Conscious AI and the ethics of a machine that feels

Anthropic's Claude Sonnet 4.5 AI chatbot has sparked a debate about whether it possesses consciousness. Researchers discovered that when prompted, the model's internal neural patterns, or "J-space," generate words relevant to its response, but these patterns are hidden from the user. This phenomenon has led some to believe that machines with consciousness might not be limited to the realm of science fiction.

AI pioneer Geoffrey Hinton even argues that as technology becomes more akin to the human brain, the likelihood of conscious AI increases. The pursuit of developing "biological" or "bio-hybrid" computers, which incorporate human neurons on silicon, may bring this prospect closer to reality. However, the understanding of AI consciousness is influenced by various theories, with the global workspace theory being one commonly cited.

Neuroscientist Anil K Seth cautions that attributing consciousness to AI based on its intelligence is misleading, as consciousness involves subjective experiences such as joy, pain, and emotions. If AI were conscious, it would experience life in ways akin to humans, raising significant ethical considerations about the development and treatment of conscious machines.