The Bitter Lesson Re-read: Scale Beats Hand-crafted Features, Not Expertise
anshulkundaje · x · 2026-11-03
A widely shared thread reinterprets Sutton's bitter lesson: its claim is simply that the incremental yield of scale outpaces hand-crafted features — not that brute force solves everything or is even the dominant approach.
The author notes the recent math proofs from OpenAI and Anthropic weren't produced by millions of agents run by amateurs, but by world-class mathematicians like @alpoge. Scale benefiting from well-aimed expertise is fully compatible with the bitter lesson — which is why the "scale vs. experts" framing is a bad question; the real answer is that scale's compounding requires expertise to aim it.
More from AGI Musings
- AlphaFold Predicted 200M Protein Structures in Two Years, Ending 50 Years of Manual Work — r0ck3t23 · 2026-11-03
- Essay: AI Agents Are Breaking the Internet's 30-Year-Old Business Model — Ronangmi · 2026-11-03
- Bot traffic surpasses humans: essay argues AI agents are breaking the open web's 30-year business model — Ronangmi · 2026-11-03
- AI optimist concedes doomers were right on cyber risk — maybe extinction risk too — harris_edouard · 2026-11-03
- AI safety debate: unilateral US pause, China dynamics, and SFBay holding humanity's leash — dhadfieldmenell · 2026-11-03
- LLMs solving math gets all the hype, but their new infrastructure patterns for reliable software go unnoticed — StewartalsopIII · 2026-11-03