Professional Documents
Culture Documents
Ethical Assessment of AI Cannot Ignore Cultural Pluralism: A Call For Broader Perspective On AI Ethics
Ethical Assessment of AI Cannot Ignore Cultural Pluralism: A Call For Broader Perspective On AI Ethics
1
Article 4 of the Universal Declaration on Cultural Diversity, adopted by
the General Conference of the United Nations Educational, Scientific and
Cultural Organization at its thirty-first session on 2 November 2001.
151 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
realization, (…) of the economic, social that have been issued in the past five
and cultural rights indispensable for his years.
dignity and the free development of his Does that mean that diversity is no longer
personality”.2 to be respected when it comes to AI
In the same vein, the United Nations ethics, which would mean that human
Charter states that the General Assembly dignity, and consequently human rights,
is expected to make recommendations should not be respected either? Or does
aiming at “promoting international that mean that the only way to have those
cooperation in the economic, social, rights respected is through the adoption of
cultural, educational, and health fields”3 the Western approach of AI ethics?
and calls for “international cultural and The following lines are not intended to
educational cooperation”4 in several provide definitive answers to these
fields, even stressing the importance of questions but rather to humbly offer new
ensuring just treatment and protection approaches to the subject to open a real
against abuses “with due respect for the debate on AI ethics that would take into
culture” of the peoples.5 account cultural pluralism.
Yet, and intriguingly when it comes to To do so, we will first try to understand
ethics, it seems that cultural differences what “AI ethics” is really about,
are left aside. Namely, in the case of deconstructing the sacrality of the phrase
artificial intelligence (AI), ethical to bring it back to the secular world of
standards clearly ignore the diversity of rough reality. We will then develop the
cultural perspectives on what is idea that philosophy is needed like never
acceptable and what is not. It is believed before to revitalize reflections on AI
that the only acceptable standpoint on AI ethics and to avoid falling into the trap of
ethics is a very superficial deontological what we call cosm-ethics. We will also
Western view on the subject. This can be elaborate on the importance of cultural
seen in the numerous AI codes of ethics diversity in the ethical assessment of AI-
2
fitted systems to open new paths to
The Universal Declaration of Human Rights, Article 22.
3
4
United Nations Charter, Chapter IV: The General Assembly, Article 13. explore and rearticulate the respect of
United Nations Charter, Chapter IX: International Economic and Social
Cooperation, Article 55.
5
United Nations Charter, Chapter XI: Declaration Regarding Non-Self-
Governing Territories, Article 73.
152 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
As we see it, norms participate in the albeit complex form of power, which has
shaping of our behaviors. In that sense, as its target population.” [13]
they represent a powerful tool to lead In other words, governmentality refers to
people without using force. Instead of all the means used by governing bodies to
coercing people, norms allow make human beings subjects through
governments to influence their citizens, to disciplinary technologies, including
bring them to act in a specific desired way mechanisms of management and
without violence. In so doing, norms are administration. Through these means,
participating in the governance of behaviors can be oriented and controlled
people’s practices through a system made in a very subtle way. Yet, this technique
of several actors that do not have any is widely used by norms entrepreneurs,
individual power or authority to namely “people interested in changing
unilaterally make decisions or to define social norms” [14], or in other words,
and put into practice any policy. As such, actors that are in some way willing to
governance is a way of governing through shape normative environments.
a variety of public or private bodies that
will try to set stable practices and The Foucauldian gouvernementalité is
institutions. therefore used to advance norms by
political bodies. The case of the European
In the late 70s, in a lecture he gave at the Union (EU) is illustrative of that type of
Collège de France, French philosopher governance influencing perceptions about
Michel Foucault introduced the concept what is ethically acceptable and what is
of “gouvernementalité” not and in shaping the normative
(governmentality), referring to the environment of AI. In setting ethical
process through which “the conduct of standards from the development of a so-
individuals or of groups might be called “trustworthy AI,” the EU is
directed” [12], or the art of governing actually influencing perceptions about
bodies through “the ensemble formed by normative expectations in the field of AI.
the institutions, procedures, analyses, It is sending a message saying that AI
reflections, calculations and tactics that must be trustworthy. Through this
allow the exercise of this very specific performative act of language, it shapes
both conduct and the institutional
155 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
156 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
According to Louis Colin, this dominance Words are weapons. Their normative
of private actors over the setting of ethical power is indisputable. Discourse is,
158 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
therefore, an extremely powerful tool as a actors raise the question of the legitimacy
performative statement modifying the of the West and of private actors to
ideal structures of agents in a set of impose a specific perspective on AI
interactions and constructing the social ethics.
realities on which institutions are based The exercise is not neutral. As a
[23]. Language is never unbiased, and performative speech act, AI ethics
even the simplest phrase can contain rhetoric insidiously imposes itself and
intention. Thus, whether they are explicit suggests that the mere evocation of the
or implicit, performative utterances will word ethics, its simple addition as a
influence our understanding of the world qualifying adjective to AI, is enough to
we live in and, consequently, the way we make the latter acceptable if not desirable.
define ourselves and the way we act.
This kind of stance asserting the
In the field of ethics, things are not importance of ethical regulations while
different. As Austin relevantly wrote, “it running for AI dominance illustrates the
has come to be commonly held that many tendency to use ethics as a marketing tool
utterances which look like statements are and to fall into ethics-washing [24], or
either not intended at all, or only intended cosm-ethics [25] [26], namely a narrative
in part, to record or impart aimed at putting some ethical make-up on
straightforward information about the the ugly truth, at disguising a reality, that
facts: for example, ‘ethical propositions’ is often difficult to face, in order to make
are perhaps intended, solely or partly, to “acceptable, if not attractive, the
evince emotion or to prescribe conduct or unacceptable” [27] [28]. Correctly used,
to influence it in special ways.” [23] cosm-ethics can give moral crusaders a
The current struggle over the normative huge power in shaping our future. Thus,
dimension of AI is a perfect added to their increasing economic
demonstration of the role communication power, cosm-ethics reinforce the
plays in shaping perceptions insidiously normative power of multinational
and, eventually, behaviors. Using companies that are “able to create
language as a vector for influence is standards through soft law or more
nothing new. In the present case, the indirectly through lobbying.” [22]
universalist aims established by some
159 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
Easy to handle, ethics is a tool of choice who is really leading and shaping the
for those who embrace cosm-ethics. debate.
Going from ethics to aesthetics, the focus E. THE CASE OF THE EU AS AN
shifted from the acceptability of the uses EXAMPLE
of this new technique to the beauty of the
discourse that has been built around the The most illustrative example of what we
subject. have seen so far is the EU’s attempt to
enter the race without competing directly
At the end of the day, put in experts’ against the two leading giants. The EU,
hands, cosm-ethics, as “a reductive like any other AI race runner, is perfectly
pseudo-ethical discourse with rhetorical aware of the benefits it could generate
value, only aims to legitimize a from this technology. Then, positioning
technology whose numerous potential itself as a normative actor, the EU has
abuses cannot make us forget the found a way to enter the competition
economic, political and diplomatic knowing that it is, nonetheless, lagging
issues” [27]. This can lead to questioning way behind many other competitors.
democracy itself since private companies Conversely to its rivals, the Union is
are now able to act “outside of any putting a lot of effort demonstrating its
democratic process” in order to advance will to see AI being framed by ethical
their own economic interests short of any standards short of legal norms.
societal considerations. [22] Besides, their
influence in political circles can lead to In June 2018, the European Commission
decisions made by public authorities that set a High-Level Expert Group on
would not aim for the greater good of the Artificial Intelligence, which issued in
greatest number but that would benefit a April 2019 a document entitled Ethics
few groups and people, potentially to the guidelines for Trustworthy AI. In this
detriment of the masses. There, the document, seven principles are listed,
relations between the private and the with the purpose of “achieving
public sector should be questioned Trustworthy AI” “in the service of
thoroughly to understand what exactly is humanity and the common good, with the
at stake when it comes to AI ethics and goal of improving human welfare and
freedom.” [29]
160 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
At the same time, one must bear in mind as a creator and a producer of this
that this posture comes with the scope of technology” and that the Union “should
a tenacious competition driven by the leverage its strengths to expand its
promises of huge economic benefits. position” and seize “the opportunity
While asserting the importance of having ahead” with the clear aim “to become a
an ethical framework for AI, the world leader in this area.” [30] The whole
European Union developed White paper is indeed built on the quest
nonconstraining tools such as the General for competitiveness supported by the
Data Protection Regulation (GDPR) and establishment of an ecosystem of trust,
principles that both lend themselves to all making the ethical stance of the EU all
kinds of interpretation and are almost but clear. No doubt, lobbyists that are
impossible to operationalize and, wandering in the EU’s corridors have
therefore, to implement. done their part advancing private interests
This led Professor Thomas Metzinger, a hidden by soothing wording.
member of the commission’s expert III. RENEWING THE REFLECTION
group that has worked on the European ON AI ETHICS
ethics guidelines for artificial intelligence, A. A WESTERN-ORIENTED AI
to write that “the Trustworthy AI story is ETHICS
a marketing narrative invented by A quick look at existing guidelines and
industry, a bedtime story for tomorrow’s other codes demonstrates that too many
customers. (…) Hence the Trustworthy norms are killing norms.
AI narrative is, in reality, about
Hence the question: “Do those ethical
developing future markets and using
guidelines have an actual impact on
ethics debates as elegant public
human decision-making in the field of AI
decorations for a large-scale investment
and machine learning?” [31]. According
strategy.” [24]
to Hagendorff the “short answer is: No,
This assertion is clearly supported by the most often not.” So, it seems that we are
content of the EU White Paper on AI now in a dead-end: a situation where we
issued in 2020, in which it is stressed that are observing a multiplication of ethical
“Europe is well placed to benefit from the guidelines applied to AI leading to
potential of AI, not only as a user but also saturation and deregulation, and where
161 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
aim at harmonious relations within and Even before thinking about our
between societies. This feminist responsibility towards the future, it would
approach, based on the idea that men and also be beneficial to question our deeply
women have a different moral rooted belief that technology is a mere
development, is “grounded in voice and tool in humans’ hands and that,
relationships, in the importance of consequently, we are and will remain in
everyone having a voice, being listened to control of AI. Indeed, “for years, experts
carefully (in their own right and on their have debated the issue of technological
own terms) and heard with respect” [36]. determinism” [2]. Yet, with AI, we have
Such a perspective would allow the embraced the instrumental stance stating
necessary open-mindedness when it that “technologies are ‘tools’ standing
comes to different voices that have ready to serve the purposes of their users”
remained or kept silent. [38]. This stance must be put in the wider
Other philosophers such as Hans Jonas competitive context we have seen earlier
would provide us with tools for a deeper in which cosm-ethics is resorting to a
reflection regarding our responsibility in reassuring vocabulary aiming at building
terms of “the survival and humanity of trust. Nonetheless, asserting that humans
man from the excesses of his own power” are and will remain in control of AI can
[37]. His ethics, applied to technology, and should be challenged using Martin
would definitely enlighten the debate on Heidegger’s critical thought on our
the future of AI and its consequences, blindness towards the very essence of
foreseeable or not, on humankind. As technology to which we are chained [39]
Jonas writes it: “the first duty of an ethics or even Gilbert Simondon and Jacques
of the future: visualizing the long-range Ellul perspectives on our alienation from
effects of technological enterprise” [37]. technology [40] [41]. This substantivist
Consequently, we all have a duty to abide approach of the relationship between
by an imperative of responsibility that he humans and technology claims that
formulates this way: “Act so that the “technology is relentlessly overtaking us”
effects of your action are compatible with shaping our perceptions, and restructuring
the permanence of genuine human life.” our social world through the
[37] establishment of “a new cultural system.”
164 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
[38] As with AI, “technology threatens to ethical particularisms are not only a
slip from human control,” the myth of reality but also humankind’s riches.
human control over technology needs As Belouali et al. put it, “ethics is
more than ever to be disputed. [39] contingent” [2]. This contingency lies on
Through those few examples, we can see the fact that human beings do not share
that ethics is far from being homogenous values. At best, we can postulate the
and one. Ethics is indeed plural, and existence of a universal structure of
facing the rise of AI, this plurality must values [42], but so far, the reality of even
feed the debate. one universal value has not formally been
C. FROM SINGULAR ETHICS TO proved. Here again, words are powerful
PLURAL ETHICS tools, and the overuse of the word
“universal” and its derivatives led us to
Interestingly, AI presents the advantage lose sight of its very meaning to accept it
of putting us in front of ourselves, as a synonym for “generally,” “widely,”
discovering our weaknesses and flaws. or even “predominantly.” Yet, strictly
Doing so, it invites us to question the very speaking, there is no universal value out
meaning of ethics and to explore its there.
complexity. Trying to mimic human
cognitive abilities is helping us to Starting from this assessment, it becomes
understand what we really are, but it also evident that without universally shared
plunges us into the intricacies of the values, there cannot be any universally
human mind. accepted ethical rules.
165 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
Paul Ricœur’s categorization can help to norm [45]. What that means for AI ethics
grasp the complexity and diversity of is that assessing AI through the unique
ethics. Separating moral from anterior lens of Kantian categorical imperatives
ethics and posterior ethics, Ricœur allows regarding what is acceptable and what is
us to operationalize ethics and withdraw it not denies individuals’ right to be
from its subjective relation to moral autonomous in their decisions, to act
obligation. According to the philosopher, according to a context.
“we need such a split, fragmented, This problem is not new, and the risk of
dispersed concept of ethics, anterior moral absolutism stemming from a strict
ethics pointing to the rooting of norms in and superficial understanding of Kantian
life and in desire, posterior ethics aiming deontology is well known and
to insert norms in concrete situations,” in documented. William D. Ross William
what he calls practical wisdom [43]. tried to actualize Kantian deontology
When it comes to AI ethics, the trap in making it adjustable to conditions through
which we have fallen is to consider that a system of deontological pluralism
there is no difference between moral and asserting the existence of both prima facie
its imperative injunctions and applied duties, incumbent to circumstances of the
ethics and the liberty of action it implies. case, and duties proper, based on the
To some extent, we can postulate that situation [46]. In any case, the appraisal
“morality is not negotiable, whereas of AI systems should free itself from rigid
ethics is reasoned” [2]. In other words, deontology to focus on a contextual
posterior ethics set the final aim of life as approach.
the quest for a “good life with and for
others within just institutions” [44], while The risk of moral absolutism carried by
anterior ethics contextualize action in moral crusaders is not mere philosophical
specific situations. Both ethics are ethereal consideration real; it is real and
articulated within the realm of moral already causes diplomatic issues in wider
norms, but Ricœur considers that ethics circles, as illustrated by the harsh
has primacy over morality and, exchange between US and Chinese
consequently, that the aim, namely the representatives in Anchorage on March
quest for a good life, has primacy over the 18th when Dr. Yang, the Chinese
Communist Party foreign affairs chief,
166 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
responding to criticism coming from the and Taiwan. That does not mean that
American side stated: “I don’t think the these countries do not have ethical
overwhelming majority of countries in the guidelines, but at least it tends to show
world would recognize that the universal that ethics might not be in their strategic
values advocated by the United States or priorities list.
that the opinion of the United States could Eventually, it is clear that our narrow
represent international public opinion, perspective on ethics as a regulation tool
and those countries would not recognize is challenged by philosophy and by the
that the rules made by a small number of reality of our world. Viewing ethics as the
people would serve as the basis for the mere application of the deontological
international order” [47]. principle is not only ineffective, but it is
Thus, ethics cannot be universalized, also dangerous. Norms are power, for
whether in the field of AI or in any other they enhance both control over valuable
field, for neither values nor their resources, of which AI is part of, and over
hierarchy are universal. The ETH Zurich the influence some actors can have on
study showed that “no single ethical others’ “affect, cognition, and behavior”
principle appeared to be common to the [49]. Norms, thus, participate in the
entire corpus of documents, although ethical tyranny of the atrophied Western
there is an emerging convergence” [16]. deontology that “alter others’ states by
In the same vein, one of the key findings providing or withholding resources or
of a study conducted by the Canadian administering punishments.” [49]
Institute for Advanced Research (CIFAR) There are still many other paths to be
is that each of the 18 national strategies explored in AI ethics, and rigid
for AI exploited was unique and that they deontology is far from being the only
do not share the same strategic priorities relevant approach to the topic. In that
[48]. This is to be related to the point we framework, there is an existential need for
made about the race for AI dominance opening the debate to non-Western
which shapes national strategies and their philosophies and wisdom. In a globalized
priorities. In some cases, ethics is not world, where diversity is praised as
even mentioned in strategic documents richness, it is paradoxical to accept that
like for Germany, Japan, South Korea, the unique Western perspective could
167 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
define for the rest of the world what is instead of requesting them to accept our
acceptable and what is not regarding AI. own programming.
D. AN OPENING TO NON- As Kai-Fu Lee wrote it, “it is becoming
WESTERN PERSPECTIVES harder to force people of all countries and
In 1980, in a seminal work recognized as cultures into a cookie-cutter mold that
a reference in cross-cultural comparative was often built in America for
research, Dutch anthropologist Geert Americans” [51]. The exchange between
Hofstede studied how organizational China and the US in Anchorage is
cultures in the workplace can differ from illustrative here.
a country to another depending on In the world of technologies, it is
specific values, customs, and behaviors. noticeable that the Institute of Electrical
In his book, Hofstede argues that “each and Electronics Engineers (IEEE) is
person carries a certain amount of mental questioning the “monopoly on ethics by
programming” that is “partly unique, Western ethical traditions” and stressing
partly shared with others” [50]. Thus, that the full benefit of autonomous and
cultures are the product of “the collective intelligent technical systems “will be
programming of the mind that attained only if they are aligned with
distinguishes the members of one group society’s defined values and ethical
or category of people from another.”[50] principles.” [32] According to the
In other words, cultures are based on Institute, there is an “urgent need to
different programmings that lie on broaden traditional ethics in its
different values and lead to different contemporary form of ‘responsible
behaviors. What is true for organizations innovation’ (RI) beyond the scope of
is also true for bigger communities or ‘Western’ ethical foundations.” [32] In its
societies. document, the IEEE even offers some
The point here is to consider that when it openings on non-Western ethical
comes to AI ethics, it is important to traditions such as Classical Buddhism,
understand the mental programming African Ubuntu, and Japanese Shinto.
through which cultures have been built This interest in non-Western ethical
traditions demonstrates that what Mahdi
168 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
Elmandjra calls the “cultural arrogance” the Moral Machine project set by the MIT
[52] has limits and that, in the field of and gathering data “on the way citizens
technology, we might have reached them. would want AVs [autonomous vehicles]
Other cultures challenging our restricted to solve moral dilemmas in the context of
and restrictive perspectives on AI ethics unavoidable accidents” provided by
will certainly open doors towards new 70,000 participants from 42 countries
options. It would mostly avoid tensions [53]. Following the experiment, a paper
regarding norms between the West and published in 2020 identifies several
rising AI powers such as China. It seems cultural clusters and demonstrates that
unlikely that, if China reaches its goal to “every culture has rules about what is
become the leader in AI by 2030, the right or wrong, but they often disagree on
Empire will keep on accepting a the particulars of moral decisions. Moral
normative diktat from the European universals are difficult to find, as they
Union and the US. China is actually an often reveal some degree of cultural
interesting illustration of our inability to variation upon closer inspection.” [54]. If
understand cultural differences in the it is clear, as the authors themselves admit
field of AI. This inability, rooted in our it, that the study is questionable on
ignorance about the Confucian grounds of several points and that its results must not
the Chinese mentality, is furthermore be used for normative purposes [55], it
reinforced by our rejection of its political nonetheless highlights trends that tend to
system, which has eventually taken show that culture is far from being neutral
precedence over the respect of its cultural when it comes to ethical appraisal, and
rights engraved in the Universal that, at least, deserve deeper scrutiny. The
Declaration on Cultural Diversity and the experiment reveals, for instance, that
UN Charter. From a Chinese perspective, there are “systematic differences between
this is seen as an infringement of the individualistic cultures and collectivistic
Chinese people’s dignity. cultures” [53].
In the field of AI, it is clear that cultural This split between individualistic and
differences can lead to very different collectivistic cultures clearly impacts the
viewpoints on what is acceptable and way different cultures will address
what is not. This has been illustrated by problems such as autonomous cars risk-
169 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
related for pedestrians, data collection and Japanese culture, is way more natural
processing, or the use of lethal than it is in the Western world. [32].
autonomous systems. Since “the “West”, In Africa, “Ubuntu philosophy tells us
however it may be defined, is an that Humans are interconnected” [58],
individualistic society, arguably more so that conversely to the West, individuals
than much of the rest of the world”, its are not an end but cogs in a larger
perspective on video surveillance and ecosystem. The quest for happiness is
facial recognition will obviously differ then not related to the satisfaction of
from other collectivistic cultures. [32] individual needs but to the building of
As Soraj Hongladarom puts it, harmonious relations within the
“Buddhism has much to offer anyone community. Thus, when it comes to
thinking about the ethical use of privacy Ubuntu emphasizes transparency
technology, including those interested in to group members rather than individual
AI” [56]. It could help bridge the ownership of data [59], stressing the gap
individualistic perspective with the between the rational personhood of
collectivist one offering “ethical Western culture and the relational one
statements formulated in a relational way, advanced in the Ubuntu tradition [60].
instead of an absolutist way” [32]. The Muslim world should also be part of
Applied to privacy, Buddhist ethics sheds the debate. As stressed by Dr. Junaid
new light on questioning the relevance of Qadir and Amana Raquib, “there’s a lack
the self and, consequently, of the of representation in AI for the two billion
reification of individuals as owners of people who profess these beliefs.” [61]
private data and showing that, given the The Muslim perspective on AI used for
absence of self in the Buddhist tradition, medical purposes would, for instance,
privacy is nonsensical, at least in the way revitalize the debate on life prolongation
the notion is understood in the West [57]. by technological means. It would also
Other cultural standpoints such as Shinto pave the way for a new discussion on the
would also challenge our relationship use of driverless cars. [61]
with robots, for the relationship between These are a few examples of other
AI systems and human beings, in the perspectives that need to be integrated to
170 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
At the end of the day, what is clear is that We have thus tried, roughly, to explain
“one of the big mistakes done by Western what is hidden behind the phrase “AI
values promoters (…), is to qualify as ethics,” shedding light on the complexity
universal all their values and conceptions of the notion, on the bigger picture it must
be placed in, and on the importance of
of democracy, human rights, justice,
wording as a normative tool shaping
equity, rationality, scientific method,
perceptions and influencing behaviors.
technology, aesthetic.” [33]
From our work, it appears that AI ethics
In doing so, the West deprives itself of is a purely Western product that defines
the richness of cultural diversity and, at for the whole world what is acceptable
the same time, this tends to deny peoples and what is not, what is to be done and
their fundamental cultural rights what is not. Furthermore, we have seen
indispensable for their dignity. that only some actors in the West are
deciding upon the norms that must frame
IV. CONCLUSION the development and use of AI. This
While we were writing these lines, tyranny of Western norms is slowly and
UNESCO held a roundtable on the ethics insidiously pervading minds all around
of artificial intelligence entitled “Shaping the world, denying the right for cultures
the Future of AI through Cultural to express their peculiarities and have a
Diversity” showing, if needed, that the say in the debate on a technology that is
subject is of utmost importance and slowly invading people’s everyday lives
topical. worldwide.
This article has not been written as an end There is a strong existential need for
in itself. It has been envisioned as a revitalizing the debate over AI ethics.
starting point for further discussions, a Discussions must integrate new
means to a greater end which is the perspectives that will both challenge
needed debate on AI ethics and its mainstream approaches, namely atrophied
relationship to cultures. Every point, deontology and the instrumental view on
171 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
technology and help us to make better [2] S. Belouali, A. Belouali, M. Saber, K. Jaafar
choices through the confrontation of and M. Belkasmi, "Ethics of AI or Ethical AI,
Topical Point of View," in International
opinions. conference on smart Information &
communication Technologies, 2019.
The technological future of humanity
depends on the respect and the integration [3] V. Goossaert, Livres de morale révélés par les
of cultural particularisms leading to them dieux, Les Belles Lettres, 2013.
to balance each other out to prevent us [4] F. D. Saussure, Course in General Linguistics,
from venturing into destructive horizons. Philosophical Library, 1959.
The cultural uniformity carried by the
[5] L. Wittgenstein, A lecture on ethics, Cambridge
crusaders of universal morality can only University, 1929.
lead to tyranny, a moral unipolarity that
[6] L. Wittgebstein, Tractatus Logico-
will crush diversity and create and
Philosophicus, Trubner & Co, 1922.
accentuate tensions between peoples.
[7] T. Züger, «The relevance of culture in the age of
Let us give the final word to the great AI,» 02 07 2019. [En ligne]. Available:
Moroccan sociologist, Mahdi Elmanjdra: https://www.hiig.de/en/the-relevance-of-culture-
in-the-age-of-ai/. [Accès le 10 01 2021].
“The blossoming of this cultural
diversity cannot, however, occur [8] K. Kesserwan, How can indigenous knowledge
shape our view of AI, Policy Options Politiques,
without freedom, the freedom to think 2018.
in a different way, the freedom to live
and act in accordance with the specific [9] P. Berger and T. Luckmann, The social
construction of reality: A treatise in the
and inalienable values of each sociology of knowledge, Anchor, Penguin
community” [33]. Books, 1966.
172 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
[15] H. Becker, Outsiders: Studies in the Sociology [24] T. 2. .. ,. 8. Metzinger, "Ethics washing made in
of Deviance, New York: The Free Press, 1963. Europe," 08 04 2019. [Online]. Available:
https://www.tagesspiegel.de/politik/eu-
[16] A. Jobin, M. Ienca and E. Vayena, "The global guidelines-ethics-washing-made-in-
landscape of AI ethics guidelines," Nature europe/24195496.html. [Accessed 28 03 2021].
Machine Intelligence, vol. 1, no. 9, p. 389–399,
Septembre 2019. [25] E. R. Goffi, «De l’éthique à la cosm-éthique (1)
: ce que l’éthique n’est pas,» Institut Sapiens,
[17] A. S. Rao and G. Verweij, "Sizing the prize. Paris, 2019.
What's the real value of AI for your business
and how can you capitalise?," PWC, 2017. [26] E. R. Goffi, «De l’éthique à la cosm-éthique (2)
: ce qu’est l’éthique,» Institut Sapiens, Paris,
[18] Chines State Council, "Notice of the State 2019.
Council Issuing the New Generation of
Artificial Intelligence Development Plan," 2017. [27] E. R. Goffi, «ETHIQUE ET CONFIANCE : LA
TYRANNIE DES MOTS, LE DESPOTISME
[19] D. Greene, A. Hoffmann and L. Stark, "Better, DU DISCOURS,» 2020. [En ligne]. Available:
nicer, clearer, fairer: A critical assessment of the http://francefintech.org/ethique-et-confiance-la-
movement for ethical artificial intelligence and tyrannie-des-mots-le-despotisme-du-discours/.
machine learning," in Proceedings of the 52nd [Accès le 28 03 2021].
Hawaii international conference on system
sciences, 2019. [28] E. R. Goffi, "The importance of cultural
diversity in AI ethics," Beyond The Horizon,
[20] L. Colin, «De quoi l'éthique de l'IA est-elle de 2021.
nom?,» chez Journée Internationales de
l'Intelligence Artificielle, Oujda, 2021. [29] European Commission, Directorate-General for
Communications Networks, Content and
[21] J. A. N. H. H. N. A. S. M. Fjeld, "Principled Technology, «Ethics guidelines for trustworthy
artificial intelligence: Mapping consensus in AI,» 2019.
ethical and rights-based approaches to principles
for AI," Berkman Klein Center Research [30] European Commission, «White Paper on
Publication, vol. 2020, no. 1, 2020. Artificial Intelligence: A European Approach to
Excellence and Trust,» Brussels, 2020.
[22] L. Colin, «Intérêt général et entreprises
multinationales,» 18 11 2020. [En ligne].
173 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco
[37] H. Jonas, The imperative of responsibility, [48] T. Dutton, B. Barron and G. Boskovic,
University of Chicago press, 1985. "Building an AI world: Report on national and
regional AI strategies," CIFAR, Toronto, 2018.
[38] A. Feenberg, Transforming Technology: A
Critical Theory Revisited, Oxford University [49] D. Keltner, D. Gruenfeld and C. Anderson,
Press, 2002. "Power, approach, and inhibition,"
Psychological review, vol. 110, no. 3, p. 265,
[39] M. Heidegger, The question concerning 2003.
technology, 1977.
[50] G. Hofstede, Culture's consequences:
[40] G. Simondon, Du monde d'existence des objets Comparing values, behaviors, institutions and
techniques, Aubier, 2012. organizations across nations, Sage publications,
2001.
[41] J. Ellul, J. Wilkinson and R. Merton, The
technological society, New York: Vintage [51] K. Lee, AI superpowers: China, Silicon Valley,
books, 1964. and the new world order, Houghton Mifflin
Harcourt, 2018.
[42] M. Wach et B. Hammer, La structure des
valeurs est-elle universelle? Genèse et [52] J. W. Botkin, M. Malitza et M. Elmandjra, On
174 | P a g e
Arribat – International Journal of Human Rights
Volume: 1, Issue: 2, September 30th, 2021 | Published by CNDH Morocco