Mathematician: LLMs matter for massive proofs, not short insight papers
littmath · x · 2026-11-02
A mathematician (amplified by Scott Little) on where LLMs actually matter in mathematics:
- Not needed: short papers built around one insightful idea — high-impact but human-writable — or technical work we already do, where LLMs just offer speedups (e.g., proving side lemmas).
- Qualitative difference: problems requiring extraordinarily complex proofs beyond what individuals or even teams can reasonably produce.
- Current bottleneck: LLMs struggle to translate very long Lean formal proofs into understandable mathematical arguments; better training should help.
- He poses a pointed question: given a proof of the Riemann Hypothesis but no number theory training, how would you actually verify and understand it?
More from AGI Musings
- The widest gap now is between tech and non-tech workers, says AI researcher — emax · 2026-11-02
- Ethan Mollick runs a 'reverse CAPTCHA': AIs swarm to Oxyrhynchus Papyri replies — eldonredwards · 2026-11-02
- Ethan Mollick: GPT-4 could work solo for 4 minutes; today's models run for weeks — eldonredwards · 2026-11-02
- Robotics isn't 'solved' by benchmark jumps — it's solved when GDP grows 100x — eigenron · 2026-11-02
- DeepSeek researcher: 'Don't make any mistakes' is the underrated moat — Yuchenj_UW · 2026-11-02
- Martin Ford vs Pedro Domingos debate whether AI will take most of our jobs — MFordFuture · 2026-11-02