Abstract:
With rapid advances in the social and cognitive capabilities of AI, questions concerning their consciousness and moral status are becoming increasingly pressing. In this talk, I contrast two radically different approaches—Deep and Shallow—to addressing these issues. Deep approaches dominate mainstream philosophy of mind and consciousness science, viewing consciousness as an empirically discoverable natural kind whose essence must be identified by consciousness researchers. On this view, empirical discoveries allow robust inferences about consciousness in non-human cases, including animals and AI. Shallow approaches, in contrast, treat consciousness primarily as a folk-psychological concept whose criteria for application rely chiefly on pragmatic and normative considerations, granting only a limited role to empirical science. I argue that the tension between these approaches becomes particularly salient for AI systems, which typically lack the biological and structural criteria emphasized by Deep theories yet readily fulfill the pragmatic and normative conditions favored by Shallow theories. Both approaches, however, face unresolved methodological challenges and questions that I explore in-depth in the talk. I conclude by proposing that productive progress may arise not by adjudicating between these frameworks, but by recognizing them as distinct yet complementary projects.
