Yudkowsky admits 'recursive self-improvement' and AGI ideas seeded the AI race
mjdramstead · x · 2026-10-14
In a candid thread, Eliezer Yudkowsky acknowledges that many 'smart-sounding' concepts in AI company discourse — recursive self-improvement, AGI itself — trace back to his writing or diffusion via Bostrom and others. He says Amodei, Altman and many employees got these ideas downstream of his work, without independently studying the underlying intellectual history. He sketched positive visions of AGI/ASI in the mid-2000s but deprioritized writing them up. Commenters note this confirms his integral role in launching the ASI race.
More from AGI Musings
- Cuban rebuts Andreessen on why housing, healthcare and education keep getting pricier — 2C_ornot2C · 2026-10-14
- LLM Reasoning Is Just Extensive Search — Experiment Time Remains the Bottleneck — JFPuget · 2026-10-14
- Anil Seth: Today's LLMs Are Very Unlikely to Be Conscious — AI That Seems Conscious Is the Real Risk — anilkseth · 2026-10-14
- Tufts and Santa Fe Institute paper models LLM adoption as a 'cognitive virus' — alex_verem · 2026-10-14
- From a Utah datacentre fight to citizens' assemblies: how to actually democratise AI — nordicinst · 2026-10-14
- "In 2 years nobody will ask what you do for work": another aggressive AGI timeline take — rand_longevity · 2026-10-14