Fields Medal winner says AI could beat humans in proof work within two years
新智元 · wechat · 2026-07-24
A new WeChat report links a Fields Medal winner’s career decision to AI’s rapid progress in mathematics.
- Jacob Tsimerman says he expects AI to surpass humans in all proof work within about two years and is moving into AI safety at OpenAI.
- He argues AI has already doubled his productivity and could make PhD training obsolete by the time current students graduate.
- The piece frames this as a sign that theorem proving, long seen as a human stronghold, may be closing faster than many expected.
Related event: Fields Medalist Warns AI May Surpass Humans in Math Within Two Years(3 posts)→
More from AGI Musings
- Playing games while an agent works now feels oddly productive — dotey · 2026-07-24
- A post argues AGI could arrive within 1-2 model generations, maybe by late 2027 — haider1 · 2026-07-24
- AI’s biggest workplace shift may be accountability, not creativity — krishnan · 2026-07-24
- The real trust measure for an agent is how much compute it controls — code_star · 2026-07-24
- A new Asimov-style AI code urges liminality, cross-disciplinary research, and creative risk — begusgasper · 2026-07-24
- AI cybersecurity needs real incentives, not first-principles guesses — sebkrier · 2026-07-24