Peking University mathematician: AI's millennium-prize wins are 'harvesting,' not discovery
新智元 · wechat · 2026-10-03
In an interview with 新智元, Peking University mathematician Li Xinyi argues OpenAI/Anthropic's claims of solving math conjectures are PR-driven 'trophy hunting': results get 'solved but never understood.' He reveals a postdoc's unpublished draft was nearly reproduced verbatim on arXiv after discussions with an early GPT-5.x build, and says an unsolved conjecture of his for 20 years fell instantly to an LLM using its proposer's old method—evidence AI is mining human knowledge, not creating. He estimates the Navier-Stokes effort burned $20M in tokens, enough to hire 100 top mathematicians for a year, and insists credit and funding must go to humans contributing understanding, echoing the open letter by 25 Fields medalists including Terence Tao.
More from AGI Musings
- Family offices are building in-house AI research systems, turning analysts into end-to-end agent builders — virattt · 2026-10-03
- AI moral circle debate: consciousness with inner life as the entry criterion — dioscuri · 2026-10-03
- Pope Leo XIV: AI-generated images and human art differ ontologically, not just aesthetically — bennash · 2026-10-03
- ML researcher: training on your emails gives a lossy token approximation, not you — gerardsans · 2026-10-03
- RIP traditional data scientists: the role is splitting into two tracks — mdancho84 · 2026-10-03
- David Patterson: solar's near-vertical cost curve will power the singularity — davidpattersonx · 2026-10-03