An 80-Year-Old Question Answered
In May 2026, the mathematics and AI communities were electrified by news that a problem first posed in 1946 had been resolved. The 'unit distance problem,' a famous conjecture from the legendary mathematician Paul Erdős, had resisted all attempts at a solution
for eight decades. The question itself seems simple enough: what is the maximum number of points you can place on a flat surface so that each point is exactly one unit of distance away from a specific number of other points? Yet finding a definitive answer proved fiendishly difficult. That is, until an advanced AI model from OpenAI produced a proof that has since been verified by leading human mathematicians, marking what many are calling a historic milestone.
The AI That Thinks Like a Mathematician
Unlike earlier AI systems that relied on brute-force computation or were trained on specific mathematical domains, this breakthrough came from a general-purpose reasoning model. Instead of just crunching numbers, the AI was able to explore complex proof strategies, have original ideas, and carry them to a logical conclusion—a process previously thought to be uniquely human. The model autonomously generated a complete proof, which was then shared with external mathematicians for review. This wasn't a case of a computer simply spitting out an answer; it was a demonstration of an AI engaging in the kind of abstract reasoning that is central to advanced mathematics.
Human Verification: The Crucial Final Step
The AI's proof, while valid, was not the end of the story. It was the beginning of a new kind of collaborative science. A team of prominent mathematicians, including Fields Medalist Timothy Gowers, was tasked with reviewing and digesting the AI's work. They confirmed the result was correct, hailing it as a landmark achievement. Furthermore, the human experts were able to refine, improve, and explain the context around the AI's logic, creating a richer and more understandable proof than the AI had produced alone. This crucial step underscores that the future isn't about AI replacing researchers, but about creating a powerful partnership where machine-scale discovery is guided and validated by human expertise and intuition.
A New Golden Age for Discovery?
The successful resolution of the unit distance problem is not an isolated incident. It is part of a rapidly accelerating trend. Across fields like knot theory and number theory, AI is helping researchers find patterns and connections that the human mind might not easily spot. In some cases, AI is helping to make progress on conjectures that have been unsolved for decades; in others, like the unit distance problem, it is delivering the final proof. Researchers believe this combination of human ingenuity and AI's ability to process vast, complex data could herald a 'golden age' of mathematics and science. The AI acts as a powerful guide, pointing mathematicians toward promising avenues of inquiry that they can then explore with their deep theoretical understanding.
What Comes After Unsolvable Problems?
With one of Erdős's famous problems now solved, the scientific community is buzzing with possibility. What other long-standing challenges might now be within reach? The event has proven that AI can be a genuine partner in frontier research, not just a tool for analyzing data. In fact, inspired by the AI's technique for the unit distance problem, human researchers were able to adapt the method to solve another significant conjecture just a week later. This rapid-fire progress demonstrates a powerful new feedback loop: AI makes a breakthrough, humans learn from its novel approach, and then apply that new knowledge to solve even more problems. The history of science may one day be divided into two eras: before and after we had AI to help us think.
