Is AI Conscious?

A mind-expanding conversation on how models actually work | Emmanuel Ameisen (Anthropic)

Topic 6 of 18

Published Oct 18, 2026 · Clip 23:02–24:45 (1:43)

Asked whether AI is conscious, Emmanuel says it's hard to say either way, but interpretability offers an unbiased window because you can look at internal structure rather than what the model says. He strongly believes people underestimate the emergent complexity inside models — concept representations, planning, mixed high- and low-level reasoning — and that the more you look inside, the less certain you become.