IEEE Technology and Society Magazine - March 2022 - 79

Some claim that the ethics of autonomous vehicles
should be " customized to the maximum common
denominator of the territory where they will be
used " [12]. Sure: a different ethics code for choosing
who to run over in New York, Johannesburg,
and Hong Kong? Can one say more barbarism? As a
technical solution, making a vehicle behave as most
drivers would, in a given area, might be a suitable
technological solution. No doubt, a vehicle programed
to operate in Stockholm would be disastrous
if driving in Rome, and vice versa, even worse! But
please do not call this ethical behavior. It may be
effective, but let us not fool ourselves into thinking
that this is ethics.
Generally, the terms " ethics " and " morality " are
used interchangeably, especially in academic contexts
[13]. Nevertheless, many distinctions between
these two terms have been proposed, even if none
has achieved universal consensus. One of the most
common distinctions assigns morality a " descriptive "
sense, whereas ethics is assigned a " normative "
sense. In other words, morality describes social customs
or codes of conduct; ethics instead refers to
what is actually right or wrong, beyond whatever is
socially accepted. In this sense, the Moral Machine is
certainly effective in its reflecting social customs, but
it will never be a true Ethical Machine.
The approach taken by the Moral Machine has
been severely criticized, questioning the importance
and the real value of finding strong ties
between the respondents' preferences and their
cultural traits, and even the morality of the whole
discussion [14]. Jaques reaches the point of calling
the Moral Machine " a monster " [15], because it, in
fact, invites people to express preferences-that is,
biases-for external indicators of social value. Not
only because humans are very bad at judging people
based on appearances, but especially because
the Moral Machine gives the impression that this
perceived social value should influence their decision.
It suggests and even advocates for the moral
relevance of those features, which is nothing less
than a direct attack to equity. Etienne [16] discusses
the dangers of the Moral Machine experiment,
alerting against both its uses for normative
ends and the whole voting system approach it is
built upon to address ethical issues. According to
Puri [17], even if moral behavior can be imitated by
algorithms, only conscious agents can take moral
decisions and assume responsibility for them. From
March 2022
a more general standpoint, the Proceedings of the
ieee special issue on machine ethics contains several
articles that emphasize the protection due to
human values in future autonomous systems; especially
significant here is Bremner et al. [18], who
argue for transparent and verifiable ethical reasoning
in AI systems.
Ethics is not an imitation game. Ethics is not about
following a code of conduct like Asimov's Three
Laws. But neither is ethics about imitating the behavior
of others. Learning by imitation is something very
human, but if there is any difference between #StayHome
and #CollectToiletPaper, that difference is not
in the number of people who behave one way or
another, but in the reasonableness of their behavior.
Ethics is not spread like a virus, but it is learned and
discussed rationally.
Rationality of ethics: Equity and
equality
So, one of the obstacles we have to overcome
in order to recover sanity is the widespread skepticism
in our society about the rationality of ethics
and values [19], which is also manifested in the
existence of multiple incompatible ethical theories
[6], [20]. Emotivism-not a healthy integration
of emotions with reason in moral judgment, but a
replacement of the latter with the former-is promoted
in approaches such as the Moral Machine,
where the respondents' reasons for their answers are
neglected; only the raw answers are counted, as if
to say: " ethics is not rational in itself, so we must be
content to reflect what most people say. " As we have
shown, this is radically insufficient to avoid biased
judgments and ensure equity.
The role of emotions in moral decision-making is
well established [21], but that does not mean that
we can reduce ethics to emotion, as emotivism does.
In a cultural context where post-truth is blowing up
current knowledge structures [22], an algorithm that
produces an appearance of consensus in emotional
reactions of a large mass of people, but without valuing
arguments, runs more risk of being manipulated,
biased or intoxicated than before the emergence of
social networks. The good news is AI is forcing us to
think ethics in a new way.
One of the most frequently heard forms of ethical
argumentation is: " if everyone acted like that,
then such a thing would happen " (which is clearly
a popular derivation of Kant's categorical/universal
79
IEEE Technology and Society Magazine - March 2022

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - March 2022
