May 20, 2026
Proof, Panic, and Petty Takes
An OpenAI model has disproved a central conjecture in discrete geometry
AI just broke a famous math puzzle, and the comments instantly turned into a vendor war
TLDR: OpenAI says one of its AI models cracked a famous old math problem and outside experts checked the proof, a big sign that AI may now help make real discoveries. In the comments, people instantly argued over whether this proves OpenAI is crushing its rivals or whether everyone is getting carried away.
A famously stubborn math puzzle that has annoyed experts for nearly 80 years just got wrecked by an OpenAI model, and honestly, the comment section is treating it like the season finale of tech Twitter. The official news is huge: OpenAI says one of its general reasoning models found a proof that knocks down a long-believed idea about arranging points on a flat surface. Translation for normal humans: a very old, very respected math belief may have just been proven wrong by AI, and outside mathematicians checked the proof. That alone is headline material. But the real fireworks came when readers immediately turned it into a three-way showdown over which AI company is winning the brains race.
One commenter flat-out declared OpenAI has a “distinct lead in academics” over Anthropic and Google, basically lighting the fuse on an instant vendor-war debate. Another jumped in to stress that this wasn’t done with some special math-only setup, which fans treated like a flex: not a niche calculator, but a broader model pulling off something elite. Then came the classic internet clapback energy, with one commenter scolding doubters that their understanding of AI is “out of date” and aging badly by the minute. Meanwhile, the most delightfully human response was someone simply asking for a picture of the strange geometric construction, because yes, after all this world-changing genius, people still want the diagram.
And the mood? A mix of awe, skepticism, and “okay, maybe I was too dismissive.” Even the cautious voices admitted AI may not solve all science overnight, but this result made it a lot harder to laugh off.
Key Points
- •OpenAI says one of its internal models disproved a longstanding conjecture related to the planar unit distance problem first posed by Paul Erdős in 1946.
- •The article states that the previous prevailing belief held square grid constructions to be essentially optimal for maximizing unit-distance pairs.
- •According to OpenAI, the model produced an infinite family of examples that achieves a polynomial improvement over those constructions.
- •OpenAI says the proof came from a general-purpose reasoning model rather than a system specialized for mathematics or the specific problem.
- •The proof was checked by external mathematicians, who also wrote a companion paper explaining the argument and its significance.