AI Solves 80-Year Math Problem Humans Couldn't Crack

A computer just cracked an 80-year-old math puzzle that has baffled the world's brightest minds since World War II ended.

OpenAI announced this week that one of its AI models disproved a long-held belief about how to arrange points on a flat surface. The problem, first posed by legendary mathematician Paul Erdős in 1946, asked a deceptively simple question: if you place points on a plane, what's the maximum number of pairs that can sit exactly one unit apart?

For nearly eight decades, mathematicians assumed square grids were the answer. Everyone's intuition pointed the same direction, and no one could prove otherwise.

Then the AI looked at it differently. Without any preconceived notions about grids, the model explored combinations humans never thought to test, pulling together ideas from different areas of mathematics that had been sitting in academic papers for years.

The result shocked experts. The AI found an infinite family of point arrangements that beat the grid approach, disproving what researchers had believed was true.

Here's what makes this special: the AI didn't just spit out an answer and walk away. It produced something strange and promising that elite mathematicians could actually work with. A team of nine top researchers then checked the proof, simplified it, connected it to existing theory, and confirmed it was legitimate.

One noted researcher called the result "a milestone in AI mathematics." The Guardian reported that while the larger problem remains unsolved, this breakthrough represents significant progress.

Why This Inspires

This isn't about computers replacing human creativity. It's about partnership. The AI served as what one might call a master curator, connecting dots across different mathematical fields that humans had scattered but never quite assembled.

The companion paper, written by nine distinguished mathematicians including Noga Alon and Timothy Gowers, describes it as a "short, digested, human-verified version" of the AI-generated discovery. The humans did what their field demands: they checked it, refined it, and explained what it really means.

This represents a fundamental shift from AI merely solving known problems to helping explore truly open questions. Previous AI achievements in mathematics focused on competition problems where answers were already known. This time, nobody knew what the right answer looked like.

The implications stretch beyond mathematics. When humans and AI work together, combining machine pattern-recognition with human insight and verification, they can tackle challenges that seemed permanently stuck.

Mathematics has always progressed when techniques from one field get applied cleverly to another. Now we have a powerful new tool that can spot those connections faster than any human could scan the literature.

The future of discovery might look exactly like this: machines finding unexpected paths, humans making sense of them, and both pushing the boundaries of what we thought was possible.