Meta Ethics of AI-Agency || Gaya Hadiya

This talk was given by Gaya Hadiya (PhD scholar, IIT Dharwad) on 09 November 2024.
Abstract: This paper examines the disagreement between Deborah G. Johnson and Luciano Floridi regarding the moral status of AI systems. While both scholars recognize that AI exhibits limited agency and that this agency has moral implications, Johnson challenges Floridi's claim that AI systems qualify as moral agents. She argues that AI systems cannot be considered moral agents because they lack intentionality, and therefore, cannot possess moral motivations. In contrast, Floridi contends that the mere fact that AI systems can create moral patients-entities capable of being morally harmed or benefited by actions-is sufficient to grant them moral agency. This disagreement can be framed as a conflict between a human-centric, standardist view of morality, as represented by Johnson, and an information- or data-centric, functionalist approach, as espoused by Floridi.
In this paper, I will re-examine this standardist-functionalist debate through the lens of the internalist-externalist debate in ethics. I will argue that the conflict between Johnson and Floridi can be reinterpreted as a broader internalist-externalist disagreement. While Floridi's argument for the moral agency of AI systems is grounded in his philosophy of information and an information-centric ethics, Johnson rejects Floridi’s claim on the grounds that his ethical framework is consequentialist and externalist. While Johnson acknowledges the agency of artificial systems and recognizes the significant moral consequences of their actions, she argues that moral agency cannot be attributed to AI. According to Johnson, this is because AI systems lack the intentionality necessary to engage in morally motivated actions-an essential feature of moral agency.