What is Erdos unit distance problem?
Erdos unit distance problem
The Erdős unit distance problem is a longstanding open problem in combinatorial geometry, first raised by Paul Erdős in 1946, concerning the maximum number of unit distances among n points in the plane. The material tracks its resolution by an AI model and the spread of the techniques that solution exposed.
How it developed
- May 2026 - Nathaniel Whittemore noted the solver was a general-purpose LLM with no specific training for this problem or for mathematics.
- Aug 2026 - Sean Carroll said the famous Erdős unit distance problem had recently been solved by an OpenAI model.
- Sep 2026 - Alex Kontorovich said that a week later the unrelated sum-product problem was also solved in the negative by human beings applying ideas and techniques exposed by the AI solution.
In the evidence
Every line below is attributed to a named speaker.
AI solving a math problem can accelerate human breakthroughs in unrelated problems: human mathematicians solved the sum-product problem within one week of an AI counterexample to the Erdos unit-distance problem, by borrowing the AI's exposed techniques.
“The most fun thing about this is, a week later, a completely unrelated problem called the sum-product problem was also solved in the negative by human beings who were applying the ideas, the techniques that were exposed to us from this AI solution.”Alex Kontorovich · 4 Sep 2026
An OpenAI model solved the Erdos unit distance problem, a longstanding open problem in discrete mathematics.
“Recently, the famous Erdos unit distance problem was solved by an OpenAI model.”Sean Carroll · 3 Aug 2026
OpenAI: its model recently solved the Erdos unit distance problem, a landmark result suggesting LLM-adjacent systems can close longstanding open problems in combinatorics.
“Recently, the famous Erdos unit distance problem was solved by an OpenAI model.”Sean Carroll · 3 Aug 2026
A general-purpose LLM with no math-specific training and no complex prompting solved the Erdos unit distance problem, an 80-year-old unsolved mathematical challenge, suggesting frontier AI capability may already exceed domain-specific tool assumptions.
“It was just a general-purpose LLM with no specific training for this problem or mathematics.”Nathaniel Whittemore · 24 May 2026