The AI Consciousness Trap with Anil Seth

Even if we could build a conscious machine, should we? Neuroscientist Anil Seth says no - and he doubts we can.

This fall, the question of whether AI might be conscious left the philosophy seminar and landed on the business pages. Anthropic has written that possibility into the "constitution" it uses to train Claude, all on the eve of what could be the largest stock offering in history. Days before we spoke, Anil Seth, Ned Block and Giulio Tononi argued in Nature that today's AI is not conscious, and perhaps never will be.

We talk about why a chatbot fools us when a protein-folding model never would; why time, and the stuff we're made of, may matter more than any algorithm; why "Would you like me to do this next?" is the new infinite scroll; and how the idea of a mind in the machine helps keep the AI bubble inflated.

Guest:

Anil Seth - Professor of Cognitive and Computational Neuroscience at the University of Sussex, director of the Sussex Centre for Consciousness Science, and author of "Being You: A New Science of Consciousness."

Website: https://www.anilseth.com/

Being You: https://www.penguinrandomhouse.com/books/566315/being-you-by-anil-seth/

TED2026, "Why AI isn't going to become conscious": https://www.ted.com/talks/anil_seth_why_ai_isn_t_going_to_become_conscious

Nature, "Three reasons why AI is not conscious and perhaps never will be": https://www.nature.com/articles/d41586-026-03313-7

Noema, "The Mythology of Conscious AI": https://www.noemamag.com/the-mythology-of-conscious-ai/

X: https://x.com/anilkseth

Bluesky: https://bsky.app/profile/anilseth.bsky.social

Chapters:

00:00 Cold open & intro

01:54 Meet Anil Seth

02:45 The Nature comment and the Behavioral and Brain Sciences paper

04:31 Why talking machines fool us

08:19 The protein-folding test

12:21 Time vs. sequence: why a brain can't be paused

14:27 How did we get to be conscious? Animals, evolution and human exceptionalism

18:05 Beyond the brain: nerves, bodies and gut feelings

19:31 Intelligence isn't consciousness, and Descartes' long shadow

22:46 Useful models vs. the real brain

25:16 Scaling, energy and data efficiency

28:23 Learning from bats

29:18 Looking back in 10 years: deep mirrors

32:00 Demoting our own humanity

35:48 Lessons from the PC hype, and the erosion of science

38:39 "Would you like me to do this?" The new infinite scroll

40:19 Power users, Claude Code and neurosymbolic AI

43:30 How would we know if AI were conscious?

46:41 "Functional pain vectors"

49:25 Substrate, analog computing and computational functionalism

54:02 The rabbi's question: should we build conscious machines?

57:00 The costs of false caution: AI rights and parasocial risk

59:43 The Anthropic IPO and the AI bubble

01:01:38 Claude's constitution and the myth of agency

01:04:00 Local models, open weights and interesting times

01:05:43 Outro & links

Follow the Anthropic IPO, the 2026 midterms and more on the America 2.0 Live Trackers: https://trackers.america2.news/

Hear them daily on the America 2.0 Tracker Briefing: https://podcasts.apple.com/us/podcast/id6825877339

Dave Troy Presents is published by America 2.0. https://america2.news