On the moral and epistemic agency of AI systems

This paper addresses the question whether moral agency can be ascribed to technical artefacts. We focus on AI systems, which are able to process information, represent and guide the decisions of individuals and the community. We establish similarities between the agency - epistemic and moral - of AI systems and human beings, but we argue that only human beings have ethical subjectivity. Finally, we attribute to human beings (but not to AI systems) the role of ‘epistemic legislators’, highlighting how this ability determines the asymmetric, non-neutral nature of the interaction between human and AI agents. Our contribution attempts to initiate an epistemic and anthropo-ethical reflection on the type of agency that the presence of algorithms can generate, to define a vocabulary to explain and justify the innovations that computer algorithms introduce into human life.