Harvard & MIT Paper: LLMs Fall Short of Real Scientific Discovery
gerardsans · x · 2026-08-02
A joint paper by Harvard and MIT researchers argues that LLMs are nowhere near capable of real scientific discovery. The researchers built a new evaluation framework called SDE, which tests models in open-ended research projects rather than static multiple-choice quizzes. Their findings suggest that recent claims from tech labs about AI achieving scientific breakthroughs are significantly overstated.
More from Research
- One ancient fold, many jobs: how the immunoglobulin domain evolved into antibodies, MHC and titin — arjunrajlab · 2026-10-16
- Universal forgery attack breaks Chinese PQC signature Origami using only the public key — matthew_d_green · 2026-10-16
- How do you attribute agent gains: model training or infrastructure? — jrxna · 2026-10-16
- First-ever paper on multi-agent swarm safety lands NeurIPS spotlight — LauraRuis · 2026-10-16
- EDEN: anchored embedding noise boosts RLVR exploration beyond policy rollouts — Daijiao Liu · 2026-10-16
- Autonomous vehicle demo shows car reversing to break narrow-road deadlock, with a robotic arm to shift gears — sanjeevs_iitr · 2026-10-16