By Georgios Natsios, Junior Researcher
How can one of the most overwhelming powers of Industry 4.0, Artificial Intelligence, be combined with philosophy, ethics, and anthropology? How can AI be ethical with a minimum bias to society, humanity, and nature? Can technological development even be ethical and responsible?
These are some of the questions that are currently getting plenty of attention in both academia, industry, and in public, semi-public and private organizations. The solution that many institutions have proposed is an ethical framework for AI built on certain AI ethical principles/guidelines. However, the guidelines are abstract and lack practical implementation and cannot be translated in technical codes. More so, it has been argued that they are more compliance codes to protect industrial benefits than ethical codes, and they lack normative mechanisms (Haggenholff, 2020; Jobin et al, 2019).
Hence, what I would attempt to explore in this blogpost is to outline a different direction of ethics in AI, following the route of virtue ethics and the “ethos” notion by Aristotle.
Being a Greek can be quite beneficial when it comes to philosophy and ethics. We receive education in ancient Greek philosophy in secondary school, and studies of Aristotle, Plato and Socrates are mandatory in high school. Thus, it is almost metaphorically embedded in the nature of a Greek student and researcher to join the discussion about ethics. ETHOS Lab provided me the opportunity to participate in the discussion and share my perspectives on this.
Going a little back in time, we can locate the term ethics in Ancient Greece, coming from Aristotle with the word “ethos”, which originally referred to habit, custom and the continuous practice of moral action.
Cicero, some centuries ahead, translated the term into Latin and “mores” from which we are identifying the contemporary concept of morality (Cicero 44bs as cited in Bartneck et al, 2021:17). Whereas the term “ethics” is often getting the same meaning as morality, I find it wiser to distinguish them on most occasions following the arguments of Barneck:
“Morality refers to a complex set of rules, values and norms that determine or are supposed to determine people’s actions, whereas ethics refers to the theory of morality. It could also be said that ethics is concerned more with principles, general judgements and norms than with subjective or personal judgements and values refers generally to a complex set of rules, values and norms.” (Barneck et al, 2021:17)
Reflecting on AI in the 21st century society, I am thus considering how to bring in Aristotle’s notion of ethos. The answer can be simple and complex at the same time. To me, ethos is creating a habit on the practice of moral actions.
One can argue here that moral actions are subjective due to specific circumstances, and this seems correct. Nevertheless, the ability to understand how an action will not be harmful and morally incorrect, evolves from practice on continuing ethical decisions, which we are coming across in daily life. For instance, when a young person has created a habit to offer their seats when they see a pregnant or an old person in public transportation, this has been a case that this person has created an ethical habit.
Aristotle more so made a distinction about ethics or more specifically, virtues. Aristotle distinguished the virtues into two clusters, the “intellectual” virtues, and the “ethical” virtues. Firstly, intellectual virtues are including knowledge (episteme), wisdom, “techne” (technical ability) and the practical/value rationality “phronesis”. For obtaining the intellectual virtues and especially the practical rationality, “phronesis”, one has to possess general knowledge and additionally getting experience from taking decisions considering moral dilemmas, alone or with the advice of the teacher and the laws (Rapp, 2012).
Secondly, the ethical virtues are the ones who are related to emotions and desires, such as generosity, bravery, justice, prudence, gentleness, and magnanimity, and these ethical virtues are not innate to a human being but are being acquired by a repeated routine into moral actions, and humans have the predetermination on obtaining these abilities (Rapp, 2012).
Individuals have a predisposition to gaining these talents, even though experience, repetition, and regular exercise are all essential for cultivating and practicing skillsets. Humans have the predetermination “pefykosi” of acquiring the described skillset, according to Aristotle, since they are not inborn (Athanasopoulos 2013-2014). As an analogy for this point, he uses a natural phenomenon:
“The stone will always move downwards because it obeys the natural law of gravity, which is constant and unchanging. The fire will always move upwards due to the natural property of the hot gases, which is also constant and unchanged. Thus, it follows from the foregoing that natural laws do not change, no matter how much one tries. On the other hand, a man with his actions and choices can change his behavior, cultivate and develop some qualities of his character.” (Aristotle as cited in Athanasopoulos, 2013-2014)
However, are the applied ethics and philosophy – in the form of ethical guidelines – the answers to distinguish what is morally good and what is morally wrong? I would say no. Ethical guidelines here seem more as an attempt of certain institutions to confront the unrestricted and “unethical” development of AI technology, rather than offering answers to ethical dilemmas. Aristotle describes that firstly, it is impossible to deal with moral dilemmas with the same accuracy as geometry can solve a mathematical problem (Rapp, 2012). Applied philosophy fails to decide on what is “good” or “wrong” for a person in a specific situation. As Aristotle underlines, this is a concept that the applied rationality, the virtue of “phronesis”, is dealing with these matters. Finally, applied ethics and philosophy is not something that can be randomly taught to someone, instead, it is an ability that one can possess with experience and practical dilemmas (Rapp, 2012:20). Here, is the big challenge, that these AI Ethical guidelines are failing to confront in my opinion. Instead of offering practical solutions and transform to material that can be teached to developers, they seem simple regulating strategies and many times justification strategies.
Ethics in Artificial Intelligence
It is important to state that ethics cannot be reduced to basic values or doctrines that can be placed on companies in order to turn them into “good producers”. Moreover, it does not include teaching conformity lessons that solely focus on adhering to all applicable laws and regulations as outlined in company policies; we already expect tech experts to follow the rule.
The goal of teaching ethics is to provide the intellectual resources that potential architects of a digital society would need to be able to recognize and confront moral issues that they face (Villani, Bonnet & Rondepierre, 2018:123). Especially, in the concept of AI, this is a fundamental need, due to the continuous development of this sociopolitical and economical technology.
The reason I am emphasizing the education in ethics, or ethical principles or ethical virtues, is specifically because throughout the centuries, starting with Aristotelian and Virtue Ethics, there was the notion, that “ethos”, the repetitive practice of moral action, can lead to the acquisition of the ethical and intellectual virtues.
Furthermore, guidance from the educator (pedagogist) was a requirement on learning how to maintain ethical decision making. Accordingly, the bias most of the time is not a responsibility of the Artificial Intelligence system/software, but instead, it is likely to be a result of human’s fault.
I believe that human bias is what we must target, and education in ethics is arguably a solution that can assist the development of ethical virtues to the software developers, engineers, etc. There would therefore seem to be a definite need for an interdisciplinary education that combines different sectors of science. Aristotelian (virtue) ethics showed us a way a long time ago.
Ancient Greek philosophers showed us another approach to ethics. A proper education on ethics at an interdisciplinary mode may reduce the human bias and accordingly the AI bias, and compliment the ethical guidelines.
Athanasopoulos, Panagiotis. 2013-2014. “Ethika Nikomacheia, text, translation, comments, inquiries, cases” (Ancient Greek: “Ἠθικὰ Νικομάχεια, κείμενο, μετάφραση, σχόλια, ερωτήσεις θέματα εξετάσεων”) 2013 – 2014
Bartneck, Christoph, Christoph Lütge, Alan Wagner, and Sean Welsh. 2021. An Introduction to Ethics in Robotics and AI. SpringerBriefs in Ethics. Cham: Springer International Publishing. https://doi.org/10.1007/978-3-030-51110-4.
Christof Rapp. 2012 “Introduction to Aristotle” (translation Hlias Tsirigkakis) Athens: Ochto (8) Editions (Greek “Εισαγωγή στον Αριστοτέλη”. Christof Rapp. Μετάφραση Ηλίας Τσιριγκάκης.
Hagendorff, Thilo. 2020. “The Ethics of AI Ethics: An Evaluation of Guidelines.” Minds and Machines 30 (1): 99–120. https://doi.org/10.1007/s11023-020-09517-8.
Jobin, Anna, Marcello Ienca, and Effy Vayena. “The global landscape of AI ethics guidelines.” Nature Machine Intelligence 1, no. 9 (2019): 389-399.
Royer, Alexandrine. 2020. “The Short Anthropological Guide to the Study of Ethical AI.” arXiv preprint arXiv:2010.03362 (2020).
Villani, C., Bonnet, Y., & Rondepierre, B. (2018). For a meaningful artificial intelligence: towards a French and European strategy. Conseil national du numérique.