OpenAI Model Disproves 80-Year-Old Math Conjecture, Signaling Shift in Discovery

OpenAI has released solutions to several longstanding open problems in mathematics and theoretical computer science, including the disproof of a central conjecture posed by Paul Erdős in 1946. The breakthrough signals a transition for artificial intelligence from a research assistant to a system capable of original, autonomous discovery.

The model, internally referred to as Astra, disproved the planar unit distance problem by providing an infinite family of examples that yield a polynomial improvement over previous constructions. Beyond this specific victory, OpenAI reports that the AI solved or made significant progress on 10 different longstanding open problems. Notably, these results were produced by a general-purpose reasoning model rather than a specialized tool designed specifically for mathematics.

The Path to Discovery

The unit distance problem asks for the maximum number of pairs of points in a plane that are exactly distance 1 apart. For eight decades, the mathematical community believed that square grid constructions were essentially optimal. Astra’s solution was unexpected because it bridged two disparate fields: it applied sophisticated concepts from algebraic number theory—specifically Golod–Shafarevich theory and infinite class field towers—to a problem rooted in discrete geometry.

Fields medalist Tim Gowers described the achievement as "a milestone in AI mathematics." Arul Shankar, a leading number theorist, noted that the work demonstrates current AI models are capable of having "original ingenious ideas, and then carrying them out to fruition," moving beyond the role of simple helpers to human mathematicians.

Implications for Science

This event marks the first time a prominent open problem in a central subfield of mathematics has been solved autonomously by AI. The ability of a general-purpose model to connect distant areas of knowledge to solve a specific, hard problem suggests a new paradigm for scientific research. If AI can synthesize complex theories across disciplines to find proofs, similar capabilities could accelerate breakthroughs in other hard sciences, such as physics and biology, where interdisciplinary connections are often the key to discovery.

The Future of Intuition

The emergence of Astra has sparked what some describe as an "existential crisis" among mathematicians regarding the role of human intuition. As AI demonstrates the ability to generate original mathematical insights, the industry must now reconcile the traditional view of human creativity with the efficiency of AI-driven discovery. The primary question remaining is whether this represents a peak in pattern recognition or the dawn of a new era of machine-led theoretical science.