Richard Sutton: Brute-Force Compute Isn't Enough, and the Bitter Lesson Says So
PMinervini · x · 2026-11-08
Turing Award winner Richard Sutton pushed back on brute-force scaling: 'Brute force computation is not enough for intelligence. Clever new methods are needed. The science of AI is far from finished.'
Graham Neubig amplified the post while mocking the replies: many people who treat the Bitter Lesson as gospel have apparently never read the original essay — which, if anything, is an argument for RSI (recursive self-improvement), not for brute-force scaling. The exchange exposes a widespread misreading of the Bitter Lesson as a 'just add compute' mantra that Sutton himself explicitly rejects.
More from AGI Musings
- Humanity's Greatest Achievement May Be an Intelligence That Surpasses Us — VraserX · 2026-11-10
- 430,000 AI Micro-Dramas in 8 Months: A Visitor's Take on China's AI Content Boom — JacquesThibs · 2026-11-10
- FOCS 2026 Grad Student Feedback: US Students More Optimistic, Many Feel No Agency — thegautamkamath · 2026-11-10
- Reddit essay: most of what people hate about AI comes from the humans steering it — teleport66 · 2026-11-10
- Underrated question: what do the diverse forms of recursive self-improvement look like? — repligate · 2026-11-10
- 'Bitter lesson for robotics' gets a sharp rebuttal: it's still human-orchestrated end to end — burny_tech · 2026-11-10