AI just solved an 80-year-old math problem that certified geniuses thought impossible
An artificially intelligent model has just solved a mathematical problem which has left experts puzzled for nearly 80 years. The model created by OpenAI has just achieved the feat that many scientists once believed to be impossible.
This comes after the OpenAI bot was able to crack part of a famous problem first proposed by legendary mathematician Paul Erdős back in 1946.
The breakthrough stunned researchers across the math world, with some calling it the first genuinely important mathematical discovery produced autonomously by AI.
Advert
What is the Planar Unit Distance problem?
The problem, which is known as the planar unit distance problem, has challenged experts for decades despite sounding surprisingly simple at first glance.
The idea asks how many pairs of points can be placed exactly one unit apart when arranging huge numbers of dots on an infinite surface.
For years, mathematicians thought the best possible arrangements would always resemble neat square grids, an idea strongly supported by Erdős himself.
Although researchers spent decades trying to prove or disprove his theory, most experts quietly assumed the famous mathematician was probably right, and that assumption has now been turned upside down.
OpenAI revealed that one of its internal AI systems managed to find a counterexample to Erdős’ long-standing conjecture using advanced reasoning techniques drawn from algebraic number theory.
The result showed there are actually arrangements of points that produce significantly more one-unit distances than the classic square-grid approach mathematicians had trusted for generations.
What makes the discovery even more remarkable is that the AI involved was not specifically designed purely for mathematics. Instead, it was a general-purpose large language model similar to the technology powering tools like ChatGPT.
The discovery immediately sparked huge discussion within the academic community.
According to a report by The Conversation, Canadian mathematician Daniel Litt described the finding as the first AI-generated mathematical result he considered genuinely interesting in its own right rather than simply a computational novelty.
'Without Hesitation': The math world reacts
Meanwhile, Fields Medal-winning mathematician Timothy Gowers said that if a human researcher had submitted the proof to one of the world’s most prestigious mathematics journals, he would have recommended publication ‘without hesitation’.
He also admitted no previous AI-generated proof had come close to this level of sophistication.
The breakthrough appears to mark a major turning point in the relationship between AI and scientific research.
However, despite the excitement, many experts still believe there are major limitations to current AI systems, with some researchers arguing that while AI can absorb huge amounts of mathematical knowledge and explore possibilities at incredible speed, it remains unclear whether machines truly ‘understand’ concepts the way humans do.