Szepesvári Defends Mathematicians' Open Letter: Target Is AI Benchmark Culture, Not AI Itself
A discussion mixing clarification and disagreement unfolded between RL researcher Csaba Szepesvári and Quanquan Gu over the mathematics community's open letter on AI. Szepesvári pointed out that the public misread the letter as "AI is bad, don't use AI for math," when its actual position is the opposite: it explicitly affirms AI's potential, and its only demand is that AI companies stop using the solving of major open math problems as benchmark-marketing stunts. In his view, it is the first open letter that hasn't slid into the "anti-AI" template.
Confirmed
- Responding to Gu, Szepesvári made clear the letter never said AI is bad, nor asked people to stop using AI to solve math problems; its criticism targets one thing only—AI companies' benchmark culture.
- Szepesvári predicts that important open math problems usable for marketing will soon run out, at which point AI companies will naturally stop, though the sooner the better.
- Gu replied that he doesn't fully agree with the demand that AI companies stop treating major math problems as benchmarks, narrowing the disagreement to that specific ask.
Why It Matters
- The discussion clarifies the letter's real demands, helping prevent the misreading of "mathematicians against AI" from spreading.
- Szepesvári's judgment that "marketable math problems will eventually run out" touches on the sustainability of AI companies using academic challenges for publicity—an incremental insight that goes beyond the open letter itself.
2026-09-13 ~ 2026-09-13 · 5 related posts
- Episode 1: 25 Fields Medalists Sign Open Letter Warning of AI's Severe Misalignment in Mathematics(2026-09-12, 93 posts)
- Episode 2: Fields Medalists' Anti-AI Letter Criticized as Self-Serving Rationalization(2026-09-12, 2 posts)
- Episode 3: After AI Solves Math, What's Left for Humans? X Users Debate(2026-09-12, 23 posts)
- Episode 4: Pachter Refuses to Sign Fields Medalists' AI Letter, Blames Math Community(2026-09-13, 3 posts)
- Episode 5: Math researcher: neither AI firms nor math community prioritize understanding(2026-09-13, 2 posts)
- Episode 6: Szepesvári Defends Mathematicians' Open Letter: Target Is AI Benchmark Culture, Not AI Itself(2026-09-13, 5 posts)
Primary sources
- Szepesvári clarifies the mathematicians' letter: it targets benchmark culture, not AI itself — CsabaSzepesvari ·
- Csaba Szepesvari says AI labs' marketing-value math benchmarks will soon run dry — CsabaSzepesvari ·
- Quanquan Gu: not convinced AI labs should stop treating math problems as benchmarks — QuanquanGu ·
- Szepesvári on the mathematicians' letter: benchmark culture, not AI itself — CsabaSzepesvari · 2026-09-13
- [source] Szepesvári clarifies the mathematicians' letter: it targets benchmark culture, not AI itself — CsabaSzepesvari · 2026-09-13
- Szepesvari on mathematicians' AI letter: it targets benchmark culture, not the people — CsabaSzepesvari · 2026-09-13
- [source] Quanquan Gu: not convinced AI labs should stop treating math problems as benchmarks — QuanquanGu · 2026-09-13
- [source] Csaba Szepesvari says AI labs' marketing-value math benchmarks will soon run dry — CsabaSzepesvari · 2026-09-13