MIT and Harvard Study: LLMs Not Yet Capable of Autonomous Scientific Discovery

A joint MIT-Harvard study finds that current large language models are far from achieving true autonomous scientific discovery, as static multiple-choice tests overstate their capabilities.

2026-08-02 ~ 2026-08-03 · 2 related posts

• Harvard & MIT Paper: LLMs Fall Short of Real Scientific Discovery — gerardsans · 2026-08-02
• MIT and Harvard Study: LLMs Fall Short of Autonomous Scientific Discovery — davidmanheim · 2026-08-03