AGI Debate: Singularity Arrived Two Years Ago, But AGI Still Needs a Society of Mind

karlwaldman · x · 2026-09-14

A threaded debate on AGI and regulation: Devon Eriksen argues AI labs calling for regulation really want a regulatory moat that burdens open-source competitors; ESR agrees but contends the 'singularity' (collapse of predictive horizons) actually passed two years ago when LLMs got good. His key point: we crossed the front wall without AGI—simple statistical inference engines proved strong enough to solve Millennium Prize Problems, but true AGI likely requires a society-of-mind architecture with many specialized subsystems we can't yet build.

Related event: Debate Rages: Are AI Labs' Regulatory Calls Self-Protection or Regulatory Capture?(11 posts)→

Original post →

More from AGI Musings

AGI Musings channel →