OpenAI AI Solves 80-Year-Old Erdős Problem With n^(1+δ) Bound
Updated
Updated · Bitcoin.com News · May 31
OpenAI AI Solves 80-Year-Old Erdős Problem With n^(1+δ) Bound
7 articles · Updated · Bitcoin.com News · May 31
OpenAI’s general-purpose model produced point configurations that prove at least n^(1+δ) unit-distance pairs, breaking past the long-assumed near-linear lower bounds for Erdős’ 1946 problem.
Princeton mathematicians verified the construction, turning the result from an AI-generated candidate into a checked mathematical advance recognized by researchers including Tim Gowers and Arul Shankar.
The unit distance problem asks how many pairs among n points in the plane can sit exactly 1 unit apart; earlier progress relied on grid-based constructions that only nudged the lower bound above n.
Researchers said the system combined geometric reasoning with algebraic number theory, suggesting a general inference model—not a math-only engine—can help find rare structures humans had missed.
The breakthrough could extend beyond geometry, with mathematicians pointing to possible AI-assisted gains in combinatorics, coding theory and cryptography.
Can AI invent truly new mathematics, or is it limited to cleverly remixing existing human knowledge?
How will AI's new ability to solve impossible math problems reshape multi-billion dollar industries?
After an AI solved a problem that stumped geniuses for 80 years, what is the future of human creativity?
OpenAI’s General-Purpose AI Disproves 80-Year-Old Erdős Unit Distance Conjecture
Overview
On May 28, 2026, OpenAI announced that its general-purpose AI model had autonomously disproved the long-standing Erdős unit distance conjecture, a major problem in mathematics first proposed in 1946. The AI generated a 125-page proof entirely on its own, without human help during its creation. This breakthrough highlights AI's advanced reasoning abilities and was confirmed valid after rigorous peer review by leading mathematicians. The achievement marks a significant milestone, showing that AI can solve complex, abstract problems and contribute new insights to mathematical research.