A profile of UCLA math professor Terence Tao, who was once enthusiastic about AI's role in math but now warns of AI destroying the field's deliberative process

The process for discovering new mathematical problems is at risk of being destroyed by AI.

UCLA math professor Terence Tao, a Fields Medal winner, has shifted his stance on AI in mathematics. Initially enthusiastic, Tao now voices concerns that AI's rapid advancement threatens the core of mathematical discovery. This change in perspective follows recent breakthroughs where AI tools, including OpenAI's GPT-5.2 Pro and Navier-Stokes solver, have successfully tackled complex problems previously unsolved for decades. These include several Erdős problems and the Millennium Prize-winning Navier-Stokes existence-and-smoothness problem. Tao notes that AI-generated proofs often lack detail and require extensive verification by human mathematicians, overwhelming academic journals and disrupting the traditional peer-review process. He argues that the "slow, deliberative process" of working through problems, which is crucial for generating new mathematical insights and formulating new questions, is being destroyed by AI's efficiency.

AI's ability to solve complex, long-standing mathematical problems signals a potential paradigm shift in how mathematical research is conducted. The swift resolution of challenges like the Navier-Stokes problem, once a Millennium Prize contender, suggests AI can overcome human "attention bottlenecks" and retrieve obscure knowledge. However, the deluge of AI-generated proofs, often lacking clarity and requiring extensive human validation, is overwhelming traditional academic structures like journals and peer review.

This disruption poses an existential threat to the generative aspect of mathematics, where new problems and insights emerge from the struggle towards solutions. If AI bypasses this deliberative process, the field risks losing its capacity for organic growth and novel inquiry. Mathematicians like Tao are now advocating for a slowdown to preserve the human element of creativity and deep understanding that has historically driven mathematical progress.