On October 11, the "Yau Mathematics and Artificial Intelligence Laboratory Alliance" was founded at Tsinghua University. Beijing News, The Paper and Sina Finance all described it as a thousand-researcher-scale push to fuse mathematics with AI; Eastmoney added that Fields medalist Yau Shing-Tung is leading eight institutions to co-develop math-specialized models, with a related laboratory to be set up in Shanghai.

"The Impact Is a Tsunami — Learn to Surf, Don't Hide"

Facing a wave of AI-generated mathematical results, Yau chose to fight back with institutions. Beijing News quoted his position: "If AI is to advance, mathematicians cannot be absent." In Sohu's coverage he compared the AI shock to a tsunami — hiding leads nowhere; one must learn to surf. The alliance turns those words into structure: joint laboratories sharing problems, corpora and compute, turning mathematicians from bystanders into builders of infrastructure.

The Same Week, Terence Tao Declared "Math 2.0"

On October 11, QbitAI reported that another Fields medalist, Terence Tao, formally proclaimed the arrival of the Math 2.0 era. His evidence maps exactly onto the week's news: OpenAI released 722 AI mathematics papers overnight, with progress on 377 open problems, and the Kakeya conjecture solved this year by Wang Hong was extended by AI to four dimensions. Math 1.0 was built on proof scarcity; as proofs flood in, the evaluation system must be rebuilt. Tao also led a joint statement demanding that AI results disclose failures and compute costs.

Math × AI: The Week in Brief

· Oct 7: OpenAI posts 722 AI math papers; progress on 377 open problems

· Oct 11: Terence Tao declares the "Math 2.0 era" has arrived

· Oct 11: Yau Math-AI Laboratory Alliance founded at Tsinghua — eight institutions, math-specialized models, a Shanghai lab

Two Fields medalists share the same verdict: the question is no longer whether to live with AI, but on what terms.

Why "Math-Specialized Models"

Tao has summarized the three conditions that make mathematics most exposed to AI: proofs can be objectively verified, reasoning is fully digitizable, and high-quality literature is already online. This also explains why general-purpose models are not enough — theorem proving demands rigorous formal verification and search strategies that generic capability cannot replace. By organizing scattered laboratories, the alliance is essentially building public infrastructure for the mathematical community in the AI era.

The lesson for industry is just as direct: the higher-value the scenario, the more it needs the combination of "specialized models + verifiable evaluation." NineZenith applies the same principle when landing Zenith-Act industry models — vertical scenarios and measurable outcomes first, not generic benchmark chasing.

(Compiled from public reports by Beijing News, The Paper, Eastmoney, Beijing Daily (jbnews), Sohu and QbitAI)