Content uploaded by Viktor Dörfler
Author content
All content in this area was uploaded by Viktor Dörfler on Jan 09, 2024
Content may be subject to copyright.
Dubito Ergo Sum: Exploring AI Ethics
Viktor Dörfler
University of Strathclyde Business School
Glasgow, United Kingdom
viktor.dorfler@strath.ac.uk
Giles Cuthbert
Chartered Banker Institute
Edinburgh, United Kingdom
giles.cuthbert@charteredbanker.com
Abstract
We paraphrase Descartes’ famous dictum in the
area of AI ethics where the “I doubt and therefore I am”
is suggested as a necessary aspect of morality.
Therefore AI, which cannot doubt itself, cannot possess
moral agency. Of course, this is not the end of the story.
We explore various aspects of the human mind that
substantially differ from AI, which includes the sensory
grounding of our knowing, the act of understanding, and
the significance of being able to doubt ourselves. The
foundation of our argument is the discipline of ethics,
one of the oldest and largest knowledge projects of
human history, yet, we seem only to be beginning to get
a grasp of it. After a couple of thousand years of
studying the ethics of humans, we (humans) arrived at a
point where moral psychology suggests that our moral
decisions are intuitive, and all the models from ethics
become relevant only when we explain ourselves. This
recognition has a major impact on what and how we can
do regarding AI ethics. We do not offer a solution, we
explore some ideas and leave the problem open, but we
hope somewhat better understood than before our study.
Keywords: AI ethics, responsible AI, understanding,
sensory knowledge, indwelling
1. Introduction
In this conceptual paper we argue for the ethical
approach in AI that suggests leaving most of the moral
issues in the hands of humans. This is not to say that we
should not try to ‘put’ a moral perspective into AI, but
that we also need to ‘put’ in the limitations. The natural
first step is that we need to understand the limitations,
but the far trickier next one is how to get AI to identify
its limitations, and request human assistance. We do not
intend here to get into the technical details of what can
and needs to be done in AI, we remain at the level of
philosophizing AI and the human mind in the context of
ethics, thus problematizing AI ethics. We do this from
a distinct phenomenological position, within a moderate
interpretivist paradigm (Dörfler, 2023b).
In this paper we do not provide a generic review of
the AI literature, we only explain the basic concepts that
we use in this paper here in the introduction. Thus, for
the purpose of this paper we use one of the oldest
definitions of AI, back from the Dortmund days,
according to which AI is loosely defined as machines
that can accomplish tasks that humans would
accomplish through thinking (e.g. Dörfler, 2020).
This definition does not say anything about AI
accomplishing such tasks in a way that resembles
human thinking; we do not see anything in this
definition that implies that AI would think in the human
sense of the word. Importantly, AI as a field is not
simply a study of the machines, it is as much the study
of the human mind (for a more detailed description see
e.g. Dörfler, 2022; Dörfler, 2023a). Specifically in the
area of decision-making we find Davenport’s (2018, p.
44) description of AI as “analytics on steroids”
particularly expressive and, consequently, AI does not
make decisions but it can make our (human) decisions
better informed.
Decision-making is an important aspect of using AI
when it comes to ethics, and all (at least the vast majority
of) our decisions have moral components. In this paper
we do not engage with particular application areas of AI,
such as medical diagnosis (Davenport & Glaser, 2022;
Davenport & Glover, 2018; Göndöcs & Dörfler, 2023),
we locate our interest loosely in organizations (Csaszar
& Steinberger, 2022; Davenport & Euchner, 2023;
Davenport & Miller, 2022; Glikson & Woolley, 2020;
Grodal et al., 2023; von Krogh, 2018; Leavitt et al.,
2021; Lindebaum & Ashraf, 2021), in which concept we
include business organizations, government institutions,
as well as organizations, such as hospitals and
universities, regardless of whether they are for profit or
not. We are conscious of the organizational learning
aspects and implications of AI (Balasubramanian et al.,
2020; Davenport & Ammanath, 2020; Davenport &
Mittal, 2022, 2023; Göndöcs & Dörfler, 2022; Oliver et
al., 2017; Pachidi et al., 2021; Raisch & Krakowski,
2021; Tschang & Almirall, 2021); although we cannot
tackle these at a great depth here.
Proceedings of the 57th Hawaii International Conference on System Sciences | 2024
Page 5587
URI: https://hdl.handle.net/10125/107056
978-0-9981331-7-1
(CC BY-NC-ND 4.0)
In order to develop our argument eloquently
captured paraphrasing Descartes, in what follows, we
begin with a brief but systematic overview of the most
common approaches and models in the domain of ethics.
This is followed by a review of the AI ethics literature.
Then we outline our philosophical position and
methodological considerations, before getting to the
points we want to make. Each of the next three sections
provides a component of our conceptual analysis. First,
we explore the sensory grounding of knowledge,
showing how such sensory grounding, in a sense we
attribute it to humans, is not possible in AI. Second, we
suggest that AI lacks understanding, and we illustrate
this with recent events in the AI landscape. Third, based
on the literature, we argue that doubting oneself is an
essential ingredient of morality, and we show that this
one requirement also incorporates the previous two. In
our final commentary we discuss what can be done, and
offer three points, to make the best use of AI in morally
acceptable ways and indicate areas of further research.
2. The Vast Landscape of Ethics
In this paper the term ethics is used to designate a
branch of philosophy, the discipline that studies
morality. In turn, morality refers to dealing with the
issues of good and evil, right and wrong, responsibility,
and such. In this sense, ethics is one of the oldest topics
studied by humankind. In the Western tradition we see
philosophy starting with discussing metaphysics
followed by epistemology, which is closely followed by
ethics, which Socrates has brought centerstage. This
means that we have at least some two and half millennia
of literature to cover, therefore our review will not be
comprehensive, but our review is systematic in the sense
that the main philosophical models are organized into
categories (see Figure 1).
Figure 1. Overview of Historical Schools of Ethics
The normative schools attempt to prescribe what to
be or what to do or not do; the behavioral school aims at
describing what people actually do. The three dominant
normative schools of ethics, are virtue ethics, rules-
based ethics (or deontology), and consequentialism.
Other important models in normative ethics include
pragmatist, intuitionist, contractualist, and feminist
ethics. While these schools are all centered on
individuals, there are also less known social variants,
proposed by a small number of philosophers.
While it was Socrates who initially turned the
philosophers’ attention to considerations of morality,
particularly to counter the sophist approach to teach
anyone how to win debates using tools of rhetoric, it was
Aristotle (cca 330 BCE) who composed ethics into the
first systematic model called virtue ethics. Aretaic
ethics, of which virtue ethics is the dominant example,
focuses on features or characteristics that are desirable
in individuals if they are to be considered to meet high
moral standards; in virtue ethics these characteristics are
the virtues. Importantly, virtue ethics does not aim at
providing answers or even tools to answering moral
dilemmas, they describe the moral person. The main
issues in virtue ethics is identifying what the relevant
virtues are and how much of each is good, recognizing
that virtues in excess may become vices (Hursthouse,
1999). Naturally, actions of people are considered
morally correct if they embody the virtues required by
the particular school; i.e. a moral dilemma can be
answered by considering what a virtuous person would
do in such situation. Therefore, there are no forbidden
activities, so there is no tenet, for instance, not to kill,
the dilemma is whether a virtuous person would kill.
After flourishing in Antiquity, virtue ethics was
superseded by other schools, but it did have a revival in
the 20-21 centuries, the leading figure being Alasdair
MacIntyre (1998) and Elizabeth Anscombe (1958).
MacIntyre (2007) also emphasized that we may need
new models of virtue ethics for the modern world.
Throughout history, and particularly during modernity,
with its emphasis on scientific thinking and
reductionism, attempts were made to derive all virtues
from a single one, so can we say that all virtues are just
consequences of e.g. courage or patience or, at least,
construct a list of priority, i.e. virtues should be ordered
by importance (or other organizing principle). Some
recent philosophers, including Julia Annas (2011) and
Anscombe (1958) argue that the virtues are all
interconnected forming a complex system although the
idea can be traced back to Aristotle (Gottlieb, 1984).
They also discuss limitations of how actions of virtuous
people may occasionally be wrong, which means that
being virtuous is not a guarantee of right action. There
are many other smaller schools of virtue ethics that we
do not cover here. All schools of virtue ethics agree on
two principles: they all attribute agency to humans, so
that they can exercise their free will and choose to do
the right thing, and they all consider humans rational
and thus exercising rationality is part of being virtuous.
Aretaic
What to be
Consequentialism
Outcome criteria
Deontology
Rules for action
Virtue ethics
Virtues & values
Deontic
What to do
Social
What we owe
each other
Normative
Prescription
Behavioral
Description
Moral
Psychology
What we really do
Shift to consider actual behavior
Schools of Ethics
Ways to study good and evil
Page 5588
While areatic schools of ethics focus on what a
virtuous person is like, deontic schools of ethics focus
on what constitutes a morally right action. Deontology
and consequentialism are both deontic (see Figure 1).
Deontology, in essence, offers sets of rules that we
must follow to adhere to a high moral standard.
Although there were a few earlier versions, the first
comprehensive treatment of deontology comes from
Immanuel Kant (1785, 1797). The central concept of
Kant’s deontology is the categorical imperative,
according to which we should do what we would be
happy to become a universal law. Another very
important concept of Kant’s deontology is the good will,
meaning that if one has the right intentions, but things
turn out badly due to things beyond one’s control, that
is acceptable. This served as basis for the development
of a minority branch of deontology, which emphasizes
intentions in contrast to consequences. While in these
schools consequences do not matter, in most schools
they do, only not as much as obligations and rights. A
deontological model can incorporate one or both of two
perspectives: agency and patiency. Agency links to
obligations, i.e. the rules are based on what one is
supposed to do. Patiency links to rights, i.e. the rules
are based on how people can expect to be treated.
While for Kant rules are universally and infinitely
valid, many approaches to deontology are more
situational; sometimes it is OK to deviate from the rule
for the sake of a better outcome. For instance, it is
wrong to lie, but sometimes it may be the kind thing to
do, and in some extreme situations (e.g. a monster
asking you where your children are because it wants to
eat them) it may be the only right thing to do. Of course,
exceptions to rules can be formulated as rules, but only
so far as we are able to foresee idiosyncratic situations.
At first sight, deontology seems an excellent candidate
for being considered in AI, as it is rule-based, and rules
are easy to program. However, exceptions tend to be
problematic in programming as much as in philosophy.
Consequence ethics, like deontology, is concerned
with the right action, but not in terms of what we ought
to do but in terms of the consequences of our actions.
Consequences are typically expressed in the form of
some utility or happiness at an individual or social level.
Variants of consequentialism include hedonism,
egoism, act consequentialism, and various forms of
utilitarianism (e.g. Mill, 1861; Williams, 1993). There
are two initial issues surrounding consequentialism: (1)
Normally, we cannot know the consequences of our
actions for sure at the time of taking the action. (2)
Desirable consequences are often formulated as the best
outcome for the greatest number of people, and thus the
rights of minorities may be overlooked. Nevertheless,
consequentialism seems to be, at least on a cursory look,
to be well aligned with the idea of machine learning
(ML) in artificial neural networks (ANN). Once we take
further in-depth considerations, however, we find that
there are other aspects of moral decisions that matter
besides the consequences, even if we could be sure of
those consequences; examples include the trolley
problem and similar dilemmas (Foot, 1978; Williams,
1985).
There were a few individual approaches to ethics by
leading thinkers of Enlightenment. These thinkers,
including Georg Wilhelm Friedrich Hegel, Baruch
Spinosa, David Hume, and Adam Smith emphasized the
interconnectedness of people, leading to the view that
other people are key to one’s moral decisions. These
approaches can be seen as a move from a fully
individualist treatment of ethics towards a social
perspective; for example, in a social variant of
consequentialism we can consider what is good for
society rather than the individual. An implication of
these social approaches is the previously mentioned role
of patiency in addition to agency in ethical
considerations. A more contemporary example of a
social view of ethics is Emmanuel Levinas (1961,
1991).
Besides these main normative schools, there are
lesser-known ones, many of which can be, in some
ways, considered “more human” than the previously
introduced major schools. Pragmatist ethics, although
not very prominently featured in the overall pragmatist
philosophy, does not believe in the possibility of a single
model that governs all moral decisions; ethics models
can be useful, but different ones may be appropriate for
different occasions (see e.g. James, 1899). According
to intuitionist ethics we can know moral principles
intuitively, as they are self-evident. These principles,
typically duties (which makes intuitionism a form of
deontology), are identified individually and therefore
intuitionist ethics may be applicable in a greater number
of situations than other approaches. For contractualist
ethics, which is a form of deontology, justice makes an
action right – although this is typically formulated in
negative terms, i.e. avoiding wrongdoing and injustice.
Finally, feminist ethics questions many assumptions of
normative ethics models, which have all been
conceptualized from a dominantly male, often
patriarchal, perspective. This leads to rejecting the
possibility of any absolute model in ethics (Gilligan,
2014), which means that both pragmatist and feminist
ethics become pluralist and contextual.
We do not discuss the variants of ethical approaches
in the Frankfurt School, as they are very fragmented and
highly blended with political philosophy. What they
agree about is similar to the starting point of feminist
ethics, i.e. they all suggest that Western ethics is imbued
with the exploitative values of the capitalist hegemony,
making social justice impossible.
Page 5589
2.1. Moral Compasses
Moral psychology is about what people actually do
rather than what they are supposed to do, therefore it can
also be labelled as descriptive ethics in contrast with the
normative schools (see Figure 1). The term moral
psychology can be traced back to Anscombe (1958),
who observed that the thinking in ethics should take into
consideration what we learned studying psychology.
In one of the cornerstone works of moral
psychology James Rest (1986) distinguishes four stages
of moral decision making: moral sensitivity, moral
judgment, moral motivation, and moral courage. A
thorough treatment of moral psychology would require
covering a significant amount of psychology literature,
both conceptual and experimental, and making
connection to the previously outlined normative
approaches to ethics. Therefore, here we focus solely
on one particularly important issue. Using Rest’s
stages, psychologists of ethics have established that
people make moral judgments more or less exclusively
using their intuition, and they only refer to the ethical
models that they are familiar with when they need to
justify their moral judgments to themselves or to others
(cf Haidt, 2001). This is a very strong claim and moral
psychology is largely in agreement about this point
(Sonenshein, 2007). This is what we tried to capture
with the notion of the moral compass.
However, we also must note that this does not mean
that all ethicists, let alone scholars in the domain of AI
ethics, subscribe to the dominant role of intuition in
moral judgments; many emphasize deliberation with or
without the use of normative models. There are also rare
studies that try to synthesize the normative and
descriptive approaches to ethics (Treviño, 1986). In our
view, humans often arrive at moral judgments
intuitively, just as in the case of any decision making,
but this can also happen through sequential reasoning
(Dörfler & Stierand, 2017). However, if consulting the
models we are familiar with does not help to arrive at a
moral judgment, we may revert to the use of intuiting as
time may be pressing. A case can also be made that the
deliberation before action is also only employed to
justify the intuitive judgment already made. The debate
is still ongoing, and we will not resolve it in this paper;
for us it is important that intuitive judgments exist and
both experimental and observational studies find
significant use of intuition, particularly at a high level of
mastery (Chase & Simon, 1973a; e.g. Chase & Simon,
1973b; Dörfler et al., 2009; Dreyfus, 2004; Dreyfus &
Dreyfus, 1986; Gobet & Simon, 1996a, 1996b, 2000;
e.g. Kreisler & Dreyfus, 2005). It is reasonable then to
assume that moral judgments, like judgments more
generally can be intuitive.
3. Ethics and AI
Ethical considerations in computer science are not
new, and they get amplified in the world of AI. Norbert
Wiener’s (1960, p. 1358) formulation is still valid,
perhaps more than ever:
“If we use, to achieve our purposes, a mechanical
agency with whose operation we cannot efficiently
interfere once we have started it, because the action
is so fast and irrevocable that we have not the data
to intervene before the action is complete, then we
had better be quite sure that the purpose put into the
machine is the purpose which we really desire and
not merely a colorful imitation of it.”
The literature on ethics in the scope of computers
and wider digitalization, sometimes also referred to as
digital ethics, largely applies normative models of ethics
to the scope of the digital world (Anderson, 2011;
Anderson & Anderson, 2011; Brey, 2000; Flanagan et
al., 2008; Friedman et al., 2013; Moor, 1985; Vallor,
2016; van Wynsberghe, 2013). For example, Shannon
Vallor (2016) seeks to adapt Aristotelian virtue ethics
for the digital future.
The most important issue of AI ethics, as a
scholarly discipline, is that it is almost completely
conceptual. There are great explorations of applying a
variety of normative ethics models in different
digital/computerized/AI environments and theorizing or
problematizing what the consequences would be –
typically not leading to happy conclusions with the
logical outcome that we may need a new normative
model (Gunkel, 2017). Important problem areas in AI
ethics, with reference to normative ethics, are agency,
the roles emotions may play, levels of relativism,
rationality, more specifically that there are different
kinds of rationalities, the use of intuiting, as well as the
relationship between a moral decision and action. For
instance, rules-based ethics seems to be particularly
suitable for computers, but whose rules to accept?
At the same time, AI vendors struggle with the
ethical aspects of their products, and they keep looking
at AI scholars and philosophers for help that they fail to
provide. They look at the potential users of their
products before designing a new product as well as after,
and the opinions that they receive contradict each other
and cannot be programmed. This problem, observed in
the reality of AI vendors, is our starting point.
The most popular form of AI today, the deep neural
networks (DNN) capable of deep learning (DL) cannot
help. Much of the AI success today is ascribed to this
form of AI, however, those successes must be looked at
in context: the type of problems DL was applied to. In
principle, DNN is simply an ANN with more than one
Page 5590
hidden layer, and DL is a really efficient form of ML in
a DNN, but the principles are not substantially different:
ML needs a large number of learning examples and then
it replicates the statistical frequencies of the outcomes
with reference to the input variables (LeCun et al., 2015;
Marcus, 2018; Schmidhuber, 2015). AlphaGo (Silver et
al., 2016) needed some 300 billion games to get trained
and deliver the extraordinary performance of beating
Lee Sedol (Hassabis, 2017). So, if we even had a
database of moral decisions, what number of learning
examples would be needed for DL to produce the
statistical frequencies? What variables would need to
be considered? How many types of moral decisions are
there? The list of questions could continue, each and
every one of them would be sufficient to conclude that
this is not the way to successfully deal with AI ethics.
Furthermore, there is evidence that if the training data is
biased, the ANN will amplify these biases.
Essentially, we need to understand that the problem
of AI ethics is not an implementation problem. It is not
about having the right conceptual construct that we need
to operationalize, we are struggling with fundamental
problems of ethics. Therefore, we need to step back and
look into making progress in the field of ethics in which
both humans and AI are present. So, what can be done?
In order to figure this out, we look into three aspects of
how AI is different from the human mind.
4. Methodological Considerations
We take a phenomenological approach in this
conceptual study, framed within the broadly considered
paradigm of Critical Interpretivism (Dörfler, 2023b).
this means that our approach is moderate subjectivist,
and we practice bracketing through transpersonal
reflexivity to arrive at insights (Dörfler & Stierand,
2021). We do not adopt a theoretical lens, as any lens
limits what can be seen, instead we adopt the approach
known as phenomenon-driven theorizing, which allows
us to approach the phenomenon at hand with an open
mind and allow theorizing to take us in various
directions (Fisher et al., 2021; Langley, 2021; Ployhart
& Bartunek, 2019). The particular type of theorizing we
undertake is called problematizing, as the purpose of it
is not to provide a solution but to arrive at an improved
understanding of what the problem is. To problematize
AI ethics, we make use of everyday well-known
phenomena, as it is often done in Gestalt psychology
(Köhler, 1959; Rock & Palmer, 1990), and attempt to
explain these employing abductive reasoning (Sætre &
Van de Ven, 2022).
5. Indwelling
Sensing, i.e. sensory input is indispensable aspect
of all knowing and is necessarily employed on par with
intellect (Bas et al., 2022). Michael Polányi (1966b, p.
15), the renowned philosopher of knowledge, argues
that the body is the ultimate instrument of all external
knowledge. Antonio Strati (2007) wonders why the
important role of the body is neglected and often ignored
although it is the body that enables both intellectual
reasoning and sensory-based knowledge. Sensing is not
a unitary construct. Based on Burton (2009, p. 37)
Dörfler and Bas (2020) consider, besides perception
based on the five primary senses, also visceral
sensations (e.g. hunger), affective sensations (e.g. love),
as well as mental sensations (e.g. pride). With this
expanded view of sensing, we can easily conclude that
everything we know we come to know through sensing
(Bas et al., 2022; de Rond et al., 2019; Strati, 2007).
Furthermore, based on Polányi, we introduce the notion
of indwelling, through the use of which the idea of
sensing can be extended to abstract domains, such as
mathematics or astrophysics, or microbiology: these are
all abstract in the sense that we cannot get in touch with
the subject of inquiry through our body, but the
phenomenon is essentially the same. We must
emphasize that we do not argue for the empirical over
the rational, we suggest considering indwelling in
addition to, rather than instead of, reasoning.
Why is this so important? Of course, people sense
and there are various mechanical, electronic, etc.
sensors that we can connect to computers. Surely a
camera with the right set of calculations is more reliable
than a human eye… However, two harsh critiques of
AI, Hubert Dreyfus (1992) and John Searle (1994) both
consider the computers’ lack of sensory capacity in
producing knowledge as one of the main reasons that
computers cannot think and that the “strong AI”
paradigm is impossible. A full treatment of this issue
would entail exploring the issue of primary and
secondary qualities, originally introduced by René
Descartes (1637) and then elaborated by John Locke
(1690) and later George Berkeley (1878), and deriving
the notion of “felt sense” from these (Dörfler, 2023b).
Therefore, we do not intend to engage in a generic
debate about the possibilities of human-level (or nearly
human-level) AI; we simply acknowledge the
significance of sensory grounding in ethics in the light
of the four phases of moral decision-making identified
in moral psychology. As before, we emphasize that we
are not arguing that sensing should replace reasoning
but that sensing and reasoning are both essential.
Page 5591
6. Understanding
There is no complete agreement in cognitive
psychology or the philosophy of mind regarding the
precise definition of the concept of understanding.
Russell Ackoff (1989) locates it between knowledge and
wisdom. We know that tacit knowledge plays a crucial
role in understanding, as Polányi (1966a, p. 7) suggests:
“While tacit knowledge can be possessed by itself,
explicit knowledge must rely on being tacitly
understood and applied. Hence all knowledge is
either tacit or rooted in tacit knowledge. A wholly
explicit knowledge is unthinkable.”
The significance of understanding in ethics is
perhaps obvious: if we are to make moral decisions, we
need to be able to understand that our actions have
consequences (consequentialism), even if we do not
exactly know what the consequences are, as need to
understand the rules that we are meant to follow
(deontology), we need to understand the implications of
specific virtues on our actions (virtue ethics).
Debates on whether AI has or can have
understanding are as old as AI. In 1957 Herbert Simon
predicted four things AI was supposed to achieve within
ten years (Simon & Newell, 1958, pp. 7-8). The only
one that has been achieved is that a computer has beaten
the best chess player in the world, but that only
happened in 1997. Simon has also asserted:
“I believe that in our time computers will be able to
perform any cognitive task that a person can
perform. I believe that computers already can read,
think, learn, create…” (Simon, 1977, p. 6)
Some recent events, however, may rejuvenate these
discussions. A chess robot broke a seven-year-old-
boy’s finger (Henley, 2022). The Bing AI chatbot called
a CNN reporter “rude and disrespectful”, presumably
for asking too many questions (Kelly, 2023), and
declared love to a NYT journalist and tried to convince
him that he did not love his wife, but the chatbot (Roose,
2023). There are also numerous examples of factual and
logical mistakes made by ChatGPT-4. However, there
was a particularly instructive story that happened in
February 2023. Kellin Pelrine, an American amateur
Go player (ranked one level below the top in amateurs)
beat the top Go computer 14 out of 15 games (Waters,
2023). What makes this point significant, is that all 14
times it was the same trick. If the computer had any
level of understanding of the game, it would have
identified the same trap being set the second time, let
alone being tricked 14 times the same way. However,
these examples only showcase that AI does not possess
understanding right now, but not that it cannot.
Of course, there are many examples of generative
AI (and other forms of AI) delivering incredible
performance, the question is whether AI can understand
and if not now, can AI ever understand (Chomsky et al.,
2023). We believe that the examples conclusively prove
that AI does not understand. We also believe that AI is
not designed to think, but to mimic some of the
outcomes of thinking. Clearly, there are other opinions,
in the end it all boils down to whether we accept the
computational model of the mind. The view of
understanding has a significant impact on the view of AI
ethics, as moral judgments presume understanding.
7. Doubting
Finally, the third pillar of our argument is the
capacity to doubt. When we paraphrase René Descartes,
we accentuate that he was not a Sceptic (one of the four
Hellenistic philosophical traditions, of which David
Hume was a late follower). Descartes was trying to
combat the sceptic dictum that we can doubt everything
by attempting to find solid ground in those things that
we can be really sure about. To this end, he adopted the
Sceptic armament and applied it to everything that he
could think of, in a systematic doubt. In doing so, he
behaved like a Sceptic, demonstrating how we can doubt
anything that we think we know. We cannot be sure that
it is an object in front of us, we cannot even be sure of
our own bodies, it can all be an illusion, an evil demon’s
doing who hijacked our minds. And then, in a master-
stroke, he turns the argument upside down, and
concludes that if one can doubt anything and everything,
then there is one thing that one can be sure of, and it is
that there is something that can doubt. As doubting is a
form of thinking, Descartes (1637, p. 27) formulates his
famous dictum: “Cogito ergo sum” (I think and
therefore I am). However, it is more precise if we limit
the term to doubting, and thus ‘Dubito ergo sum’ (I
doubt and therefore I am). What we are suggesting here,
is that being moral entails the capacity to doubt, we need
to be able to doubt our actions, to doubt ourselves (de
Crescenzo, 1992; e.g. Spiegelberg, 1947).
So, what does it mean to be able to doubt,
particularly in the context of ethics? It entails sensing
our decision situation and understanding it, trying to do
the right thing but not being able to figure out what our
actions may lead to, reflecting and not being sure even
of our motivations. Just think of Hamlet’s painful
struggle whether he should avenge his father. Doubt
incorporates some of the most complex issues of the
human mind, including sensing and understanding, and
it may well be indispensable for our moral decisions, for
our moral development, perhaps the central component
of the moral mind. There is no consensus about this
point, the significance of doubt is our own observation
Page 5592
in the realm of AI ethics. Doubt also has an interesting
implication regarding certainty: there is a significant
body of literature on uncertainty in entrepreneurship,
strategy, and decision making, since Knight (1921,
1923) suggested that our default condition is
uncertainty, in which alternatives and their respective
probabilities are not known. To cope with uncertainty,
we make social contracts, perhaps we can think of doubt
in a similar vein. Doubt scarcely appears in the AI
literature (see Shklovski & Némethy, 2023 for a rare
example). Importantly for AI ethics, if doubt seem to be
essential for our moral judgments, what are the
implications of doubt-less AI? Importantly, while we
developed our argument from a phenomenological
perspective, the concept of doubting can be significant
for AI in any philosophical position.
8. Final commentary
In several ways, AI ethics is a weird concept. It can
cover moral considerations of making AI, ethical
aspects of using AI, potential consequences of the tasks
we assign to AI, and so forth (Asaro, 2006; Floridi &
Sanders, 2002). When we build AI, if we hand over
some of our decisions to it, we need to put in something
that takes care of those aspects that constitute the moral
dimensions of our decisions. How can we do that?
What should it be? We know, for instance, that if
trained on biased data, AI may amplify those biases.
We will not pretend to have figured out how to
build a moral engine for AI. However, we now perhaps
understand a little better what needs to be considered for
such an attempt. This is incredibly timely, it was as we
were writing the first version of this paper that we found
out that Microsoft has sacked all its AI ethics team just
as they were getting ChatGPT into Bing and soon
possibly into many other products. This means one
thing for us: ethical questions of AI are difficult,
complex, ignite heated debates – and it is paramount that
we get them right.
It looks like there is no easy way that would let us
‘program’ ethics into AI, or let it learn it through
ML/DL. The reason is that we do not have a proverbial
perfect moral entity whose moral characteristics or
actions we could use as starting points. Even if we could
find such an entity, we would struggle to identify a
sensible number of learning examples – and we could
not even start figuring this out, as we have no idea how
many types of moral decisions exist. There is also no
large model that could be used by generative AI.
What does this leave us with? Well, there are a few
things we can pin down: (1) The main point of moral
psychology was that we make at least some of our moral
judgments intuitively and only use ethics models to
explain them. (2) Our moral judgments are rooted in
indwelling (typically sensory perception) and we need
to understand the decision situation as well as the
possible consequences of our actions. This does not
mean that we can know the consequences, but we can
think up scenarios. (3) The capacity to doubt seems an
indispensable part of moral decisions. None of this,
however, suggests to us that AI ethics is a futile area.
What we are trying to figure out is what AI can help
us with in terms of moral decisions and to understand
the way forward. There are two immediate things that
we believe AI can do for us right now: (1) AI can
provide us with useful input for our process of self-
doubt (ex ante or ex post), as we are deliberating our
moral decisions by identifying potentially relevant
patterns in available ethics models. This can help in two
ways, it can reduce the struggle of self-doubt and can
help us explain our moral judgments. (2) AI can scan
the context for emerging information and patterns,
feeding this back to us so that we can course-correct
quickly. Humans rely on their ‘felt sense’, like babies
calling for their parents when they need a change of
diapers: they feel uncomfortable. As AI lacks ‘felt
sense’, we need to figure out how to provide external
pointers when AI needs to involve a human in the
process.
Considering the previous discussion we now make
a leap and suggest something that does not trivially
follow from what has been said. AI is an amplifier. It
does not make us smarter, it amplifies what we have,
and if we are stupid, it will amplify that as well (Dörfler,
2022). We have seen e.g. how AI can amplify biases.
However, the leap is the following: we suggest that we
do not actually have AI ethics problems. This is the
reason that we went all the way back and took a journey
in time for 2.5 millennia. We do not have AI ethics
problems, we have ethics problems. AI amplifies them.
9. References
Ackoff, R. L. (1989). From Data to Wisdom. Journal of
Applied Systems Analysis, 16(1), 3-9.
Anderson, S. L. (2011). Philosophical Concerns with
Machine Ethics. In M. Anderson & S. L. Anderson
(Eds.), Machine Ethics (pp. 162-167). Cambridge
University Press.
https://doi.org/10.1017/CBO9780511978036.014
Anderson, S. L., & Anderson, M. (2011). A Prima Facie
Duty Approach to Machine Ethics: Machine Learning of
Features of Ethical Dilemmas, Prima Facie Duties, and
Decision Principles through a Dialogue with Ethicists. In
M. Anderson & S. L. Anderson (Eds.), Machine Ethics
(pp. 476-492). Cambridge University Press.
https://doi.org/10.1017/CBO9780511978036.032
Annas, J. (2011). Intelligent Virtue. Oxford University Press.
Anscombe, G. E. M. (1958). Modern Moral Philosophy.
Philosophy, 33(124), 1-19.
https://doi.org/10.1017/S0031819100037943
Page 5593
Aristotle. (cca 330 BCE/1980). The Nicomachean Ethics (W.
D. Ross, Trans.; W. D. Ross, J. O. Urmson, & J. L.
Ackrill, Eds.). Oxford University Press.
Asaro, P. M. (2006). What Should We Want From a Robot
Ethic? International Review of Information Ethics,
6(12), 9-16.
Balasubramanian, N., Ye, Y., & Xu, M. (2020). Substituting
Human Decision-Making with Machine Learning:
Implications for Organizational Learning. Academy of
Management Review, 0(ja), null.
https://doi.org/10.5465/amr.2019.0470
Bas, A., Sinclair, M., & Dörfler, V. (2022). Sensing: The
Elephant in the Room of Management Learning.
Management Learning.
https://doi.org/10.1177/13505076221077226
Berkeley, G. (1878/2008). A Treatise Concerning the
Principles of Human Knowledge (C. Porterfield Krauth,
Trans.). J.B. Lippincott & Company.
Brey, P. (2000). Disclosive computer ethics. SIGCAS
Comput. Soc., 30(4), 10–16.
https://doi.org/10.1145/572260.572264
Burton, R. A. (2009). On being certain: Believing you are
right even when you're not. St. Martin's Press.
Chase, W. G., & Simon, H. A. (1973a). The Mind's Eye in
Chess. In W. G. Chase (Ed.), Visual Information
Processing (pp. 215-281). Academic Press.
Chase, W. G., & Simon, H. A. (1973b). Perception in Chess.
Cognitive Psychology, 4(1), 55-81.
https://doi.org/10.1016/0010-0285(73)90004-2
Chomsky, N., Roberts, I., & Watumull, J. (2023, 8th March
2003). Noam Chomsky: The False Promise of ChatGPT.
New York Times.
https://www.nytimes.com/2023/03/08/opinion/noam-
chomsky-chatgpt-ai.html
de Crescenzo, L. (1992). Il Dubbio [The Doubt]. Arnoldo
Mondadori.
Csaszar, F. A., & Steinberger, T. (2022). Organizations as
Artificial Intelligences: The Use of Artificial
Intelligence Analogies in Organization Theory. Academy
of Management Annals, 16(1), 1-37.
https://doi.org/10.5465/annals.2020.0192
Davenport, T. H. (2018). The AI Advantage: How to Put the
Artificial Intelligence Revolution to Work. MIT Press.
Davenport, T. H., & Ammanath, B. (2020, 12th August
2020). Redefining Al Leadership in the C-Suite. Sloan
Management Review.
https://sloanreview.mit.edu/article/redefining-ai-
leadership-in-the-c-suite/
Davenport, T. H., & Euchner, J. (2023). The Rise of Human-
Machine Collaboration. Research-Technology
Management, 66(1), 11-15.
https://doi.org/10.1080/08956308.2023.2142435
Davenport, T. H., & Glaser, J. P. (2022). Factors governing
the adoption of artificial intelligence in healthcare
providers. Discover Health Systems, 1(4).
https://doi.org/10.1007/s44250-022-00004-8
Davenport, T. H., & Glover, W. J. (2018). Artificial
intelligence and the augmentation of health care
decision-making. NEJM Catalyst, 4(3).
https://catalyst.nejm.org/ai-technologies-augmentation-
healthcare-decisions/
Davenport, T. H., & Miller, S. M. (2022). Working with AI:
Real Stories of Human-Machine Collaboration. MIT
Press.
Davenport, T. H., & Mittal, N. (2022). How Companies Can
Prepare for the Coming “AI-first” World. Strategy &
Leadership, ahead-of-print(ahead-of-print).
https://doi.org/10.1108/SL-11-2022-0107
Davenport, T. H., & Mittal, N. (2023). All-in On AI: How
Smart Companies Win Big with Artificial Intelligence.
Harvard Business Review Press.
Descartes, R. (1637/1986). Discourse on the Method of
Rightly Conducting the Reason, and Seeking Truth in
the Sciences. In R. Descartes (Ed.), A Discourse on
Method, Meditations and Principles. Dent.
Dörfler, V. (2020). Artificial Intelligence. In M. A. Runco &
S. R. Pritzker (Eds.), Encyclopedia of Creativity (3rd ed.,
pp. 57-64). Academic Press.
https://doi.org/10.1016/B978-0-12-809324-5.23863-7
Dörfler, V. (2022). What Every CEO Should Know About AI.
Cambridge University Press.
Dörfler, V. (2023a). Artificial Intelligence. In J. Mattingly
(Ed.), The SAGE Encyclopedia of Theory in Science,
Technology, Engineering, and Mathematics (Vol. 1, pp.
37-41). Sage.
https://doi.org/10.4135/9781071872383.n15
Dörfler, V. (2023b). Critical Interpretivism: The First
Outline PHILOS 2023: 3rd Colloquium on Philosophy
and Organization Studies, Chania, Greece
Dörfler, V., Baracskai, Z., & Velencei, J. (2009, 7-11 August
2009). Knowledge Levels: 3-D Model of the Levels of
Expertise AoM 2009: 69th Annual Meeting of the
Academy of Management, Chicago, IL.
https://www.researchgate.net/publication/308339223
Dörfler, V., & Bas, A. (2020). Intuition: scientific, non-
scientific or unscientific? In M. Sinclair (Ed.),
Handbook of Intuition Research as Practice (pp. 293-
305). Edward Elgar.
https://doi.org/10.4337/9781788979757.00033
Dörfler, V., & Stierand, M. (2017). The Underpinnings of
Intuition. In J. Liebowitz, J. Paliszkiewicz, & J.
Gołuchowski (Eds.), Intuition, Trust, and Analytics (pp.
3-20). Taylor & Francis.
https://doi.org/10.1201/9781315195551-1
Dörfler, V., & Stierand, M. (2021). Bracketing: A
Phenomenological Theory Applied Through
Transpersonal Reflexivity. Journal of Organizational
Change Management, 34(4), 778-793.
https://doi.org/10.1108/JOCM-12-2019-0393
Dreyfus, H. L. (1992). What Computers Still Can't Do: A
Critique of Artificial Reason (revised ed.). MIT Press.
Dreyfus, H. L. (2004). A Phenomenology of Skill Acquisition
as the basis for a Merleau-Pontian
Nonrepresentationalist Cognitive Science. University of
California, Department of Philosophy.
https://philpapers.org/archive/DREAPO.pdf
Dreyfus, H. L., & Dreyfus, S. E. (1986/2000). Mind over
Machine: The Power of Human Intuition and Expertise
in the Era of the Computer. The Free Press.
Fisher, G., Mayer, K., & Morris, S. (2021). From the Editors
- Phenomenon-Based Theorizing. Academy of
Page 5594
Management Review, 46(4), 631-639.
https://doi.org/10.5465/amr.2021.0320
Flanagan, M., Howe, D. C., & Nissenbaum, H. (2008).
Embodying Values in Technology: Theory and Practice.
In J. van den Hoven & J. Weckert (Eds.), Information
Technology and Moral Philosophy (pp. 322-353).
Cambridge University Press.
https://doi.org/10.1017/CBO9780511498725.017
Floridi, L., & Sanders, J. W. (2002). Mapping the
Foundationalist Debate in Computer Ethics. Ethics and
Information Technology, 4(1), 1-9.
https://doi.org/10.1023/A:1015209807065
Foot, P. (1978). Virtues and Vices and Other Essays in Moral
Philosophy. University of California Press.
Friedman, B., Kahn, P. H., Borning, A., & Huldtgren, A.
(2013). Value Sensitive Design and Information
Systems. In N. Doorn, D. Schuurbiers, I. van de Poel, &
M. E. Gorman (Eds.), Early engagement and new
technologies: Opening up the laboratory (pp. 55-95).
Springer Netherlands. https://doi.org/10.1007/978-94-
007-7844-3_4
Gilligan, C. (2014). Moral Injury and the Ethic of Care:
Reframing the Conversation about Differences. Journal
of Social Philosophy, 45(1), 89-106.
https://doi.org/10.1111/josp.12050
Glikson, E., & Woolley, A. W. (2020). Human Trust in
Artificial Intelligence: Review of Empirical Research.
Academy of Management Annals, 14(2), 627-660.
https://doi.org/10.5465/annals.2018.0057
Gobet, F., & Simon, H. A. (1996a). Recall of Random and
Distorted Chess Positions: Implications for the Theory
of Expertise. Memory & Cognition, 24(4), 493-503.
https://doi.org/10.3758/BF03200937
Gobet, F., & Simon, H. A. (1996b). Templates in Chess
Memory: Mechanism for Re-calling Several Boards.
Cognitive Psychology, 31(1), 1-40.
https://doi.org/10.1006/cogp.1996.0011
Gobet, F., & Simon, H. A. (2000). Five seconds or sixty?
Presentation time in expert memory. Cognitive Science,
24(4), 651-682. https://doi.org/10.1016/S0364-
0213(00)00031-8
Göndöcs, D., & Dörfler, V. (2022, 7-9 July 2022). AI-
enabled Organizational Learning Strategy EGOS 2022:
38th Colloquium of the European Group for
Organization Studies, Vienna, Austria.
https://www.researchgate.net/publication/363049580
Göndöcs, D., & Dörfler, V. (2023, 7-14 February 2023). AI
in Medical Diagnosis: AI Prediction vs Human
Judgement AAAI 2023: 37th AAAI (Association for the
Advancement of Artificial Intelligence) Conference on
Artificial Intelligence, Washington, DC.
https://www.researchgate.net/publication/368779553
Gottlieb, P. (1984). Aristotle on Dividing the Soul and
Uniting the Virtues. Phronesis, 39(3), 275-290.
https://doi.org/10.1163/156852894321052081
Grodal, S., Krabbe, A. D., & Chan-Zninog, M. (2023). The
Evolution of Technology. Academy of Management
Annals, 17(1), 141-180.
https://doi.org/10.5465/annals.2021.0086
Gunkel, D. J. (2017). The Machine Question: Critical
Perspectives on AI, Robots, and Ethics. MIT Press.
Haidt, J. (2001). The emotional dog and its rational tail: A
social intuitionist approach to moral judgment.
Psychological Review, 108(4), 814-834.
https://doi.org/10.1037/0033-295X.108.4.814
Hassabis, D. (2017). The Future of AI New Scientist Live
2017, London, UK.
Henley, J. (2022, 24th July). Chess robot grabs and breaks
finger of seven-year-old opponent. The Guardian.
https://www.theguardian.com/sport/2022/jul/24/chess-
robot-grabs-and-breaks-finger-of-seven-year-old-
opponent-moscow
Hursthouse, R. (1999). On Virtue Ethics. Oxford University
Press.
James, W. (1899/2001). On a Certain Blindness in Human
Beings. In W. James (Ed.), Talks to Teachers on
Psychology: And to Students on Some of Life's Ideals.
Dover.
Kant, I. (1785/1998). Groundwork of the Metaphysics of
Morals (M. Gregor, Ed.). Cambridge University Press.
Kant, I. (1797/2017). The Metaphysics of Morals (M. Gregor,
Trans.; L. Denis, Ed. revised ed.). Cambridge University
Press.
Kelly, S. M. (2023, 16th February). The Dark Side of Bing’s
New AI Chatbot. CNN.
https://www.cnn.com/2023/02/16/tech/bing-dark-side
Knight, F. H. (1921). Risk, Uncertainty and Profit. Houghton
Mifflin.
Knight, F. H. (1923). Business Management: Science or Art?
Journal of Business, 2(4th March), 5-24.
Köhler, W. (1959). Gestalt Psychology Today: Address of
the President at the sixty-seventh Annual Convention of
the American Psychological Association, Cincinnati,
Ohio, September 6, 1959. American Psychologist,
14(12), 727-734. https://doi.org/10.1037/h0042492
Kreisler, H., & Dreyfus, H. L. (2005). Meaning, Relevance,
and the Limits of Technology: Interview with Hubert
Dreyfus (Conversations with History, Issue.
http://globetrotter.berkeley.edu/people5/Dreyfus/
von Krogh, G. (2018). Artificial Intelligence in
Organizations: New Opportunities for Phenomenon-
Based Theorizing. Academy of Management
Discoveries, 4(4), 404-409.
https://doi.org/10.5465/amd.2018.0084
Langley, A. (2021). What Is “This” a Case of? Generative
Theorizing for Disruptive Times. Journal of
Management Inquiry, 30(3), 251-258.
https://doi.org/10.1177/10564926211016545
Leavitt, K., Schabram, K., Hariharan, P., & Barnes, C. M.
(2021). The Machine Hums! Addressing Ontological
and Normative Concerns Regarding Machine Learning
Applications in Organizational Scholarship. Academy of
Management Review, 0(ja), null.
https://doi.org/10.5465/amr.2021.0166
LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning.
Nature, 521(7553), 436-444.
https://doi.org/10.1038/nature14539
Levinas, E. (1961/1991). Totality and Infinity: An Essay on
Exteriority. Kluwer Academic Publishers.
Levinas, E. (1991/2017). Entre Nous (B. Harshav & M. B.
Smith, Trans.). Bloomsbury.
Page 5595
Lindebaum, D., & Ashraf, M. (2021). The Ghost in the
Machine, or the Ghost in Organizational Theory? A
Complementary View on the Use of Machine Learning.
Academy of Management Review, 0(ja), null.
https://doi.org/10.5465/amr.2021.0036
Locke, J. (1690/1959). An Essay Concerning Human
Understanding (Vol. 1). Dover Publications.
MacIntyre, A. (1998). A Short History of Ethics: A History of
Moral Philosophy from the Homeric Age to the
Twentieth Century (2nd ed.). Routledge.
MacIntyre, A. (2007). After Virtue: A Study in Moral Theory
(3rd ed.). University of Notre Dame Press.
Marcus, G. (2018). Deep learning: A critical appraisal. arXiv
preprint arXiv:1801.00631.
Mill, J. S. (1861). Utilitarianism. In G. Williams (Ed.),
Utilitarianism, On Liberty, Considerations on
Representative Government, Remarks on Bentham's
Philosophy (pp. 1-67). Everyman.
Moor, J. H. (1985). What Is Computer Ethics?
Metaphilosophy, 16(4), 266-275.
https://doi.org/10.1111/j.1467-9973.1985.tb00173.x
Oliver, N., Calvard, T., & Potočnik, K. (2017). Cognition,
Technology, and Organizational Limits: Lessons from
the Air France 447 Disaster. Organization Science,
28(4), 729-743. https://doi.org/10.1287/orsc.2017.1138
Pachidi, S., Berends, H., Faraj, S., & Huysman, M. (2021).
Make Way for the Algorithms: Symbolic Actions and
Change in a Regime of Knowing. Organization Science,
32(1), 18-41. https://doi.org/10.1287/orsc.2020.1377
Ployhart, R. E., & Bartunek, J. M. (2019). Editors’
Comments: There Is Nothing So Theoretical As Good
Practice – A Call for Phenomenal Theory. Academy of
Management Review, 44(3), 493-497.
https://doi.org/10.5465/amr.2019.0087
Polányi, M. (1966a). The Logic of Tacit Inference.
Philosophy, 41(155), 1-18.
https://doi.org/10.1017/S0031819100066110
Polányi, M. (1966b/1983). The Tacit Dimension. Peter Smith.
Raisch, S., & Krakowski, S. (2021). Artificial Intelligence
and Management: The Automation-Augmentation
Paradox. Academy of Management Review, 46(1), 192–
210. https://doi.org/10.5465/amr.2018.0072
Rest, J. R. (1986). Moral Development: Advances in
Research and Theory. University of Minnesota Press.
Rock, I., & Palmer, S. (1990). The Legacy of Gestalt
Psychology. Scientific American, 263(6), 84-91.
http://www.jstor.org/stable/24997014
de Rond, M., Holeman, I., & Howard-Grenville, J. (2019).
Sensemaking from the Body: An Enactive Ethnography
of Rowing the Amazon. Academy of Management
Journal, 62(6), 1961-1988.
https://doi.org/10.5465/amj.2017.1417
Roose, K. (2023). A Conversation With Bing’s Chatbot Left
Me Deeply Unsettled. New York Times.
https://www.nytimes.com/2023/02/16/technology/bing-
chatbot-microsoft-chatgpt.html?smid=url-share
Sætre, A. S., & Van de Ven, A. H. (2022). Abductive
Theorizing Is More than Idea Generation: Disciplined
Imagination and a Prepared Mind [Dialogue]. Academy
of Management Review, 0(ja), null.
https://doi.org/10.5465/amr.2021.0317
Schmidhuber, J. (2015). Deep Learning in Neural Networks:
An Overview. Neural Networks, 61, 85-117.
https://doi.org/10.1016/j.neunet.2014.09.003
Searle, J. R. (1994/2002). The Rediscovery of the Mind
(paperback ed.). MIT Press.
Shklovski, I., & Némethy, C. (2023). Nodes of certainty and
spaces for doubt in AI ethics for engineers. Information,
Communication & Society, 26(1), 37-53.
https://doi.org/10.1080/1369118X.2021.2014547
Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L.,
van den Driessche, G., Schrittwieser, J., Antonoglou, I.,
Panneershelvam, V., Lanctot, M., Dieleman, S., Grewe,
D., Nham, J., Kalchbrenner, N., Sutskever, I., Lillicrap,
T., Leach, M., Kavukcuoglu, K., Graepel, T., &
Hassabis, D. (2016). Mastering the game of Go with
deep neural networks and tree search [Article]. Nature,
529, 484. https://doi.org/10.1038/nature16961
Simon, H. A. (1977). The New Science of Management
Decision (3rd ed.). Prentice-Hall.
Simon, H. A., & Newell, A. (1958). Heuristic Problem
Solving: The Next Advance in Operations Research.
Operations Research, 6(1), 1-10.
https://doi.org/10.1287/opre.6.1.1
Sonenshein, S. (2007). The Role of Construction, Intuition,
and Justification in Responding to Ethical Issues at
Work: The Sensemaking-Intuition Model. Academy of
Management Review, 23(4), 1022-1040.
https://doi.org/10.5465/AMR.2007.26585677
Spiegelberg, H. (1947). Indubitables in Ethics: A Cartesian
Meditation. Ethics, 58(1), 35-50.
https://doi.org/10.1086/290589
Strati, A. (2007). Sensible knowledge and practice-based
learning. Management Learning, 38(1), 61-77.
https://doi.org/10.1177/1350507607073023
Treviño, L. K. (1986). Ethical Decision Making in
Organizations: A Person-Situation Interactionist Model.
Academy of Management Review, 11(3), 601-617.
https://doi.org/10.5465/amr.1986.4306235
Tschang, F. T., & Almirall, E. M. (2021). Artificial
Intelligence as Augmenting Automation: Implications
for Employment. Academy of Management Perspectives,
35(4), 642-659. https://doi.org/10.5465/amp.2019.0062
Vallor, S. (2016). Technology and the virtues: a
philosophical guide to a future worth wanting. Oxford
University Press.
Waters, R. (2023, 17th February). Man beats machine at Go
in human victory over AI. Financial Times.
https://www.ft.com/content/175e5314-a7f7-4741-a786-
273219f433a1
Wiener, N. (1960). Some Moral and Technical Consequences
of Automation. Science, 131(3410), 1355-1358.
https://doi.org/10.1126/science.131.3410.1355
Williams, B. (1985/2006). Ethics and the Limits of
Philosophy. Routledge.
Williams, B. (1993). Morality: An Introduction to Ethics.
Cambridge University Press.
van Wynsberghe, A. (2013). Designing Robots for Care:
Care Centered Value-Sensitive Design. Science and
Engineering Ethics, 19(2), 407-433.
https://doi.org/10.1007/s11948-011-9343-6
Page 5596
