Yudkowsky: AI fiction lags code badly — 'we may only be alive because gradient descent generalizes policies so poorly'
sebkrier · x · 2026-11-10
Pinboard's founder mocks Eliezer Yudkowsky's ongoing doom rationalizations, but the quoted tweet carries the substance: Yudkowsky observes AIs are much worse at fiction than code — constantly losing track of which facts depend on which other facts and fixing conflicts the wrong way. He concludes we may only still be alive because gradient descent on neural networks is such a bad way to generalize over policies.
More from AGI Musings
- Frontier lab compute shifts hard: pre-training falls from 67% to 7%, post-training/RL now 55% — akbirthko · 2026-11-10
- Why the AI Race Discourages Guardrails: Even CoT Monitoring Slows You Down — burny_tech · 2026-11-10
- Researcher argues LLM generalization and understanding is a spectrum, not binary — burny_tech · 2026-11-10
- LeCun clarifies PRH hype as researchers decry polarized AI video titles — burny_tech · 2026-11-10
- Study of 1,600 Data Centers Finds Each Adds ~1,000-1,300 Local Jobs Within 3-5 Years — RachelVT42 · 2026-11-10
- Claude builds a 500,000-tile mosaic engine, sparking the 'tasteslop' debate — teortaxesTex · 2026-11-10