TypeSafe AI's 'Bitterest Lesson': picking the right task beats data, compute and algorithms
ZeYanjie · x · 2026-09-20
A TypeSafe AI blog post extends Sutton's bitter lesson one layer further: the true ordering is right task > data > compute > algorithms. Games make the bitter lesson visible because the task is obvious and self-play generates endless data; in the real world someone still has to choose the objective, which requires studying users, products and organizations outside ML itself. Pick the wrong task and you get beautiful loss and scaling curves for a useless model. ML research, the author argues, attacks these problems in exactly the reverse order.
Related event: The Bitterest Lesson: Picking the Right Task Beats Data and Compute(2 posts)→
More from Research
- Gradient-free weight editing cuts VLM FigStep attack success from 49% to 2% — lateinteraction · 2026-11-04
- Anthropic says it may imminently hit RSP threshold for AI-accelerated R&D, releases provable inference work — tobyordoxford · 2026-11-04
- Google scanned 1,033 real household products, packed into a 345 MB GLB for three.js — victormustar · 2026-11-04
- These 10 GPU optimization papers are why your model runs fast today — NandoDF · 2026-11-04
- INT8 isn't the problem: I benchmarked samplers on INT8 flow models — YakovSava · 2026-11-04
- Karpathy Boosts Open Hacker News Dataset: 10M+ Posts Usable as a Personal Knowledge Base — huangyun_122 · 2026-11-04