算力便宜。選對工作不便宜。

作者: Diogo Almeida

2026-09-10

The Bitterest Lesson

Complete Skeptic

把巨大模型訓練在錯的工作上,它就會把錯的工作做得很好。Diogo 的重點是:還是得有人為工作命名。Classify 也從這裡開始。Jev 該決定什麼?

The Bitterest Lesson

TL;DR: Compute drives progress in AI, but what good is progress if you are not doing the right task!

Rich Sutton's bitter lesson states that compute beats algorithms.