Key questions about artificial sentience: an opinionated guide

Abstract

The question of artificial sentience is explored, with the main argument being that it is important to develop a precise computational theory of sentience that applies to both biological organisms and artificial systems. This theory should explain how various conscious valenced experiences, such as pleasure and pain, arise in both biological and artificial systems. The article discusses the five key assumptions that underpin this argument, and then explores the challenges of formulating such a theory. In particular, the article raises questions about the nature of valenced states, their relation to reward and reinforcement learning, and the difficulty of applying existing theories of consciousness to specific AI systems. It is argued that more detailed and rigorous cross-talk between consciousness scientists and AI researchers is needed to advance our understanding of AI sentience. – AI-generated abstract.