Steven Pinker on AI Doom and Consciousness: Why He Thinks Doomerism Fails
sapinker · x · 2026-10-08
Conspicuous Cognition releases a long interview with Steven Pinker, recorded amid his public spat with Scott Alexander over AI existential risk: Alexander publicly challenged him to a debate, Pinker declined with an open letter arguing doom scenarios rest on confused assumptions about intelligence, motivation, and engineering, and Alexander later published a lengthy rebuttal.
Key points:
- Pinker argues intelligence is not a superpower and intelligence alone doesn't confer dangerous agency.
- He thinks AI systems (probably) aren't conscious and that extinction scenarios don't hold up.
- The conversation ties his AI views to decades of work on intelligence, motivation, evolution, and progress.
- Hosts press him with objections, several drawn from Alexander's latest response, and ask what evidence would change his mind.
The episode also notes where Pinker was wrong about neural networks. Full transcript available.
Related event: Steven Pinker Pushes Back on AGI Doom Scenarios(2 posts)→
More from AGI Musings
- AI Bottlenecks: a 2am side project now read by investors, semis and AI labs — demian_ai · 2026-10-08
- OpenAI preprint claims matrix multiplication exponent drops to 2.25, biggest leap since 1979 — BorisMPower · 2026-10-08
- Math was like Go until brute-force compute won — now mathematicians should ask better questions — xwang_lk · 2026-10-08
- Taste isn't the differentiator — just throw more tokens and intelligence at it — zebird0 · 2026-10-08
- Ugly, Brute-Force Solutions Win: AI Circle Debates What Our Taste for Elegance Cost Us — shakoistsLog · 2026-10-08
- tszzl: Training models to do AI R&D is far harder than training them to do math — tszzl · 2026-10-08