OpenAI Model Solves Math Problem Humans Couldn’t Crack For 80 Years

Times Now ^ | 05/21/2026 | Govind Choudhary

[SeekAndFind]

OpenAI says its AI model solved a famous 80-year-old maths problem that puzzled experts for decades, marking a major breakthrough in AI-powered research and reasoning.

The artificial intelligence boom is real. Sectors like healthcare, IT, education and many others are rapidly moving towards AI adoption. Now mathematicians have also acknowledged how AI is proving its mettle. OpenAI announced on Wednesday that one of its reasoning models has solved a famous maths problem that humans could not solve in 80 years.

Notably, the maths problem, known as the ‘planet unit distance’, was initially proposed in 1946 by legendary mathematician Paul Erdős. Since that time, this problem has remained one of the most discussed puzzles in geometry. OpenAI claims that it is for the first time an AI system has solved a major open maths problem on its own that is central to an active field of research.

What Was The Problem?

The maths problem talked about placing many dots on a flat surface. This challenge was to figure out how many pairs of dots can be exactly one unit apart from each other. For decades, mathematicians believed that the best possible arrangements looked similar to soiree grids. But OpenAI says that its AI model discovered a new method that performs better than the old approach.

OpenAI claims that mathematicians have underestimated the outcomes. The Sam Altman-led firm explained that previous construction could create unit-distance pairs at a rate only slightly better than linear growth. In simple words, as the number of points increased, the number of perfect one-unit pairs also increased, but not dramatically. OpenAI’s model discovered an entirely new family of point arrangements that creates significantly more unit-distance pairs than the old square-grid method.

Experts Call It A Major Moment

Mathematicians have praised the AI’s work. As per OpenAI, Fields Medal winner Tim Gowers called this achievement “a milestone in AI mathematics". Number theorist Arul Shankar said the work shows AI systems are now capable of creating “original ingenious ideas” instead of only assisting humans.

This development shows how AI is turning out to be a strong and brainy research partner rather than being a simple assistant for mundane tasks. This paints a picture of the future where AI could possibly become the assistant for many, affecting the basic level jobs.

---

TOPICS: Computers/Internet; Science
KEYWORDS: ai; math; openai; reasoning

[SeekAndFind]

https://x.com/OpenAI/status/2057176201782075690

Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946.

For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids.

An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better.

This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.

[Scrambler Bob]

OK

What is Infinity divided by Zero?

[TBP]

AI can be helpful wiht some things. But only some. In others, it can be harmful.

[nickcarraway]

Did humans create the tool that solved it?

[Paal Gulli]

"OpenAI Model Solves Math Problem Humans Couldn’t Crack For 80 Years"

Just because humans didn't "crack" it doesn't necessarily mean they couldn't.

It reads like they used AI to create the (self-aggrandizing) headline.

[Ge0ffrey]

Was that a problem?

[Regulator]

So how come the AI instance didn’t propose the problem and then the solution?

[Governor Dinwiddie]

It sounds like the perfect math problem for an autistic spaz
to work on. Give the kid a crayon and a stack of notebooks.

[Jonty30]

The answer is 42.

[dfwgator]

[Deaf Smith]

>>AI instance didn’t propose the problem and then the solution?

Then AI would have shown it was a Democrat; having a solution for a problem they made up or created.

[Revel]

Having AI solve math problems is one thing. Having it interact and control the population is another thing. The 2nd thing is going to bring great despair to the human race.

[dayglored]

The headline is false, and the description of what the AI did, if read carefully, bears this out. The AI did not “solve” the problem, it merely found a better class of solutions, better than the existing human-generated solutions.

The problem is not definitively solved until a human or AI can demonstrate that their new class of solutions is indeed The Best Possible. In other words a “Proof” in the mathematical sense.

Having a solution to a problem (which humans already had), and then finding a better solution to that problem (which the AI did), does not preclude finding yet another solution that’s even better.

So the headline is clickbait. That said, it’s an interesting article and kudos to OpenAI for finding the new solution.

[Tymesup]

The excerpt repeatedly uses “maths”, yuck.