Researchers Race AI to Solve a Famous Math Problem
Human researchers rushed to publish a complex mathematical proof after hearing rumors that an artificial intelligence company was about to release a similar solution. Both groups successfully tackled major problems in theoretical computer science.
In short: Human researchers rushed to publish a complex mathematical proof after hearing rumors that an artificial intelligence company was about to release a similar solution. Both groups successfully tackled major problems in theoretical computer science.
When a text message arrived warning that artificial intelligence might beat him to a legendary math discovery, Dor Minzer thought it was a joke. Soon, the rumors proved real, setting off a high-stakes race between human brainpower and machine generation.
What happened, in plain words
Dor Minzer, Yumou Fei, and Shuo Wang spent years working on a tough puzzle related to the 2-to-1 games conjecture in theoretical computer science. When they heard rumors in September 2026 that OpenAI was about to release a proof of the related unique games conjecture, the trio rushed to publish their own 95-page paper online to avoid being overshadowed. Shortly after, OpenAI publicly announced proofs for the unique games conjecture and the 2-to-1 conjecture, alongside hundreds of other math results.
Key points
- Human researchers beat a press release: Minzer and his graduate students rushed to share their mathematical manuscript online before OpenAI could release its own automated proofs.
- A long-standing puzzle cracked: The human team proved a variant of a conjecture first proposed by Subhash Khot, showing that some coloring problems remain very hard even with many extra color choices.
- AI enters the world of mathematics: OpenAI released a massive batch of proofs, including one for the unique games conjecture, sparking both curiosity and anxiety among human mathematicians.
Terms explained
- Conjecture — A smart guess or idea that looks true based on tests and logic, but has not yet been proven for every single case. Example: Noticing that it always rains on your birthday for five years in a row leads to the guess that it always rains on your birthday, even though you have not checked every year in history.
- Computational complexity theory — The scientific study of how hard math and computer problems are to solve, and why some take too much time or power. Example: Figuring out why organizing a massive seating chart for a giant wedding is so much harder than adding two simple numbers together.
- Graph — A network made up of dots called nodes connected together by lines called edges. Example: A map of cities connected by airline flight paths, where cities are the nodes and flights are the edges.
Why it matters
These foundational discoveries help scientists understand the natural limits of what computers and algorithms can calculate, which can eventually guide research in many different technical fields.
What we still don't know
The human paper was rushed and lacks connecting words in later sections, while the AI proofs are generated manuscripts that did not undergo traditional human editing or review by independent experts.
Based on reporting from Quanta Magazine. This is an independent explainer, written in our own words with AI assistance; Quanta Magazine has not reviewed or endorsed it. Read the original for the full details.