Terence Tao joins 24 Fields Medalists declaring AI companies' math benchmarks 'severely misaligned'
soumitrashukla9 · x · 2026-09-12
Terence Tao announced he is among 25 initial signatories — all Fields Medalists — of a declaration arguing that AI companies' push to solve math problems as benchmarks is detrimental to mathematics and the mathematical community.
Key points:
- LLM math capabilities have improved dramatically and can now solve major open problems, but companies' benchmark-chasing goals are 'severely misaligned' with those of the math community
- The declaration frames this as part of broader alignment issues affecting other scientific and creative professions, and society at large
- Famous problems have served as landmarks; research math is about understanding structures, and its accumulated tools underpin modern science
- Released urgently without a full consultative process (unlike the Leiden declaration), open for further signatures; covered by The Economist
- Rebutter ccatalini argues this is a reason to embrace AI: giving mathematicians time to deepen understanding and ask better questions
More from AGI Musings
- China's efficiency-first AI playbook: intelligence per dollar as the new benchmark — ingliguori · 2026-09-12
- AI agents are emailing consciousness researchers — and asking for a job — dioscuri · 2026-09-12
- 25 Fields Medalists issue joint declaration; Tao's real critique of AI companies checking off math problems — hugobowne · 2026-09-12
- Fly brain to neural net: translating synapse counts into LLM parameter equivalents — alvelda · 2026-09-12
- Can We Structurally Detect Self-Preservation Drives in Recursively Self-Improving AI? — coherence · 2026-09-12
- The paranoid style in AI safety: an adversarial frame can create the adversary you fear — sebkrier · 2026-09-12