Episode Details
Back to EpisodesAre Chatbots Conscious? A Conversation with Henry Shevlin
Description
Henry Shevlin, DeepMind's first in-house philosopher and associate director of Cambridge's Leverhulme Centre for the Future of Intelligence, joins the show to sort through the hardest question in AI: are these systems conscious, and would we even know? Drawing on his background in animal cognition, Shevlin argues that the consciousness debate has never been settled by a single decisive test, in animals or in machines, but by the slow accumulation of behavioral and mechanistic evidence.
From there, the conversation ranges across Ted Chiang's recent essay dismissing machine consciousness, the "just an X-ism" fallacy, what spiders can teach us about the difference between perceiving the world and caring about it, Anthropic's constitutional AI project, and the uncomfortable historical parallels raised by building minds engineered to serve us. Along the way: reward hacking, the "anthropomimetic turn" in AI design, and why philosophy is suddenly a growth industry at the frontier labs.
Show Notes:“No, Artificial Intelligence Is Not Conscious” Ted Chiang
"Behaviourism's Revenge" Shevlin
"The House Elf Problem" Shevlin
"Contra Chiang" Shevlin
As always, listeners can contact us at prostheticgods@gmail.com with questions, comments, and suggestions for future topics.
Credits:Hosted by James Hughes and Nir Eisikovits
Music & Production by Jake Burley