Three frontier artificial intelligence labs claimed victory over the same 80-year-old mathematical conjecture this week, sparking backlash from mathematicians who say the breakthrough was turned into a corporate publicity stunt rather than celebrated as genuine scientific progress. OpenAI's reasoning model, DeepMind's AlphaProof, and Anthropic's Claude Mythos each independently cracked the Erdős unit-distance conjecture, an open problem in discrete geometry that has stumped human mathematicians since 1946.

What Is the Erdős Conjecture and Why Does It Matter?

The Erdős unit-distance conjecture is a fundamental problem in discrete geometry that asks how many times a unit distance can appear in a finite set of points in the plane. Named after mathematician Paul Erdős, the problem has remained unsolved for eight decades, making it one of the most prestigious open questions in mathematics. The fact that three separate AI systems solved it within days of each other demonstrates the raw problem-solving capability that modern large language models (LLMs), which are AI systems trained on vast amounts of text data, have developed.

Beyond the unit-distance conjecture, DeepMind's AlphaProof Nexus, which pairs the Gemini 3.1 Pro model with the Lean formal proof assistant, solved nine open Erdős problems and proved 44 open conjectures from the Online Encyclopedia of Integer Sequences (OEIS), with some of those problems unsolved for 56 years. The inference cost for solving each problem was reportedly only a few hundred dollars.

Why Are Mathematicians Upset About the Breakthrough?

The mathematical community's frustration centers not on the achievement itself, but on how it was announced and framed. A mathematics graduate student took to social media to call OpenAI's announcement "exceedingly tacky and in bad taste," noting that the proof was previously considered "unapproachable" and that turning it into a corporate launch tweet represented a fundamental shift in how mathematical discovery is treated.

The core complaint reflects a deeper cultural tension: mathematics has historically been a discipline that valued patient, long-term inquiry over speed and publicity. For nearly a century, the Erdős conjecture remained an open problem that mathematicians worked on gradually, without pressure to announce results on a specific timeline. The simultaneous claims by three AI labs transformed the problem into a competitive benchmark, where the proofs themselves became secondary to the press releases announcing them.

This moment marks what observers describe as the end of an era. As one source noted, "the era of patient mathematics ended on a Tuesday," signaling that mathematical breakthroughs are now subject to the same competitive dynamics and publicity cycles as software releases or product launches.

How Are AI Labs Competing in Mathematical Problem-Solving?

The race to solve the Erdős conjecture reflects a broader trend in which frontier AI companies use mathematical benchmarks as proof points for their models' reasoning capabilities. Each lab approached the problem differently, showcasing distinct technical strategies:

• OpenAI's Approach: Used an internal reasoning model to disprove the unit-distance conjecture by applying algebraic number theory to a geometric question, demonstrating cross-domain problem-solving ability.
• DeepMind's Strategy: Deployed AlphaProof Nexus, which combines Gemini 3.1 Pro with the Lean formal proof assistant, solving nine Erdős problems and 44 OEIS conjectures at a cost of a few hundred dollars per problem.
• Anthropic's Method: Leveraged Claude Mythos to tackle the same unit-distance conjecture using parallel hypothesis exploration, a technique that tests multiple mathematical approaches simultaneously.

The fact that three labs solved the same problem independently within the same week suggests that modern AI systems have reached a threshold where they can tackle previously intractable mathematical problems. However, it also demonstrates that these capabilities are now being weaponized as competitive differentiators rather than being shared openly with the mathematical community.

What This Means for the Future of Mathematics and AI

The Erdős conjecture breakthrough reveals a fundamental shift in how mathematical discovery happens in the age of AI. Rather than individual mathematicians or small research teams working on problems over years or decades, frontier AI labs can now solve them in weeks, with the primary motivation being corporate reputation and model capability demonstration.

This raises important questions about how mathematical knowledge will be valued and shared going forward. If AI systems can solve open problems faster than humans, but those solutions are announced as corporate achievements rather than scientific contributions, the discipline risks losing the collaborative, open-ended inquiry that has defined it for centuries. The mathematical community's frustration suggests that the industry may need to develop new norms around how AI breakthroughs in pure mathematics are announced and credited.

For developers and technologists watching this unfold, the message is clear: AI reasoning capabilities have matured to the point where they can tackle problems that were previously considered beyond reach. However, the way these capabilities are deployed and communicated matters as much as the capabilities themselves. The Erdős conjecture solved by three labs simultaneously is not just a technical achievement; it is a cultural inflection point for how AI and human expertise will coexist in knowledge-intensive fields.

" }