OpenAI's AI systems have begun solving long-standing unsolved math problems, prompting intense debate within the mathematical community about what this means for the discipline itself.
The breakthrough came when OpenAI refuted the Unit Distance Conjecture, a problem mathematicians had struggled with for years. This success triggered broader progress. Fields Medal winner Timothy Gowers reports that GPT 5.6 Pro solved two problems he personally spent considerable time on, each solved on the first attempt.
But Gowers raises a serious concern. He warns of the "possible destruction of mathematical culture" if mathematicians stop developing the expertise needed to understand these AI-generated results. The danger lies not in solved problems, but in a potential hollowing-out of mathematical knowledge. If researchers outsource problem-solving to AI without building deep understanding, the field loses its foundation.
The mathematical community splits on interpretation. Some view AI as a productivity tool that accelerates discovery. Others see a threat to how mathematics develops. Understanding how a problem gets solved matters as much as the solution itself. That process builds intuition, reveals connections, and trains the next generation of mathematicians.
The practical question becomes immediate. When an AI solves a problem in seconds that took humans years, what happens to the incentive to learn the underlying theory? Does a mathematician need to verify an AI's proof line-by-line, or can they trust the output? How do you publish results that AI generated but you don't fully understand?
These aren't abstract concerns. Mathematical breakthroughs often create new fields or methods with applications far beyond the original problem. If the problem-solving process gets abstracted away into black boxes, those secondary benefits may never materialize.
The real issue isn't whether AI should solve math problems. It's whether the mathematics profession can integrate these tools without abandoning the human expertise that makes mathematics a coherent body of knowledge rather than a collection