AI solves what humans could not
An artificial intelligence model has solved an 80-year-old geometry problem posed by the legendary Hungarian mathematician Paul Erdős, stunning the math world. The Erdős problem, one of approximately 1,200 unsolved challenges Erdős left behind, had resisted all human attempts at a solution since the 1940s. Researchers found that the AI discovered a more complex arrangement of paired points — one that human mathematicians had never imagined — yielding a larger growth rate than conventional approaches allowed.
Mathematicians call for guardrails
The breakthrough has sparked both excitement and concern. 'This is what AI found, how we missed it, and why it matters,' wrote the Wall Street Journal. But the math community is also urging caution. The Leiden Declaration, signed by leading mathematicians, calls for oversight on AI use in mathematical research, warning that over-reliance on AI could undermine the human creativity at the core of mathematical discovery.
A new era for AI in science
Scientific American called the achievement 'AI's biggest math breakthrough yet.' The solution demonstrates that AI can now contribute to authentic mathematical discovery, not just computation. While some celebrate this as a tool for accelerating research, others fear it could change the nature of mathematical proof itself. The debate mirrors broader conversations across science about where to draw the line between AI assistance and human-led discovery.