AI Solves 372 Major Math Problems in Single Day

A new AI just accomplished what would take human mathematicians decades to achieve, solving hundreds of major math problems in one sweep.

OpenAI released 372 mathematical proofs on Monday, each resolving or making major progress on important unsolved questions in mathematics and computer science. The results include advances on the four-dimensional Kakeya conjecture, improvements to vital computer algorithms, and even steps toward the notoriously difficult Riemann hypothesis.

What makes this moment remarkable isn't just the volume of breakthroughs. It's how they happened.

The company says nearly all 372 solutions came from a single AI agent responding to one prompt. That's a dramatic shift from last month's celebrated Navier-Stokes solution, which required 10,000 AI agents working together at a cost of millions of dollars in computing power.

Most of the proofs have already been verified using Lean, a programming language that checks mathematical logic. That means they're almost certainly correct, even if mathematicians don't yet fully understand how the AI reached its conclusions.

The speed is both thrilling and unsettling for the math community. OpenAI's own mathematicians haven't had time to understand many of the results they just published. The company acknowledges the pace feels overwhelming but says it can't slow down because solving hard math problems proves their AI is genuinely getting smarter.

Why This Inspires

This technology could democratize mathematical discovery in ways we've never seen before. If a single affordable AI agent can solve problems that have stumped brilliant minds for generations, that power won't stay locked in expensive labs forever.

Daniel Litt, a mathematician at the University of Toronto, sees the upside clearly. "If we want to know the answers to these math questions, I see no reason why we should ask the company to keep them secret from us," he says.

The company is working to release this model to the public as quickly as possible. When that happens, students, researchers, and curious minds everywhere could gain access to a mathematical assistant more powerful than anything that's existed before.

Some mathematicians worry about AI companies flooding their field with unexplained proofs faster than humans can process them. Others question whether these proofs contain genuinely new ideas or just remix existing techniques. Fair concerns, but the solutions are real and the math checks out.

An independent advisory group recommended that companies releasing AI-generated math should share their models, prompts, and computing times publicly. OpenAI isn't following all those guidelines yet but says it's working toward greater transparency.

What started as a trickle of AI-assisted mathematical discovery has become a flood, and the tools behind it are getting cheaper and more accessible every day.