In this lecture we will discuss some codes of ethics and guidelines to regulate the design and the use of Artificial Intelligence. More precisely, the goal of these codes and guidelines is to regulate the behavior of the different actors involved in the development of AI products. For example to clarify the responsibilities of professionals or to discuss which values to adopt in order to promote human well-being. Codes of ethics and conduct are a core process in the so-called professional ethics. Professional ethics concerns duties, rights and responsibilities of professionals. It is connected to professional responsibility that is based on one’s role as a professional, as far as it stays within the limits of what is morally allowed. A code of ethics is a document in which organizations (like companies or professional associations) lay down guidelines for the responsible behavior of their members. More precisely, a professional code is a code of conduct formulated by a professional association. And a corporate code is a code of conduct formulated by a company. Codes of conduct are formulated for a variety of reasons, such as increasing moral awareness, identifying and interpreting the moral norms and values of a profession or a company, increasing accountability to the outside world, improving the image of a profession or a company. At the core of professional codes lie these values: Integrity: it means living by one’s own (moral) values, norms and commitments. Honesty: it means telling what is believed to be true and disclosing all relevant information. Avoidance of conflicts of interest. A conflict of interest is a situation in which one has an interest (personal or professional) that, when achieved, can conflict with the fulfillment of one’s professional obligations to an employer or to other clients. Let’s see some examples. ACM, the Association of Computing Machinery, one of the largest computing professional associations of the world, has adopted a code of ethics and professional conduct in 1992. A revised version, that takes into account the advances of AI, has been adopted in 2018. Another interesting example from a group of professionals is the document “IEEE Ethically Aligned Design”, released in 2019. IEEE, the Institute of Electrical and Electronics Engineers, is a professional association for electronic and electrical engineering (and associated disciplines). The document has been created to establish societal and policy guidelines in order for autonomous and intelligent systems to remain human-centric, serving humanity’s values and ethical principles. The focus is to develop systems behaving in a way that is beneficial to people beyond reaching functional goals and addressing technical problems. In 2018, a High-Level Expert Group on AI presented Ethics Guidelines for Trustworthy Artificial Intelligence. These guidelines put forward a set of key requirements that AI systems should meet in order to be deemed trustworthy. For example, AI systems should empower human beings, allowing them to make informed decisions and fostering their fundamental rights. At the same time, AI systems need to be resilient and secure. Moreover, besides ensuring full respect for privacy and data protection, adequate data governance mechanisms must also be ensured. Also data, system and AI business models should be transparent. Unfair bias must be avoided, and AI systems should be accessible to all. AI systems should benefit all human beings. Hence, it must be ensured that they are sustainable and environmentally friendly. And finally, mechanisms should be put in place to ensure responsibility and accountability for AI systems and their outcomes. In general, a number of objections can be leveled with respect to codes and guidelines. Sometimes codes of conduct are formulated only for reasons of self-interest: this is the case of window-dressing, that is to present a favorable impression that is not based on actual facts. Codes can result in contradictory recommendations about what to do in a specific situation; for example uncritical loyalty is to place the interests of the employer above any other considerations. Also, some scholars argue that drafting a code of conduct is misperceived because ethics cannot be codified. Sometimes codes of conduct contain provisions that are difficult to put into practice. Besides these issues, it’s fair to say that both codes of ethics and guidelines do not define once and for all what is good or bad. Rather, they represent a shared starting point that needs to be declined and implemented case by case. Moreover, since AI today has a wide public dimension, it’s necessary to move from single individuals to social groups to adopt ethical approaches able to represent as much as possible the values shared by the different parts involved.
