Terence Tao Explains the Math Behind AI¶

Source: YouTube · OpenAI Forum · Terence Tao (UCLA) & Mark Chen (OpenAI CRO) · March 2026

Overview¶

Fields Medal recipient Terence Tao, widely considered the world's greatest living mathematician, explains how AI is changing the way mathematicians work. The interview was part of an OpenAI Forum event held in partnership with IPAM (Institute for Pure and Applied Mathematics). OpenAI Chief Research Officer Mark Chen also discusses OpenAI's broader goal of building tools that help scientists achieve breakthroughs at scale.

Key Topics¶

1. AI as a Research Assistant¶

Tao describes how AI tools are making it easier for mathematicians to: - Experiment rapidly — test conjectures and explore new ideas without full manual proofs - Explore new ideas — AI can suggest paths that a human might not consider - Collaborate — AI bridges gaps between mathematicians working on different aspects of a problem

2. The Erdős Problems Breakthrough¶

Tao served as the adjudicator for a series of Erdős Problems solved with generative AI (including ChatGPT). These are over 1,000 mathematical questions set forth by Paul Erdős. The AI-written proofs represent a landmark moment — not because AI solved the hardest problems, but because it demonstrated it could handle the "long tail" of obscure problems that mathematicians haven't gotten to.

3. AI and Mathematical Understanding¶

Tao and others (including Akshay Venkatesh of IAS) note a risk: as AI becomes more powerful, mathematicians may lose direct experience with mathematical understanding — the "muscle memory" of proving things. The tension between AI as a tool and the preservation of human mathematical intuition remains an open question.

4. The AI Revolution in Math¶

Assessments from Quanta Magazine and The Atlantic confirm: "2025 was the year when AI really started being useful" for mathematical tasks. The combination of machine learning insights with the precision of mathematics is seen as a path toward general intelligence.

Key Takeaways¶

AI is changing mathematics by enabling rapid experimentation and collaboration
Tao served as adjudicator for AI-solved Erdős Problems — a milestone in AI mathematics
The risk: mathematicians may lose direct proving experience as AI takes over routine work
AI tools preserve human creativity and curiosity while expanding what's possible
OpenAI is building toward tools that help scientists achieve breakthroughs at scale