Dan Williams: "The Crazy AI People Are Directionally Correct" — on the limits of AI scepticism
dioscuri · x · 2026-11-23
Philosopher Dan Williams publishes "The Crazy AI People Are Directionally Correct" on Conspicuous Cognition, dissecting the "superintelligence perspective" and the limits of responsible AI scepticism. Shared by Dioscuri.
The essay decomposes the perspective into three claims:
- Imminent recursive self-improvement: within a few years AI will automate most AI-building work, creating an accelerating feedback loop;
- Imminent superintelligence: this quickly yields systems far beyond the best humans at nearly all cognitive tasks;
- Imminent transformation: unlike electricity or the internet, superintelligence could compress a century of progress into a decade.
Williams argues the view can't be dismissed as charlatanism — many very smart people with strong forecasting records take it seriously, and it rests on a technology that verifiably works — and examines where scepticism should and shouldn't stop.
More from AGI Musings
- Papers on AI timelines are growing super-exponentially, now doubling every 3.8 months — peterwildeford · 2026-11-23
- OpenAI Chief Scientist: Avalanche of Pretraining Breakthroughs Possible by 2027 — Neurogence · 2026-11-23
- Unreleased frontier AI formalizes Fermat's Last Theorem proof, fells Millennium Problems, roils mathematicians — Afinetheorem · 2026-11-23
- The pragmatic path: gen AI for rendering only, before touching game state — gandamu_ml · 2026-11-23
- Elliptic-curve discrete log already has a poly-time solution — and could an LLM have found it? — jamestagg · 2026-11-23
- AI-native org math upends tech staffing as Microsoft offers exits to 60,000 — thedealdirector · 2026-11-23