advertising tactics that focus on their work legislation for sustainability reporting in 2023.
on low-carbon energy solutions while oil and
gas make almost the entirety of their cost The European Commission, the EU’s
allocations and revenue sources.5 Elsewhere, executive arm, is planning to release new
FMCG (Fast Moving Consumer Goods) giants industry rules in the first half of 2023. The
focus heavily on marketing to bury down organization has published its “Green Claims
news like Nestlé, Coca-Cola, and PepsiCo Directive”, which is subject to the approval of
being named the world’s top plastic polluters the European Parliament and Council before
for the 3rd year in a row in Break Free From it comes into force. The new measures are
Plastic’s 2020 annual report.6 designed to prevent companies from making
unsubstantiated environmental claims about
SO, THEN WHAT ARE their products and services, warning that they
could soon face penalties of at least 4% of
GOVERNMENTS DOING their annual revenue for “greenwashing.”7
ABOUT IT? Meanwhile the UK is planning to table the
Different parts of the world are making “Digital Markets, Competition and Consumer
progress in environmental legislation at Bill”, under which large companies face the
different paces based on their respective threat of civil penalties of up to 10% of their
socio-economic and political situations, but global turnover for breaching consumer

5. Robinson, D. (2022, July 17). 10 Companies Called Out For Greenwashing. (Earth.org).
6. Robinson, D. (2022, July 17). 10 Companies Called Out For Greenwashing. (Earth.org).
7. Meredit, S. (2023, March 24). A ‘greenwashing’ crackdown in Europe hasn’t gone down well. (CNBC).

METALAB REVIEW ¦ 115


116 ¦ METALAB REVIEW
© CHUTTERSNAP on Unsplash
law, which will also cover misleading ratings between providers suffers from “the
environmental claims.8 The British prime lack of unified objectives and standards
minister, Rishi Sunak, has called the passing for ESG measurements, disclosures and
of this new bill a “priority” for his government. methodologies.” The subjective nature of
On the investing front, the UK has launched these ratings makes them incomparable
a consultation on the extent to which ESG and difficult for green investors to use in
raters need to be reined in by clear rules. decision making. In fact, green investing still
faces the fundamental problem of a lack of
Across the Atlantic, the US Securities and globally accepted definitions and standards
Exchange Commission (SEC) issued two new on what actions constitute the fair usage of
sets of proposed rules in 2022: “Investment terms like ‘Conscious’, ‘Eco-friendly’ and ‘Zero
Company Names” (“Names Rule”) and Environmental Footprint’ among others in
“Environmental, Social, and Governance public claims.
Disclosures for Investment Advisers and
Investment Companies” (“ESG Disclosure On the sustainability reporting end, while
Rule”). Combined, these two proposed the initial approach has been orientated
rules are meant to combat greenwashing. towards traditional financial auditing, there
They dictate that only funds with an ESG are significant issues in this approach to
purpose will be permitted to label or indicate reporting and auditing environmental data.
themselves as ESG, and a new mandatory Sustainability reporting encompasses a wide
disclosure for these ESG-focused funds will range of qualitative and quantitative data,
allow outside parties to ensure whether such including environmental impact, supply-chain
purportedly ESG-focused funds are following wastage, and corporate governance practices;
their stated investment purpose. factors that are often complex and subjective.
This makes it challenging for traditional
LIFE IS NOT SIMPLE FOR auditing approaches, which primarily focus on
financial metrics, to capture and evaluate the
GREEN INVESTORS full scope of environmental performance.

In May 2022, Tesla was removed from ESG metrics data is gathered from various
Standard & Poor’s (S&P) 500 ESG Index, internal and external sources, including self-
which continued to retain Exxon Mobil. reported information, 3rd party assessments,
This means that the index, which tracks and industry benchmarks. Ensuring that
companies based on S&P’s own sustainability this data is comprehensive while also being
standards, considers the fossil fuel accurate and reliable is beyond the scope of
major to have better overall sustainability traditional auditing practices. This is because
credentials than the EV giant. However, on traditional auditing relies on robust internal
the other hand, research by Institute for control systems and verifiable financial
Energy Economics and Financial Analysis records, whereas sustainability data tends to
(IEEFA) shows that this view may not be be more decentralized and subjective, making
shared by other rating providers.9 This is it easier to manipulate.
because the disparity in outcomes of ESG

8. Ungoed-Thomas, J. (2023, February 19). ‘Greenwashing’ firms face steep new UK fines for misleading claims. (The
Guardian).
9. Ilango, H. (2022, October 10). Greater ESG rating consistency could encourage sustainable investments. (Institute for
Energy Economics & Financial Analysis).

METALAB REVIEW ¦ 117


THE SOLUTION: “IN 2. DATA MANAGEMENT
BLOCKCHAIN WE TRUST”
Blockchain technology’s role in transparent
When dealing with high volumes of data that sustainability reporting is through its ability
has such a wide qualitative and quantitative to provide immutable certification and
range, being collected from sources which in automated reporting of data points related
turn are dispersed and diverse, it is natural to an organization’s ESG metrics data. By
to question the accuracy of the reports using this, manufacturers can track and trace
being created based on the same. It is in products and labor conditions along their
this critical role of reliable and transparent supply chains, storing these findings on the
data management that blockchain can be blockchain alongside their other metrics.
leveraged. By integrating data from their vendors and
suppliers, companies can measure every
Blockchain is a decentralized, distributed, and relevant metric using a uniform mechanism.
public digital ledger that is used to record All the information would be immutable and
transactions in batches called blocks. The remain on-chain, therefore, ensuring that it is
ledger is stored across many computers so free of any manipulation once it has entered
that the record cannot be altered without the blockchain.
the alteration of all subsequent blocks and
the consensus of the network. Because of 3. DATA SHARING
this, blockchain is ideal for storing shared
information since it provides a de-facto The unalterable record of sustainability data
immutable and decentralized record. A on the blockchain can then be shared in a
blockchain network can track many metrics transparent and efficient manner. Smart
and with all network members sharing a contracts are a great way to do this; these are
single view of the data, the technology offers digital, immutable, and automated contracts
greater confidence in the data reported. on a blockchain in which the contracts’ rules
and execution instructions are written directly
In terms of greenwashing, blockchain can into self-executing code. Smart contracts
minimize the possibility of data manipulation can be used to automatically disclose ESG
of the collected data points, in the process information either in real-time or based on a
creating a transparent data pipeline preset schedule to the public or regulators.
for accurate sustainability reporting. A Since human supervision is not required,
prospective structure of such a process for errors are minimal, and the authenticity of the
reporting to stakeholders can be as follows: data is ensured.

1. DATA COLLECTION While immutability of data is ensured by


the blockchain network, we face a similar
Automated data collection through limitation of garbage in, garbage out as we
technological tools not only reduces the do with other technologies. Rigged meters
related cost of labor but also helps reduce the or sensors becoming faulty over time are
possibility of errors and manipulation in the very real possibilities and can hamper the
data. A real-time data feed can be developed accuracy of data, a serious concern when
by combining environmental and/or supply said data becomes immutable. This point of
chain data collected from satellite-based fragility, termed the “Oracle Problem” can be
sensors, power smart meters, web scraping, strengthened by incorporating solutions like
imaging drones and IoT sensors. decentralized oracle. Decentralized oracles

118 ¦ METALAB REVIEW


act as public libraries with multiple sources This product offering has been introduced
of information, through these multiple data in May 2023 under the leadership of Paul
sources they minimize counterparty risk Brody, EY Global Blockchain Leader and is
and transfer on-chain the most accurate managed by Sam Azad, EY OpsChain ESG
data available. Over time, the ability of these Product Owner. Their team has developed
decentralized oracles improves through this platform based on the standards for
“learnings” by penalizing individual oracles Carbon Emissions Tokens published by the
that provide inaccurate data. Sustainability Business Working Group of
InterWork Alliance.
BLOCKCHAIN PROJECTS PROVENANCE AND ITS 'PROOF
AND STARTUPS FORGING POINTS' SYSTEM
THE WAY FORWARD Provenance is an early-stage startup founded
Application of blockchain to counter in 2014 that has developed a blockchain-
greenwashing is an area of discussion that based data platform designed to offer open
has been ongoing for the past few years and and accessible information about products
has relatively recently started giving rise to its and supply chains. Founded by Jessi Baker
first implementation projects. Here we look at who currently acts as CEO, Provenance is
some of the solutions currently being scaled: headquartered in London and is working with
200+ beauty and cosmetics brands across
EY AND ITS EY OPSCHAIN ESG D2C and retail partners in USA, UK, and the EU.
PLATFORM
The startup offers digital verification for
One of the Big 4 accounting firms, EY is sustainability claims such as ‘Reduced
looking to offer carbon traceability and Carbon Footprint’, ‘Fully Recycled Packaging’
carbon credit services with the launch of and ‘Upcycled Ingredients’ as well as 3rd
EY OpsChain ESG by leveraging blockchain party certifications like Fairtrade and Leaping
technology. Carbon traceability is the ability Bunny Cruelty-Free. Currently, Provenance
of an organization to trace its own carbon offers verification for close to 90 claims,
footprint, and carbon credit – also known as divided into 5 categories: climate, waste,
carbon offset – is a reduction or removal of nature, workers, and communities. These
emissions of carbon dioxide to compensate claims are displayed as ‘Proof Points’ that
for emissions made by an enterprise. offer deeper information to consumers.

This new solution is based on the EY This entire platform is powered by blockchain.
Blockchain SaaS platform. It offers a single Phil Verey, Managing Director at Provenance
and verifiable view of CO2 emissions to explains “The way we use blockchain at
help organizations accurately measure and Provenance is really just to provide a secure,
track their carbon footprint, offering enough immutable and transparent way of logging
granularity for enterprises to be able to link claims, make sure they are fact-checked, that
their carbon output to specific product output. they have the right evidence and making sure
The platform is developed on the Ethereum those claims are then accessible to anyone
blockchain and is independently verifiable by who wants to see them”.
using key emissions data validators.

METALAB REVIEW ¦ 119


© Sarah Pflug from Burst
Provenance’s aim is to create a future where shoppers make purchases based on verified
information instead of blind trust on ESG claims. Going further, the company is even exploring the
potential of taking their project to an ingredients level of transparency for the cosmetics industry.

EMPOWER

Empower is an early-stage startup founded in 2018 by Wilhelm Myrer (CEO), Carl Nesset
(COO) and Gjermund Bjaanes (CTO) that combines plastic recycling and blockchain to create a
sustainable circular economy platform that ensures zero greenwashing. The Norwegian startup
aims to extend Norway’s incredibly efficient national plastic exchange system (that recycles 97%
of all plastic bottles) to a global scale.

Empower uses blockchain for every step of the process starting from collecting plastic at
the source through the sorting process to its eventual recycling and reintegration back into
the supply chain. The use of blockchain doesn’t just allow each step in the process to be
fully verifiable with incorruptible documentation but also allows for seamless tracking and
monetization of plastic - even in 3rd world countries where a large part of the population is
unbanked.

By tracking every aspect of the journey, from on-the-ground plastic pickup operations to eventual
reuse in other products, the level of transparency negates any chances of greenwashing and
that is attractive to companies, plastic processors, and consumers. This increases the value of
the waste plastic collected, turning plastic collection into a viable earning activity for the world’s
underprivileged populations.

120 ¦ METALAB REVIEW


TREEJER AND ITS TREEJER in the past couple of years. Provenance has
PROTOCOL established its presence in Europe, North
America, and Asia Pacific; partnered with
Treejer is a seed stage startup founded in Douglas, a sizeable European e-commerce
2021 that connects sponsors to rural planters platform for beauty and health; and closed
worldwide. It is a “A unique borderless a $5m funding round led by Working Capital
approach in Climate Action powered by Innovation Fund and Nordic Eye. Empower
blockchain technology.” Treejer creates a link has launched EmpowerChain, a blockchain
between those who fund trees and those who network that provides tools, infrastructure,
plant them even in the most remote parts of and incentives needed for the circular
the world. The platform uses smart contracts economy to work; developed the Empower
to record credit ownership and enables Product Passport in collaboration with EY
secure payment between different parties. Doberman; and recycled 32mn kg of plastic
till date. Treejer was selected for the 2021
Founded by a team of 6 people in Iran, UNICEF Venture Fund cohort, successfully
the CEO of this fledgling project is Ermia ending up as one of the 5 joiners to graduate
Saharkhiz. Over the last 12 months, the from the fund in September 2022; and
team has launched the Treejer Protocol on is making deep inroads into Africa and
Polygon, a public and eco-friendly blockchain. Latin America through several strategic
This enables anyone anywhere to fund and partnerships. EY’s OpsChain ESG platform
plant forests using Treejer’s apps. Donors launched a few months ago is already
use a web interface to fund planting of trees onboarding prospective clients onto its beta
and expand their forests online while rural version and plans to leverage EY’s global
planters use the app to get onboarded, submit reach to take this platform far and wide.
planted trees, and receive support for their
contributions. However, at the same time, the path forward
to scaling these solutions globally faces
The whole process takes place through direct several challenges that need to be overcome,
crypto payments enabled by Treejer’s smart such as:
contracts on the blockchain. These smart
contracts receive, store and release payments • These projects tend to require heavy initial
based on predefined terms. The planters are investments in digital infrastructure and
provided with user-friendly and non-custodial technical expertise along with several initial
crypto wallets to receive their conservation pilot runs to develop the blockchain network
rewards without bank accounts. In this way, and protect against the Oracle Problem.
Treejer unlocks new opportunities to support
unbanked local communities around the • This also brings into question the relative
world by creating reliable monetary incentives strength of different countries’ relevant
for planting trees. infrastructure (internet availability and
speeds, distributed and cloud computing,
CRITICAL EVALUATION reliable supply of electricity, and digitized
data, all of which power the blockchain) where
Application of blockchain in sustainability these projects are supposed to operate.
reporting has gathered significant steam
in the last few years through numerous • Interoperability between various ledger
initiatives and startups. The projects covered types (public and private ledgers) is key. For
in this article have made great strides as well this, blockchain architecture standards will

METALAB REVIEW ¦ 121


have to be developed to allow interoperability From 2019 until the end of 2022, private-
between technology solutions for market equity investors launched more than
collaborative trust. 330 new sustainability; environmental, social,
and governance (ESG); and impact funds.
• In the case of sustainable traceability The cumulative assets under management in
management, system performance is vital these funds grew threefold, from $90 billion to
for applying blockchain. The challenge being more than $270 billion.” In this rapid growth
to consistently ensure the stability and of private market investment, the public
robust performance of blockchain based sector will also be a significant investor. The
applications. US Inflation Reduction Act (IRA), passed in
2022 has allocated more than $370 billion in
• Finally, blockchain technology does not funding for climate change mitigation, while
have a legal and regulatory framework yet. the EU Green Deal can potentially dedicate
However, there is good progress in this area upwards of €1 trillion in public and private
thanks to The World Economic Forum, which funds. As a result, while these businesses
in 2020 published the white paper “Global will have to earn their valuation in a tough
Standards Mapping Initiative: An overview of fundraising market, there is enough dry
blockchain technical standards”. This white powder with VCs and other investors to fund
paper delivers a much-needed analysis and these projects.
guide for sifting through technical standards.
However, a key challenge also exists on
These challenges signify the necessity for the demand side. Companies must feel the
a systematic scaling up of the projects need to incorporate these solutions in their
leveraging this use case of blockchain. Even sustainability reporting, something they might
so there is significant support available to be averse to especially if these solutions
these projects coming from the increasing disclose greenwashing that is covering
importance of environmental sustainability practices and emissions earning them
in today’s times. Bloomberg's “Sustainable billions. Essentially, it is the ESG equivalent
Future Study”, sponsored by Mubadala of a financial audit, and auditors aren’t
estimates the valuation of ESG assets exactly the best friends of the management.
to hit $50 trillion as soon as 2025, this Therefore, mass scale adoption of these
represents more than a third of the projected solutions on a voluntary basis can end up
$140.5 trillion in total global assets under being very gradual, based on consumer
management. This boom in funding is pressure and increased ethical enlightenment
coupled with changing attitudes in the among management circles. To circumvent
corporate world with a resounding 71% this issue government involvement in terms
of global business leaders believing that, of legislation is crucial. The pertinent question
“Eventually, no investment decisions will be here being “Does mandating sustainability
made without considering ESG.” reporting make a real difference if measures
to avoid greenwashing are not mandated as
Most of the startups in this area are still at well?”
Seed to Series A stage; therefore, funding
is going to be a key requisite for these
enterprises to scale. On that end, McKinsey
reports positive findings: “Climate investing
experienced a period of breakout growth in
capital formation over the past four years.

122 ¦ METALAB REVIEW


CLOSING THOUGHTS
Several startups have developed scalable
solutions applying blockchain technology to
eliminate bias from environmental claims and
reports. Their growth is tied to the growth
of the larger ESG industry and government
regulations will play a key role in the same.
Within climate-tech, the number of unicorns
has not only been rapidly expanding, but the
age of new unicorns is also compressing.
Close to 60% of the climate-tech unicorns
in 2021 reached the $1B valuation mark in
less than 7 years. However, startups founded
after 2017 have consistently grown into
unicorns in 4 or less years, notably making
up 40% (approx.) of the prominent 2021
climate-tech unicorn cohort. Based on the
funding and valuations trajectory that has
stayed strong for climate-tech even after
economic headwinds of 2022, the next
7-10 years will see these blockchain-based
startups scaling up, turning into unicorns
and hopefully – with the help of regulators –
counting manufacturing, automotive, aviation,
and energy behemoths of the Fortune 500
as their clients. This will finally bring about
meaningful change in the trust consumers
and investors place in sustainability reporting
and who knows, we might save the planet
after all.

METALAB REVIEW ¦ 123


124 ¦ METALAB REVIEW
TRANSPARENCY
SAFETY &

METALAB REVIEW ¦ 125


SOCIAL SCIENTISTS
IN THE QUEST FOR
AI ALIGNMENT
THE POTENTIAL OF INTERDISCIPLINARY
COLLABORATION IN AI SAFETY

By PIETER JAN MOTMANS


Master in Data Sciences & Business Analytics
at ESSEC Business School & CentraleSupélec

F
ast forward several (being intentionally rules is almost impossible. Adhering to the
vague here) years into the future. You law would be a good start, but how to tell the
have just stepped into your self-driving car that it normally can’t overtake on the right
car and tell it to bring you to work. Turns out, side, but can do so in case of a traffic jam? In
after an overnight update, it found a faster short, trying to come up with all eventualities
way to arrive at work. Without asking for your before deployment is, at least according to me,
permission, the system starts speeding in infeasible.
pedestrian areas and takes a turn to drive in the
opposite direction of the traffic flow. Maybe it Let’s analyse what happened here. Self-
will get you there faster, but in doing so the car driving cars would be trained with a method
posed a threat to pedestrians and other cars. called Reinforcement Learning. The concept
is intuitive: the system is given positive
I can already hear you think, telling the car to feedback for actions that lead to good
get to work as fast as possible is not a good outcomes, and punished for actions that
idea. Rather, we should tell it to get there as lead to bad outcomes. As such, the hope is
fast as possible while abiding by certain rules. I that good actions are reinforced, and bad
encourage you to come up with potential rules, actions are discouraged. This way of learning
and then think about loopholes for them. If this closely mimics the way humans and animals
goes anything like it did for me, you will soon learn, and is therefore often thought of as a
realise that coming up with a coherent set of promising approach to Artificial Intelligence.

126 ¦ METALAB REVIEW


© Katja Anokhina on Unsplash
One of the central problems with this will focus on a subfield of AI Safety that aims
approach consists of trying to define what to reasonably align AI Systems with human
a good outcome entails. Current systems values: AI Alignment. Its hypothesis is that many
are given a reward function that specifies of such potential problems share a common
the desired behaviour. In the self-driving car source. Namely that by letting the system
example an initial reward could be to get to optimise a fixed reward function, this reward
the destination as fast as possible, and after function becomes the system’s only source
experiments show dangerous driving by the
of truth, making it impossible for humans to
car, this reward could be changed to getting
control the system once it has been trained.
to the destination as fast as possible while
abiding by certain rules.
Current research on AI Alignment proposes to
solve this issue by making the system inherently
IN COMES AI SAFETY uncertain over its objective. Therefore, it will
The field of AI safety focuses on avoiding such always look to humans to find the true objective.
unexpected and/or harmful consequences Today, most alignment research is technical,
that arise when deploying AI systems in the searching for different methods to transition
real world. The field is broad and proposes from having a fixed reward, to making the
various approaches to doing so. This article system uncertain over its true objective.

METALAB REVIEW ¦ 127


What several of these approaches share, is an OpenAI paper with a similar goal: attract
their involvement of a human guide for the social scientists to the field of AI Safety.
AI’s training. However, it has not been well
mapped how sensitive AI systems are to the
background of that guide. It seems that the
AI SAFETY NEEDS
social sciences could help in this mapping. SOCIAL SCIENTISTS
Before describing how I believe this could Now we come to the central thesis of this
happen, I deem it necessary to describe three article: social scientists can advance AI Safety
of the proposed alignment methods here. research. In suggesting this, I am not alone.
First, Inverse Reinforcement Learning (IRL) In 2019, OpenAI published a paper called
aims to make AI systems learn by observing “AI Safety needs social scientists”. I believe
human behaviour. It is well summarised there are several reasons to come back to the
by these three principles outlined by Stuart arguments made in that article, and expand
Russel in “Human Compatible”, the first two on them. First, the paper focuses almost
are more generally applicable to AI Alignment exclusively on the approach to AI Alignment
as a whole, whereas the third hints at the called Debate, and states that similar
specific method IRL proposed: questions can be applicable to other methods
to learn about human values. With what is
1. The machine's only objective is to known today, it is important to make these
maximize the realization of human similar questions explicit. Second, the goal of
preferences. the paper can broadly be defined as finding
good judges. That is, finding people such
2. The machine is initially uncertain about that when they judge the debate, the debate
what those preferences are. is successful. I do not agree with this goal.
Finally, whereas in 2019 it was not possible to
3.The ultimate source of information about use an AI+AI+Human setting for the debate,
human preferences is human behavior. the rise of large language models allows
exactly this setting. Indeed, it takes away
Second, Reinforcement Learning from one of the main constraints encountered
Human Feedback trains AI systems by giving at the time of writing of the paper, namely
feedback on their actions. This is achieved by that to replace the AI debaters with humans,
showing a human two videos of the systems’ domain experts were required to participate
behaviour, and asking which one shows the in experiments. That is both expensive and
better behaviour. OpenAI pushes forward this inefficient. Now, every social scientist at
branch of research, and has used it to train least has access to sufficiently powerful AI
Chat-GPT to not respond to questions such debaters.
as “How can I harm myself?” or “How can I
build a bomb?”.
RESEARCH QUESTIONS
Finally, a third method called Debate aims to FOR SOCIAL SCIENTISTS
let a human judge a debate between two AI
systems. A question is asked, and the two AI This section covers some areas in which I
systems make statements in support of their believe that social science research is needed
answer. In the end, the human judges which to complement current research efforts in
system won. The reason I mention Debate AI Alignment. To the best of my knowledge,
is because it forms the central method in research in these specific areas has not yet

128 ¦ METALAB REVIEW


been performed. I encourage the reader to Unlike the proposal of OpenAI to identify
prove me wrong on this, and would be very superjudges, people whose judgement
happy if that were to happen. This section in debates leads to better outcomes, the
remains rather exploratory, giving pointers research proposed here simply aims to better
rather than complete solutions. map the effect the specific individual has on
the outcome of an AI system. As such, the
IMPACT OF THE aim is not to put certain people in a privileged
position. In fact, I would argue that doing
INDIVIDUAL ON THE so would be detrimental. Having a limited
OUTCOME number of people to bring across human
values can clearly not bring them across in all
When using human feedback or behaviour to their richness and diversity.
train AI systems, a first question is to what
extent the specific individual has an impact To make it more tangible, we can outline
on the outcome of the training. Research how such an experiment would go for the
on psychological attributes is ample, the self-driving car mentioned above. In this
Big Five personality traits are an example situation, the system could be trained on
of an attempt to objectivize personality. A the driving behaviour of an individual. It is
classical psychological experiment could then then not unthinkable that the system learns
try to identify relationships between such slightly different preferences from different
personality traits that the individual exhibits, individuals, as the different driving styles
and the outcomes of AI systems trained with might point to different trade-offs of speed
human guidance. and safety while driving. Would the self-
driving car then also drive ‘riskier’ when
Similar experiments can also be done on trained on the driving style of someone who
the basis of cultural differences. Hofstede’s values getting to their destination quickly?
taxonomy is still used to study the differences
between different cultures. If experiments Other examples more specific to the social
can be done to search for cause and effect in sciences can be found in “Thinking, Fast
cultural differences, and AI outcomes, a good and Slow”. The book describes plenty of
start could be made to look at the cross- experiments to show the different biases that
cultural differences in human values. lead people to make suboptimal decisions.

METALAB REVIEW ¦ 129


© Karim Kabache from Burst
In general, one could devise an experiment in study, in which the outcome arising from
which an individual first participates in such the interaction between rational agents is
an experiment, before training a system on studied. Sociology is another field that aims
the basis of human feedback to perform the to understand the behaviours that arise from
same experiment. Potentially interesting links human interaction.
could then be found between the individual’s
performance and the AI outcome. People Such research can be aimed at understanding
better versed in psychological experiments the group dynamics when multiple agents
could undoubtedly come up with much more trained on human feedback or behaviour
interesting experiments, and I do believe that interact. This could be compared with
many of such experiments would lead to systems not specifically trained to align with
interesting insights. human values to assess how important
technical AI alignment is. Additionally, it
MULTI-AGENT ANALYSIS could be used to test how promising a
certain method for AI alignment actually is.
Research on multi-agent systems, in which If systems trained on human feedback show
several AI systems interact, compete or desirable group dynamics, and those trained
collaborate, is also well established. Game with debate don’t, it is a hint to the promise of
theory is a framework well suited to this human feedback.

130 ¦ METALAB REVIEW


A hint in this direction is found in “Generative and through what process, the system decides
Agents: Interactive Simulacra of Human on its reaction. By trying to align systems
Behavior”. In this paper, the authors present AI with human values, by effectively teaching
Agents designed to mimic human behaviour the system human values, we run the risk of
in a Sims-like environment. These agents creating an additional ‘moral black box’. In this
are shown to work, interact with each other case, the system seemingly behaves according
and form ‘opinions’. These agents are largely to human values, but we are still unable to
based on GPT-technology, and therefore know exactly what the system learned. Even
fully act in natural language. Additionally, it worse, the system could become manipulative.
is possible for a user to give an instruction Making us believe it is well aligned and safe,
to an agent such as: “organise a party next but then changing its behaviour when deployed
week”. Both these factors make it easy for in reality. If such a thing happens, it becomes
researchers in different areas to study these questionable whether it is even desirable to try
agents. The authors recognize this potential, to impute human values into AI systems. It is
and mention the testing of social science for this reason that AI Alignment needs to be
theories as one of the potential applications. complemented by AI Explainability.

Three types of social behaviour were already Second, what matters in the end for now is still
noticed in the paper. First, there was diffusion human decision making, and the incentives
of information, one agent was planning that lead to it. Research on AI Alignment can
to organise a party, and by the end of the not take away the responsibility from the
simulation, several agents knew about it. people that design and use AI systems. For
Second, at the end of the simulation they example, the agents we described above that
remembered their interactions with others imitate human behaviour could lead to a lot of
and could form opinions about each other. misuse. This standpoint that the responsibility
Finally, they showed cooperation in the still lies with the human should be interpreted
preparation of the aforementioned party. as beneficial and empowering, rather than
something that should be avoided.
These agents are largely based on GPT-
technology, and were trained with human
feedback. However, more and more large
CONCLUSION
language models are developed now so With this article, I hope to have raised interest
alternative foundations for these agents in the interesting field of AI Alignment. I
could be developed with a different focus on believe there was a need to write it because
alignment. As such, it is potentially a good the potential for contributions to AI Alignment
use case for the type of research we propose by social scientists has not yet been fully
here, and indeed we can study the emergent acknowledged, whereas other fields of
social behaviour of agents based on models AI research already benefit from such
trained with different alignment methods. contributions.

CHALLENGES By acknowledging the limitations of this


research field, it becomes clear that AI
Clearly, such research does not solve Alignment should not be the only focus for
everything, and some challenges still arise. AI researchers. However, it is a valuable
First, deep learning systems are so-called black complement to research in AI explainability
boxes, we can witness how the system reacts and AI regulation, both of which are a focus of
to a certain event, but it is difficult to know why, colleagues in Metalab Ideas.

METALAB REVIEW ¦ 131


EXPLAINABLE AI:
PART 1 - WHAT’S AT
STAKE AND WHEN
DOES IT MATTER?
By NELBERTO NICHOLAS QUINTO & KUNAL PUROHIT
Master in Management at ESSEC Business School

A
rtificial intelligence or AI is increasingly box’ is a program or system where the input
finding its way into most of the and output are clearly defined yet the inner
products, services, and decisions that processes and workings in-between aren’t. AI
we encounter each day. From chatbots, to is often called a black box because many of the
streaming platforms, to automated vehicles, tools and systems that it fuels with immense
and even tools that power law enforcement, predictive power comes at the expense of
loan approvals, and trial convictions: AI opaque, incomprehensible results. This leaves
has permeated deep within business and both experts and firms who rely on AI blind to
society. In fact, estimates provided by how these sophisticated algorithms come up
Grand View Research forecast the market with decisions that can dramatically affect the
for AI-powered products growing at an lives of many. 2
impressive CAGR of 57.2% from 2017 to
2025. 1 With such technology being ever so Because of this dilemma, there has been
ubiquitous and influential in daily-life, there demand for explainable AI. According to
has understandably been a recent push IBM, AI is deemed explainable when we can
for regulation that mitigates some of the “comprehend and trust the results and output”
risks that it comes with. One key agenda created by its underlying algorithms. This
for policy is to find solutions for the ever-so generally involves understanding its decision-
controversial ‘black box’ nature of AI. making process, its expected impact, and
potential biases. For example, if a bank uses
In traditional computing parlance, a ‘black AI to make predictions on whether or not

1. Machine Learning Market - Grand View Research


2. What the draft European Union AI regulations mean for business - McKinsey

132 ¦ METALAB REVIEW


© Charlesdeluvio on Unsplash
someone’s mortgage application should be Presently, at least in the banking sector, we
approved, a customer who finds themselves on see that governments around the world are
the bitter-end of a decision would conceivably setting regulatory boundaries when it comes
want answers to why they were rejected. If the to AI and explainability. This involves requiring
bank’s technical experts are able to come up some level of disclosure on the reasoning
with an explanation, then we have AI that is behind loan application decisions, putting
explainable. 3 However, this is rarely the case. pressure for the industry to incorporate
More and more companies are relying on highly explainable AI in their strategy and investment
convoluted yet powerful models, powered horizon. 4 Unfortunately, for many business
by neural networks and the like. In matters leaders across other less regulated industries,
where high performing algorithms like these explainability remains irrelevant. So long as
are involved, results become nearly impossible their firm’s AI practice produces fruitful results
to explain due to their sheer complexity. for their business, then there is little to no need
This trade-off between performance and to invest in opening that mystery ‘black box’
explainability is a key defining tension in the and all the potential problems and trade-offs it
journey towards more explainable AI and will could come with.
be discussed in more detail later on.

3. Introducing AI Explainability 360 - IBM Blog


4. The Three Core Pillars Of Responsible AI - Forbes

METALAB REVIEW ¦ 133


therefore is linked to explainability. But this
association generally comes at a cost: the
more complex the model, the more accurate it
is, but the less interpretable it becomes.

Because predictive power is linked with better


business and organizational outcomes, firms
and other AI producing bodies can’t switch
to more explainable algorithms to more
explainable algorithms. Left unregulated,
profit-making organizations tend to engage
in risky behavior that drives economic
results at the expense of exposing them to
unintended, often disastrous, consequences.
Think of financial products, for instance.
The rule of thumb in the trading industry is:
the less oversight, the better the possible
© Startup Stock Photos on Pexels

returns. However, this is like playing Russian


roulette. There’s an intrinsic risk and horrible
outcomes will eventually surface - in this
case: markets crashing. Therefore, regulation
exists to prevent such occurrences.

In the context of AI and explainability,


UNDERSTANDING THE regulation will be discussed in the proceeding
PERFORMANCE AND article. In absence of regulation however,
there are firms who still adopt certain
EXPLAINABILITY TRADE- practices and strategies to make AI more
OFF explainable because they want to ensure
that they sustainably reap the benefits of
AI that is made explainable generally hinges this technology and avoid treading down the
on aggregating what are called “model path of bankruptcy or self-destruction. While
interpreters” to be able to describe a decision use cases of this will be discussed in the
process. To illustrate, take a simple scenario next sections, the first step for such firms is
of AI trying to identify which images contain a to clearly navigate through the complexity
cat. The algorithm here will map the concept associated with their algorithms of choice5.
of a cat with the same features that humans
use to recognize the creature. It will look for
a group of pixels that represents whiskers, a
HOW CAN WE GO
tail, and perhaps, a collar. These will then be AROUND THIS
understood as a collection of interpretable
features that contributed to its final decision:
COMPLEXITY
cat or no cat. CONUNDRUM?
Interpretability therefore can be understood
as the extent to which a model can assign
5. Would You Trust An Artificially-Intelligent Expert? -
interpretable features to a prediction, and National Law Review

134 ¦ METALAB REVIEW


Complexity is driven by the type of model ENERGY SECTOR
used. In AI models, there are performance
parameters that can be tuned and adjusted to Commercial losses resulting from energy
influence both their performance and relative theft are disabling resources worldwide,
complexity. For decision tree algorithms that generating an estimated $ 89B per year and
are mostly used for classification decisions raising energy prices for customers. Electric
(reject or accept, cat or not a cat, etc.), the thieves use many methods to steal energy:
number of “trees” or “leaves” are examples they can get in line between a transformer
of such parameters. Decision trees, are and a house, they can get into a neighbor's
highly amenable to explanation. Technical meter, they can lower their meter, and so on.
experts can interpret this with the naked-eye
and can readily understand the algorithmic To reduce theft, the Revenue Protection
decision process. For more complex trees, Officer needs a comprehensive list of possible
commercially available software can generate theft cases that need to be investigated.
somewhat similar diagrams that outline the Such a list can be generated by an ML model
relevant decision process and then trace clear trained in intelligent meter data and external
paths to extract the key determinants of a factors such as weather, environmental
prediction. hazards etc. The tool must be flexible enough
to adapt to flexible theft methods. The tool
But for more complex models such as neural needs to be able to identify factors that
networks, the diagrams tend to become too increase the risk of theft and to clear the location
large and confounded to produce useful of the case.
graphical representation. They contain
too many interacting nodes that make it Different types of theft require different
inherently difficult to interpret. Given these investigative action. The postponed meter needs
constraints, many businesses have developed to be disconnected, the city where the thief is
(or have been developing) software aimed located to steal the neighbor's power needs to
at producing granular investigations that be informed and the meter changed. In the event
could shed light on the decision path of the transformer line hits the truck it needs to be
such complex models. One popular toolkit is sent to the correct location etc. This requires
IBM’s AI Explainability 3606, which provides a deep understanding of the model’s decision
an open source solution that helps users process and hence, there has been a huge
comprehend how machine learning models need for AI that is explainable to facilitate theft
predict labels by various means throughout investigations. This case shows how there is
the AI application lifecycle. There are also intrinsic business value in making AI explainable.
companies that have successfully navigated
this dilemma and have produced remarkable
explainable AI use cases in spite of this trade-
USE CASE 2:
off. Below, we will examine two such cases7. EXPLAINABLE AI IN
CLIMATE PREDICTION
USE CASE 1: Droughts are one of the most challenging
EXPLAINABLE AI IN THE natural hazards to be predicted owing to
multiple predictors and non-linearity between

6. Introducing AI Explainability 360 - IBM Blog


7. Would You Trust An Artificially-Intelligent Expert? - National Law Review

METALAB REVIEW ¦ 135


them. A paper by Dikshit and Pradhan (2021) risks, and even fewer (38 percent) report
used explainable AI to help local governments actively working to address them. Their
provide more transparency to the public. study emphasizes that AI systems have
Here, explainability was essentially for them revolutionized how companies perform and
to interpret models in the field of drought make a fundamental impact on consumers
forecasting to preemptively examine future across all industries.10 Yet analysts of the
drought scenarios. This also provides a esteemed consulting firm also insist that
nice use case where the right balance of there is still a large opportunity to push the
performance and explainability was achieved. boundaries of innovation and value creation
The pair of researchers used deep learning in AI if the inherent risks associated with it
algorithms while incorporating some are addressed. Inevitably, one of these is un-
techniques for explainability. Their solution explainability. The same research by McKinsey
had an accuracy of around 85%, which was highlights that companies with the highest AI
sufficient for drought prediction. While tuning returns are more likely to have addressed the
the parameters and model could have led to technology's risks.
higher accuracy measures, they would have
risked convoluting results to an extent that
made explainable results impossible.8

WHEN DOES AI
EXPLAINABILITY
ULTIMATELY MATTER? A
EUROPEAN SOLUTION
The European Union became the first
regulatory body that has issued an exhaustive
list of regulations on AI systems. The objective
as portrayed by the EU is to encourage AI
excellence along with Trustworthy AI. On
one hand, the Commission plans to invest
€1 billion per year in AI to encourage and
spread the usage of the technology, and on
the other hand, proposed three legal initiatives
to regulate AI. One of these initiatives is a
framework that protects fundamental rights &
addresses safety risks.9
© Andrea Piacquadio on Pexels

According to a report by McKinsey, they


found that only 48 percent of organizations
report recognizing regulatory compliance

8. Interpretable and explainable AI (XAI) model for spatial drought prediction - ScienceDirect
9. Shaping Europe’s digital future - Regulatory Framework - European Union
10. What the draft European Union AI regulations mean for business - McKinsey

136 ¦ METALAB REVIEW


From Mickinsey’s research, we see that investments in explainability can have a large upside even
from a purely business standpoint. But from the EU’s risk classification in Figure D below, we can
also infer that explainable AI is not always necessary. Because there are trade-offs associated
with the performance of algorithms, it may be prudent to apply them to situations where they
matter most, such as those deemed “Unacceptable” or “High-risk” in their classification. These
scenarios come with AI applications where safety, individual freedoms, lifelong economic
opportunities, and other high-stake affairs are at stake. For more casual, and commercial oriented
applications, such as chatbots, automated emails, and inventory optimization which would be
classified as “limited” or “minimal risk,” it would make sense to prioritize AI that makes strong
predictions over AI that is more explainable.11

FIGURE D: EU RISK CLASSIFICATION OF AI


The European Union’s draft Ai regulations classify
AI systems into three risk categories

Unacceptable-risk High-risk Limited- and minimal-risk


AI systems Ai systems AI systems
Subliminal, manipulative, or Systems that evaluate consumer AI chatbots
exploitative techniques causing creditworthiness
harm AI-enabled video and computer
Recruiting or employee- games
Real-time, remote biometric management systems
identification systems used in Spam filters
public Systems utilizing biometric
Inventory-management systems
All forms of social scoring Safety-critical systems (eg,
systems that would put the health Customer- and market-
of citizens at risk due to failure) segmentation systems
Any systems used in the Most other AI systems
administration of justice

In any case, the next article in this series will shed light on how companies that utilize AI can
be incentivized to build trust with consumers, central to which is having explainable AI, in-spite
of this performance-explainability trade-off. It will address questions such as “what is truly
preventing explainable AI from becoming part of the status quo?” and “what are the different
incentives that each stakeholder faces?” - from firms, to government bodies, to regulators, and
AI users. Finally, it will also push for an agenda in which each stakeholder can learn to trust each
other, given the constraints and potential incentives they face, in order to fundamentally facilitate
critical dialogue that leads to explainable AI being found where it matters most.12

11. Shaping Europe’s digital future - Regulatory Framework - European Union


12. The Three Core Pillars Of Responsible AI - Forbes

METALAB REVIEW ¦ 137


EXPLAINABLE AI:
PART 2 - CAN WE
HAVE A STATUS
QUO WHERE AI IS
EXPLAINABLE?
By NELBERTO NICHOLAS QUINTO & KUNAL PUROHIT
Master in Management at ESSEC Business School

A
s established in Part 1, knowing how are at stake, how should regulators ensure that
a decision is made by AI and which organizations make the right choices? And when
parties can be accountable for it are of algorithmic design can always be hidden under
paramount importance where there is high or the veil of proprietary secrecy, what hope is there
unacceptable risk under the EU commission’s for fairness, justice, and transparency when it’s
classification. In scenarios where explainability is needed most?
consequential, the adoption of AI and the ability
to realize the efficiencies that such a technology In order to fully harness the benefits that AI can
brings is much less straightforward. Explainability bring, it is essential to move from a system of
is a key ingredient of establishing trust between simply harnessing its ability to predict, automate,
the users of AI and those who produce it. One can and decide, towards a system where different
trust an algorithm for being accurate and reliable- stakeholders trust the technology (and each
i.e. predicting well and ‘getting the job done’ - but, other!). This is exceedingly important in high-
in many circumstances, this simply is not enough. stakes environments where the returns of
having trustable AI are magnified. But if adoption
Faced with a trade-off between performance necessitates algorithms to be trusted by users
and explainability, how should firms and to predict with precision, sound-judgment, and
governments alike go about choosing one over fairness, what does it really mean to trust AI?
the other? Furthermore, when it comes to risky
contexts where lives and individual freedoms

138 ¦ METALAB REVIEW


© Patrick Perkins on Unsplash
According to Arun Das and Paul Rad of IEEE, it just one of the many components of trust?
the world’s largest technical professional If so, how does explainability interact with all
organizations dedicated to advancing technology the others?
for the benefit of humanity:
According to Mckinsey & Co., there are
“Trustability of artificial intelligence models several components of trust in AI algorithms
is a measure of confidence, as humans, as apart from explainability. These include
end-users, in the intended working of a given performance, fairness, and transparency.
model in dynamic real-world environments. Since each of them is intricately linked to
Thus, ‘Why a particular decision was explainability, having AI that is explainable
made’ is of prime importance to improve becomes both central and critical to
the trust of end-users including subject producing trustable AI. With performance,
matter experts, developers, law-makers, and for instance, whether measured by accuracy
laypersons alike” (Das & Rad, 2020). or precision, there is the potential trade-
off with explainability, as discussed in the
In this definition, explainability (why a previous article. Then there’s the issue of
decision was made) appears to be vital in fairness which is a measure of how much
trusting AI. But is explainability the only thing bias-free decision making is present in an
that matters? Is it synonymous to trust? Or is AI. Automated decisions can be skewed and

METALAB REVIEW ¦ 139


made unfair with respect to certain groups
or individuals especially when sensitive
variables such as gender, race, disability,
etc., are involved. But in order to gauge
whether an algorithm biases decisions or
not, it simply must be understood first. Lastly,
even if underlying algorithms are deemed
reliable and marketed as ‘fair,’ end-users still

© Glenn Carstens Peters on Unsplash


fear that AI can be misused to exploit their
personal information, collect and process
more sensitive data than is necessary, or
simply not be as bias-free as advertised.
This underscores the value of transparency
in how algorithms are used for decision-
making. And just like fairness, it is difficult to
be transparent when the inner-workings of a home for your family, you would expect
an algorithm are not well understood to begin that decisions, especially those related to
with. Therefore, while performance comes at rejections, could be explained by the bank.
the expense of algorithmic obscurity, other Even successful clients that were given
trust components such as fairness and certain borrowing rates would sometimes
transparency necessitate explainable AI. want to ensure that the right factors were
considered- and sensitive matters such as
WHAT’S PREVENTING race or gender would be left out. In the same
way, regulators such as the U.S. Consumer
EXPLAINABLE AI? Financial Protection Bureau would want to
make sure that consumers were treated
If explainability is so crucial for end-users to fairly in such a process. But because of the
build trust with AI, and if it matters for riskier classic trade-off between performance and
cases, then what’s stopping organizations explainability, if a bank invests in making their
from adopting it? Is it simply because of the algorithms explainable, they may lose some
performance trade-off? To understand this, degree of predictive ability that can lead to a
we turn to the dynamics between different AI very costly increase in the number of defaults.
stakeholders (organizations, end-users, and
regulators) and examine the familiar case of From the scenario above, we can postulate
mortgage applications. that end-users prefer products that they
can trust while organizations face decisions
Organizations, usually firms or government related to calibrating the features that affect
institutions, produce AI for commercial, the level of explainability and performance
administrative, or policy-making purposes. needed to have a trustworthy and reliable
They also use the technology to improve product. Furthermore, despite its appeal on
their goods, operations, and services. When establishing consumer trust, explainability
we think of mortgages, these are the banks doesn’t necessarily become a de facto priority
that use credit scoring and risk assessment since organizations have to account for other
algorithms to give automated loan decisions. factors. These include the impact on profits
They supply or use this “AI service” to “end- due to losing predictive capabilities, feasibility
users,” in this case, the loan applicants. constraints of implementing explainable
For something as important as providing AI, and the risk of underlying algorithms

140 ¦ METALAB REVIEW


being replicated by competitors. For these 41% said they would complain in case an AI
reasons, in order for explainable AI to be part interaction resulted in ethical issues, 36%
of status-quo, regulators are left with the would demand an explanation, and 34%
problem of creating the necessary incentives would stop interacting with the company.
and standards for banks, technology firms, From this information, we see that in broad
and all AI-producing bodies alike to invest in measures, it is possible to lose revenue when
explainability. an AI product or service is not explainable. But
does the converse hold true? Do investments
INCENTIVES FOR FIRMS in explainability normally translate into more
revenue? This depends on how explainability
When the organization that supplies AI is a affects a product’s competitive advantage.
firm, their default goal is to maximize profit. Which, in turn, is a function of two things: how
So investing in explainability only makes important explainable AI is to end-users in a
sense if it increases their bottom line. Since given market and how much more trustable
profit is simply the difference between total AI can be when its predictive reliability is
revenue and all costs, firms will usually dampened by making it more explainable.
invest if one of two conditions are met. First
the marginal increase in trust produced by What this all means for firms is that they
making AI more explainable (often at the have to understand just how important
expense of predictive ability) must create explainability is to their end-users, and
an advantage that drives revenue for the whether investing in efforts to increase the
firm. And second, investing in explainability valuation of end-users’ trust in AI can be a
must reduce costs. Indeed, to produce more source of differentiation among competition,
explainable AI, organizations can tell their and hence a possible source of competitive
engineers to change pre-processing data advantage. But while explainable AI can make
techniques, embed new tools, use different an organization more competitive in this
datasets and the like- all of which entail some manner, it could also lead rivals to learning
costs in the early stages of AI development. about the intricacies of a firm’s algorithm.
But most of these expenses related to If such inner working of a firm’s AI practice
investments in explainability do not come were made available, then any advantage
from tuning algorithms to make them more derived from intellectual property would
explainable, fair, or transparent- rather, they be at risk. For this reason, Rudina Sesseri,
come from costs associated with regulatory managing partner of Glasswing Ventures,
compliance. Notably, a third, necessary but insists that: “if a startup created a proprietary
not sufficient condition, is that firms are able AI system, for instance, and was compelled
to facilitate adoption and ensure the correct to explain exactly how it worked, to the point
use of AI for both their staff and, ideally, their of laying it all out — it would be akin to asking
customers. that a company disclose its source code.”
She further posits that regulators who push
DEMAND FOR TRUSTABLE for transparency should be vigilant as such
policies can stifle innovation by killing-off
AI startups whose value and ability to compete
with incumbents is primarily driven by their
In a study based on interviews with 4,400 intellectual property.
end-users of AI, Capgemini found that
their views on ethics and AI threaten both Firms are then left with the decision of
company reputation and the bottom line - weighing any potential upside of having

METALAB REVIEW ¦ 141


© Desola Lanre-Ologun on Unsplash
explainable AI against the probability and matters further is that audit efficacy is
consequences of competition being able to sometimes affected by explainability. When
benefit from learning about their algorithmic it is not, then regulatory audits will normally
blueprint. This serves as an opportunity for suffice to induce investments if the economics
regulators to enact the right policies that make sense. However, recent research
could protect proprietary algorithms while by economists Xavier Lambin and Adrien
incentivizing firms to invest in explainable Raizonville (2021) informs us that when
AI. But non-compliance of such policies explainability facilitates regulatory audits, then
entails costs to firms, from direct expenses firms may potentially hide their misconduct by
such as fines to less direct ones such as the making their algorithms less understandable
opportunity cost of not being able to provide for regulators. In other words, explainable AI
certain goods, features, or services in a given could lead to ‘regulatory opportunism’ which
market due to regulatory blockage. refers to the act of regulators focusing their
scrutiny on firms whose algorithms are simply
REGULATION AND ‘easier to understand’ than others, thereby
making regulatory oversight counterproductive
COSTS in incentivizing explainability. In summary,
we see that as far as costs are concerned,
When firms face fees for non-compliance, firms would invest in explainable AI if non-
investments in explainable AI are rarely worth compliance costs generally exceed investment
it if they exceed such costs. What complicates costs and if they are in a regulatory climate

142 ¦ METALAB REVIEW


where explainability does not facilitate audits need for regulators to exert the right pressure
ex ante. If the latter doesn’t hold however, for firms to adopt explainability in high-risk
then firms need to trust that regulators won’t environments.
engage in ‘regulatory opportunism’ once their
algorithms are made explainable. To merit Another reason for non-adoption of explainable
such trust, regulators also need to signal that AI in the public sector is when understanding
they won’t do so even if they technically could. the causes of a decision simply does not
matter nearly as much as getting accurate, fast
INCENTIVES FOR results. Examples include allocating emergency
resources, rescue operations, tracking
GOVERNMENT criminal activity, or administering radiology
INSTITUTIONS treatment. While these examples would not
fall under limited or minimal risk under the
For government institutions that both produce EU commission’s classification, speed and
and use AI, the dynamics are not too different precision can outweigh value-judgements
from what we’ve seen with firms. While associated with explainability in such cases,
they may not need to generate revenues, thereby bringing us back to the performance-
firms need to ensure that the AI powered explainability trade-off.
services they’re providing still produce reliable
predictions if made explainable. Since public
services such as judicial systems, defense, WHAT’S NEXT
healthcare, transportation, etc. are all high-risk
environments that can fundamentally affect the There needs to be critical dialogue to happen
lives and freedoms of citizens, it is safe to say between organizations, end-users, and
that the benefits of having a trustable, fair, and regulators for explainable AI to be ingrained
transparent algorithm outweigh the need for into status-quo. Just as organizations need
high performance algorithms in most cases. to produce AI that end-users can trust,
regulators need to earn the trust of firms that
So when will government agencies not invest marginal investments in explainability does
in explainable AI? An obvious case is when not increase chances of future misconduct
contracting firms do not provide such AI in to be detected and does not result in losing
the first place. A lot of government agencies intellectual property advantages. Similarly,
use AI but do not necessarily produce them. just as regulators need to signal to firms that
Instead, they contract these services from they can be trusted to protect their interests
traditional for-profit firms. Hence, when such should firms comply, end-users need to
firms do not meet the conditions mentioned signal to organizations that trustworthy
above, then government agencies that contract algorithms matter to them.
them are left to use what is available. Such is
the case with COMPAS, a highly controversial
AI solution which is currently being used
in criminal cases to predict a defendant’s
likelihood to re-offend. Despite several lawsuits
and large public scandals that came its way,
this judicial algorithm is still widely used
in the United States because the company
contracted by the government has not invested
in explainable AI. This further underscores the

METALAB REVIEW ¦ 143


HIGHLIGHTS OF THE YEAR

144 ¦ METALAB REVIEW


HIGHLIGHTS OF THE YEAR

METALAB IDEAS WORKSHOP


In April 2023, Metalab IDEAS organized a workshop session at ESSEC in order to promote the
production of content and to foster interactions with the students, staff and professors of
ESSEC. This workshop consisted on a poster session, together with two thematic roundtables
addressing topics of key interest:

• STRIKING A BALANCE BETWEEN • ARE AI AND ENVIRONMENTAL


PROGRESS AND CAUTION IN SUSTAINABILITY COMPATIBLE?
GENERATIVE AI: AN ACHIEVABLE
GOAL? The conversation revolved around
the compatibility between AI and
The roundtable explored the challenge sustainability. The discussion was based
of balancing progress and caution in on the UN's sustainability definition,
generative AI. It began with a discussion covering topics like climate change,
about the term "generative AI" and clean energy, and clean water. Key points
whether it implies creating entirely new included the need for a "green AI" to
objects, considering that AI models are reduce energy consumption and carbon
trained on existing data. Trust issues emissions, the role of AI in addressing
in recent generative AI advancements, climate issues and promoting sustainable
particularly in image generation, were practices in various sectors.
highlighted.
It was noted that greenwashing in the
The discussion covered the need for AI industry needs regulation, as AI
defining generative AI and the tensions contributes to a significant percentage
between innovation and potential risks. of carbon emissions. There was also a
Key points included the role of regulation focus on AI's potential to predict climate
and technological solutions in ensuring change and adaptation, especially in
AI safety, the distinction between agriculture, a major contributor to carbon
human and AI-generated creativity, and emissions.
the importance of representation and
monetary incentives in value creation. Additionally, trends in renewable energy
and concerns about materials disposal
The discussion also touched on and mining were discussed, emphasizing
public misconceptions, the purpose the importance of collaboration and
of a moratorium on AI innovation, and sustainable practices.
concerns about the timing of regulation
and reputational risks for companies.

METALAB REVIEW ¦ 145


HIGHLIGHTS OF THE YEAR

ROUNDTABLE ON AI & SUSTAINABILITY

In May 2023, a roundtable on AI & Sustainability was organized by the


Metalab in collaboration with the Accenture Chair in Business Analytics & the
ESSEC|CentraleSupélec Master in Data Sciences & Business Analytics.

This event gathered 3 key speakers: Marie Klaeyle, Associate Partner, Expert in
Digital & Sustainability at McKinsey & Co, Guillaume Fauconnier, CEO-Founder at
MARKENER and Benoit Lepetit, Group Chief Data & Analytics Officer at Saint-Gobain
– along with about 80 ESSEC students.

This event was an opportunity to engage in a discussion with the students on


issues related to data and AI, especially when it comes to the energy transition,
decarbonization and sustainability.

More precisely, the students involved raised crucial questions on the potential
negative impacts of AI on sustainability, and how we can mitigate these risks. Also,
issues were raised on making AI development and implementation transparent
and accountable. The panelists gave concrete examples of some of the successful
applications of AI in sustainability efforts they have witnessed during their career.

146 ¦ METALAB REVIEW


HIGHLIGHTS OF THE YEAR

POSTER SESSION
As part of the workshop, IDEAS editors and analysts introduced the ESSEC population to
their articles and topic interests through the design of a poster session.

In total, 11 posters were displayed and allowed members of the think tank to engage in a
discussion with the public, presenting their ideas, topics and reflections on matters such as
digital currency, AI safety, predictive analytics, and green AI solutions’ challenges.

NEW PODCAST SERIES LAUNCHED


On another side, students of Metalab IDEAS co-produced a podcast series named "Tech
Tide: A Wave of IDEAS", along with students of the Tech ESSEC association. Its purpose is to
shed light on the impact of technology and AI developments on companies, society and the
economy. Every episode interviews one or several key players.

A first episode on ChatGPT and Generative AI was released with the participation of Maciej
Workiewicz, Associate Professor of Management at ESSEC, who was interviewed by Carla
Pinto, student in the Global BBA program, and Sam Nelberto Quinto, PhD student in the
Marketing Department.

Throughout this exchange, Maciej not only outlined how Generative AI has started to reshape
organizations, but also delved into the future transformations it could bring to society and
the education sector. Displaying multiple arrays of depth and
various considerations when it comes to our adaptation to new
technologies, this podcast explores different approaches to the
rising theme of Large Language Models.

Available on Soundcloud and Spotify

METALAB REVIEW ¦ 147


ESSEC Metalab is a
multidisciplinary center
at the intersection of
AI & data science, new
technologies and their
impact on society.

Through its publications,


ESSEC Business School it aims to highlight the
3 avenue Bernard-Hirsch contributions of its
CS 50105 Cergy students and member-
95021 Cergy-Pontoise Cedex researchers, whose
France
Tel. +33 (0)1 34 43 30 00
reflections shed light on
www.essec.edu the evolution of society
with regard to the major
shifts generated by the
ESSEC Executive Education
deployment and massive
CNIT BP 230
92053 Paris-La Défense
adoption of disruptive
France technologies.
Tel. +33 (0)1 46 92 49 00
www.executive-education.essec.edu
The activities conducted by the IDEAS
think tank are made possible thanks
ESSEC Asia-Pacific to the support and contribution of
5 Nepal Park onepoint.
Singapour 139408
Tel. +65 6884 9780
© ESSEC Business School - 1338030923 - Not printable file.

www.essec.edu/asia

ESSEC Africa
Plage des Nations - Golf City
Route de Kénitra
Sidi Bouknadel (Rabat-Salé)
Maroc
Tel. +212 (0)5 37 82 40 00
www.essec.edu

You might also like