Professional Documents
Culture Documents
Ai For The Earth Jan 2018
Ai For The Earth Jan 2018
Industrialisation has led to many of the world’s current Harnessing these opportunities and proactively
environmental problems. For example, climate change, managing these risks will require a transformation of
unsafe levels of air pollution, the depletion of fishing the “enabling environment”, namely the governance
stocks, toxins in rivers and soils, overflowing levels of frameworks and policy protocols, investment and
waste on land and in the ocean, loss of biodiversity and financing models, the prevailing incentives for
deforestation can all be traced to industrialisation. technology development, and the nature of societal
engagement. This transformation will not happen
As the Fourth Industrial Revolution gathers pace,
automatically. It will require proactive collaboration
innovations are becoming faster, more efficient and
between policymakers, scientists, civil society,
more widely accessible than before. Technology is also
technology champions and investors.
becoming increasingly connected; in particular we are
seeing a merging of digital, physical and biological If we get it right, it could create a sustainability
realms. New technologies are enabling societal shifts revolution.
by having an effect on economics, values, identities and
This “Fourth Industrial Revolution for the Earth”
possibilities for future generations.
series is designed to illustrate the potential of Fourth
We have a unique opportunity to harness this Fourth Industrial Revolution innovations and their application
Industrial Revolution, and the societal shifts it triggers, to the world’s most pressing environmental challenges.
to help address environmental issues and redesign how It offers insights into the emerging opportunities and
we manage our shared global environment. The Fourth risks, and highlights the roles various actors could play
Industrial Revolution could, however, also exacerbate to ensure these technologies are harnessed and scaled
existing threats to environmental security or create effectively. It is not intended to be conclusive, but
entirely new risks that will need to be considered rather to stimulate a discussion between diverse
and managed. stakeholders to provide a foundation for further
collaborative work. This paper looks at artificial
intelligence and the Earth.
Climate change Biodiversity and Healthy Oceans Water security Clean air Weather and
conservation disaster
resilience
• Clean power • Habitat • Fishing • Water supply • Filtering and • Prediction and
• Smart transport protection and sustainably • Catchment capture forecasting
options restoration • Preventing control • Monitoring and • Early warning
• Sustainable • Sustainable pollution • Water prevention systems
production and trade • Protecting efficiency • Early warning • Resilient
consumption • Pollution habitats • Adequate • Clean fuels infrastructure
• Sustainable control • Protecting sanitation • Financial
• Real-time,
land-use • Invasive species • Drought integrated, instruments
• Smart cities species and • Impacts from planning adaptive urban • Resilience
and homes disease control climate change management planning
• Realising (including
natural capital acidification)
In meeting these challenges, there is wide scope for the most prominent innovations, and to provide an
innovation and investment. AI in particular has initial overview.
immense potential to help unlock solutions. Indeed,
the annex section provides a summary of research into Currently, most of these applications focus on
more than 80 existing AI use cases for the automated and assisted intelligence to unlock value
environment that we uncovered through desk-based from large unstructured real-time datasets. Future
research and interviews with a range of stakeholders applications will likely involve more systems propelled
at the forefront of applying AI across industry, big by autonomous decision-making where AI acts
tech, entrepreneurs, research and government. independently, thus creating new opportunities and
risks. The challenge for innovators, investors and
In the next section, we highlight, by environmental
governments is to identify and scale these pioneering
challenge area, the broad range of emerging use cases
innovations, and also to make sustainability
across relevant action areas. As can be seen from the
considerations central to wider AI development and
Figure 3, each priority Earth challenge stands to
benefit in a myriad of ways from AI. The snapshots are use.
not meant to be exhaustive, but to be representative of
Climate change
• Optimised energy system • On-demand shared transport
forecasting mobility
• Smart grids for electricity use • AI-enabled electric cars
• Predict solar flares for protecting • Autonomous vehicles for efficient
power grids transport
• Renewable energy plant • Vehicle to infrastructure
assessments communication and optimisation
• Optimised decentralised & • Optimised traffic flows
peer-to-peer renewable • Integrated cost-efficient transport
energy systems systems
• Optimised virtual power plants • Demand-response charging
infrastructure
• Smart traffic light & parking Climate change • Supply chain monitoring and
systems for urban mobility transparency
management • Active optimisation of industrial
• Optimised sustainable machinery & manufacturing
building design • Digital twins for lifespan
• Energy-efficient building Sustainable performance optimisation
management systems land-use
• Smarter fresh-food replenishment
• Auditory responsive lighting • Smart recycling systems
& heating
• Integrated municipal &
• Optimised urban-level energy industrial waste management
generation and use
• Early crop yield prediction
• Analytics & automation for smart
• Precision agriculture & nutrition
urban planning
• Hyper-local weather forecasting
for crop management
• Early detection of crop issues
• Automated & enhanced land-use
change detection for avoided
deforestation
• Monitoring health & well-being in
livestock farming
Source: PwC research
AI has the potential to transform the way in which to solar and wind power generation plants to supply
climate change is tackled. In clean power, for example, data for machine learning monitoring capability,
machine learning is being used to match energy enabling remote inspection of sites, predictive
generation and demand in real-time, realising more maintenance, and energy resource forecasting 23. This
fully the potential of “smart grids”, decreasing increases control and maintenance efficiency lowering
unpredictability, and increasing efficiency, power costs of solar and wind energy.
balancing, use, and storage of renewable energy21. For
Within buildings, machine learning algorithms are
example, Agder Energi22 is using AI and the Cloud to
also being deployed to analyse data from millions of
predict and prepare for changing energy needs in
smart sensors and meters to provide predictions on
Norway, particularly given the rapidly-increasing
energy usage requirements and cost24. AI is also being
penetration of electric vehicles. Such approaches can
used to provide auditory cue responsive lighting and
also lower the need for excess ‘idle’ capacity. Neural
heating from buildings to streets to optimise energy
networks for renewable power are also being
use, while JTC25 of Singapore is using AI to monitor,
developed to improve the energy efficiency and
analyse and optimise energy efficiency in buildings.
reliability. For example, DNV GL use sensors attached
Machine learning algorithms are also being used at the
• Machine-automated
biodiversity analysis
• Smart mosquito traps
• Plant disease identification
& detection
AI has the potential to transform the ways by which we PlanetWatchers insights - using precision monitoring
monitor and conserve habitats. For example, AI of landscapes - provides a resource for management of
provides the backbone for applications that, combined forest habitats to address the challenges presented by
with satellite imagery, can automatically detect land- climate change related disturbances such as pests,
use changes, including cover analysis and forests, damage, drought and fire.32
vegetation and monitoring of floods. For example,
AI techniques are opening up various new approaches partnership with the Ocean Alliance uses drones to
to protect and sustainably manage oceans.35 Systems collect mucus samples from whales off the coasts of
that use AI in combination with other techniques to Patagonia, Mexico and Alaska to obtain DNA
gather data in hard-to-reach ocean locations support information, and scientists use AI to gauge the
efforts to track provenance and fish sustainably, mammals’ health – and by extension, measure the
protect species and habitats, and to monitor the ocean habitat in which they live – in real-time37.
impacts of climate change.
Ocean conditions can also be monitored using
AI is also unlocking new solutions to tackle illegal AI-powered robots for detecting pollution levels and
fishing. Machine learning techniques are being tracking changes in temperature and pH of the oceans
pioneered to guide more accurate patrol schedules, due to climate change. Moreover, NASA uses satellite
and early efforts are underway to apply vessel imagery and machine learning computer modelling to
algorithmic patterns to satellite data combined with predict the current and future conditions of the
Automatic Identification System (AIS) data from ships world’s oceanic phytoplankton38. Autonomous ocean
to monitor illegal fishing activities (e.g., Global Fishing exploration technologies - utilising advances in AI,
Watch)36. Such tracking will enable authorities to robotics and nanotechnology - are also under
prevent overfishing and to control fisheries. development to help survey the ocean floor at high
resolution to help with species identification and
For species protection, some systems use image
mapping and natural resource management.39
analytics and machine learning to track the numbers
and locations of invasive species. One industry-NGO
Water is at the nexus of food, energy, environment and Syracuse, N.Y. uses an AI system to analyse its aging
urban issues. Enabled by AI, scientists and engineers water infrastructure to identify specific locations of
can simulate the performance of reservoirs and project leaks-prone pipes to repair42. While Water Planet’s
water usage for a geographical area, in combination IntelliFlux incorporates AI to analyse data from
with weather forecasts, making better informed policy pressure sensors and determine optimal performance
decisions. Valor Water Analytics, meanwhile, is of filtration systems, minimising water loss43.
combining AI with industry intelligence and As well as supply and efficiency, AI - working with
operational interactions to manage smart meter satellite data - can help forecast weather patterns and
assets40. Their approach enables them to identify analyse soil and surface water conditions to predict
leaks, understand water flows in real-time, and see drought conditions to help people and sectors
whether meters are malfunctioning. Elsewhere, Water affected44. Scientists can also use machine learning
Smart Software offers a data analytics platform, combined with physical models to conduct water plan
utilising machine learning, to provide utilities with scenarios and evaluate capital investments, crisis
information and strategies, including the ability to management plans, and potential outcomes of
check water flows or spot anomalies41. Moreover, Flo water-planning decisions.
Technologies uses machine learning to provide real-
time data on water quality sending alerts to user's
smartphones.
• Advanced battery and fuel-cell design Clean Air • 2-10 day pollution level
• Advanced battery components forecasting
• Pollution forecasting for transport • Air quality alerts
management
For clean air solutions, some of the early examples traditional physics-based models of atmospheric
concentrated around filtration and capture. For chemistry and weather with machine learning models.
filtration, air purifiers (e.g. ARCADYA’s) use machine In terms of air quality alerts, such AI-based systems
learning to record air quality and environment data can now provide forecasts of resource intensive and
in real-time and adapt filtration efficiency45. AI polluting behaviours. Simulations powered by AI can
applications are also driving advances in real-time air enable residents of urban areas, such as Beijing, to
quality monitoring; for example, the company AirTick receive warnings about air quality.49
uses smartphone cameras as a proxy for air pollution
Moreover, the use of AI in new connected platforms
sensors harnessing image recognition and machine
that harness data from vehicles, radar sensors, and
algorithms to analyse images across a city at low
cameras to optimise traffic flow in urban areas is also
cost46. Elsewhere, air pollution forecasting tools are
improving air pollution due to its impact on reducing
being developed by start-up AirVisual, IBM, and
stationary vehicles and stop-start driving.50 In terms of
Microsoft for cities like Beijing47. IBM’s Green
mobility, AI is also being used to optimise advanced
Horizons initiative combines machine learning and
battery design to improve the effectiveness and
IoT, harnessing data from air quality stations and
efficiency of electric vehicles, whose increased uptake
more widespread sources, such as traffic systems,
will further improve air quality.
weather satellites, and stations, as well as industrial
activity, topographic maps, and even social media,
to develop predictive analytics for 2 to 7-10 day
forecasts48. Both IBM’s and Microsoft’s tools blend
Many of the emerging applications for weather and using big data analytics and AI combined with more
disaster resilience focus on the ability to forecast traditional physics-based modelling approaches to
extreme weather and natural disasters. Predictive model the impact of extreme weather events on
analytics powered by AI, along with IoT, drones, infrastructure and systems to inform disaster risk
blockchain, and advanced sensor platforms can help management strategies. The models can be used
governments and the scientific communities monitor to predict direct damages in addition to loss
tremors, floods and windstorms, as well as sea level amplification due to business interruption risks
changes and other possible natural hazards, in real- from electricity outages or transport closures. AI
time with thresholds for automated triggers, that simulations are also being applied to evaluate
enable early evacuations when needed. In Indonesia, disaster resilience strategies.
PetaBencana.id combines multiple open-source In addition to predicting extreme weather and natural
sensors, AI, and people’s social media reports for real- disasters, natural language processing and machine
time flood mapping in the capital, Jakarta.51 AI is also learning techniques are increasingly being used to
being used with image analytics to process social communicate disaster information to the public in
media information to provide real-time extreme response to queries. Moreover, in terms of real-time
weather forecasts based on people’s images and posts, response planning, deep-learning algorithms and
for example, IBM building on The Weather Company image analytics can use seismic data, structural data
whom they acquired52. In addition, The Yield53 is a for buildings (age of structure, materials used etc.),
Tasmanian agtech company using sensors, analytics social media data, and also satellite images to
and apps to produce real-time weather data, helping coordinate and prioritise disaster relief efforts, from
growers make smarter decisions reducing water and determine which parts of a city will be most at risk to
other inputs. monitoring the flow of people and resources54.
A number of meteorological agencies, tech companies
(e.g. IBM, Palantir), insurers, and utilities are also
The speed and effectiveness with which organisations Household smart meters can produce large volumes of
and people can respond to disasters has a substantial data that can be sued to predict water flows, spot
impact on the extent of economic losses and human inconsistencies and check leaks. The next stage will be
suffering, particularly in the most catastrophic events. to combine machine learning, the Internet of Things
But delays often occur due to a lack of information, and blockchain to create a truly decentralised water
analytical insight and awareness of the best course of system, where local resources and closed-loop water
action. Often the necessary data exists in large part, recycling gain value. Water resources could even be
but is segregated among various organisations and is traded via blockchain.
thus mostly inaccessible to communities. Furthermore, machine learning, predictive modelling
Better resiliency planning is also an important and robotics can be combined to transform current
component to mitigate the damage of future natural approaches to building and managing water
catastrophes. AI can be used to sort through infrastructure and to accelerate innovation in
multidimensional data about a region and identify environmental engineering. Rivers, for example, could
which aspects have the biggest impact on resilience. AI be engineered to adjust autonomously their own
can run and analyse simulations of different weather sediment flows. Coupled with AI-informed pricing,
events and disasters in a region to seek out such approaches could optimise water usage and drive
vulnerabilities and identify the resiliency plans that behaviour change by providing incentives for water
are most robust across a range of event types. conservation.
Performance Security risks Control risks Ethical risks Economic risks Societal risks
risks
• Risk of errors • Cyber intrusion • Risk of AI • “Lack of • Job- • Risk of
• Risk of bias risks going “rogue” values” risk displacement autonomous
• Privacy risks • Inability to • Value risks weapons
• Risk of proliferation
opaqueness or • Open-source control alignment risk • “Winner-takes-
“black box” risk software risks malevolent AI • Goal all” concentration • Risk of
alignment risk of power risk “intelligence
• Risk of divide”
explainability • Liability risk
• Risk of stability • Reputation risk
of performance
Algorithmic game Systems that address the economic and social computing
theory and dimensions of AI, such as how systems can handle potentially
computational social misaligned incentives, including self-interested human
choice participants or firms and the automated AI-based agents
representing them.
Soft robotics (robotic Automation of repetitive tasks and common processes such
process automation) as IT, customer servicing and sales without the need to
transform existing IT system maps.
1Clark, Liat, Google’s Global Fishing Watch is using ‘manipulated’ data, Wired, November 2014, available at:
http://www.wired.co.uk/article/global-fishing-watch-false-data-windward
2 Gilbert, Elissa,
Scientists equipped “SnotBots” — drones using sophisticated AI programs — to learn about whales, oceans and even human
health, August 2017, available at: https://iq.intel.com/whale-snot-hold-secret-ocean-
health/?cid=sem43700027467499372&intel_term=parley+for+the+oceans&gclid=EAIaIQobChMIhO6LzPqW2AIVAtVkCh1kMgRSEAAYAiAAEg
L5p_D_BwE&gclsrc=aw.ds&dclid=CKuCpfP6ltgCFVIFgQodTewB-g
The Fourth Industrial Revolution for the Earth Develop a set of insight papers, taking a deep
initiative is designed to raise awareness and accelerate dive into the possibilities of the Fourth Industrial
progress across this agenda for the benefit of society. Revolution and each of these issues.
In the first phase of the project, specific environmental Build new networks of practitioners and
focus areas will be considered in depth, exploring in support them to co-design and innovate for action
detail how to harness Fourth Industrial Revolution on the environment in each of these issue areas,
innovations to better manage the world’s most leveraging the latest technologies and research that
pressing environmental challenges. Initial focus areas the Fourth Industrial Revolution offers
will include:
Design a public-private accelerator for
Air pollution action, enabling both government, foundational,
Biodiversity research organisation, and commercial funds to be
pooled and deployed into scaling innovative Fourth
Cities Industrial Revolution solutions for the
environment.
Climate change and greenhouse gas monitoring
Help government stakeholders to develop and
Food systems
trial the requisite policy protocols that will
Oceans help Fourth Industrial Revolution solutions for the
environment to take hold and develop.
Water resources and sanitation.
The Fourth Industrial Revolution for the Earth
Working from these thematic areas, the World initiative will be driven jointly out of the World
Economic Forum, supported by Stanford University Economic Forum Center for the Fourth Industrial
and PwC (as project adviser) and advised by the Revolution in San Francisco and other Forum offices
members of the Global Future Councils on the Future in New York, Geneva and Beijing.
of Environment and Natural Resource Security and
specific Fourth Industrial Revolution technology
At PwC, our purpose is to build trust in society and solve important problems. We’re a network of firms in 158 countries with more than 236,000 people who are committed to delivering quality in
assurance, advisory and tax services. Fine out more and tell us what matters to you by visiting us at www.pwc.com.
© 2018 PwC. All rights reserved. PwC refers to the PwC network and/or one or more of its member firms, each of which is a separate legal entity. Please see www.pwc.com/structure for further
details. This content is for general information purposes only, and should not be used as a substitute for consultation with professional advisors.