hn.today

As AI Closed in on 'Unique Games' Proof, Researchers Raced to Beat the Machines

quantamagazine.org4 points0 comments
Screenshot of As AI Closed in on 'Unique Games' Proof, Researchers Raced to Beat the Machines

When rumors circulated that an AI had found a proof of the unique games conjecture, Dor Minzer and his students Yumou Fei and Shuo Wang rushed to finish and post a long-awaited paper they had been refining for years. They published a 95-page draft quickly and candidly - complete mathematically but not polished - partly to avoid being overshadowed by an imminent AI announcement. OpenAI later publicized a large batch of machine-generated proofs, including one for unique games, provoking a mix of alarm and curiosity across complexity theory; peers still hailed Minzer’s result as “truly great.”

The work tackles constraint-satisfaction problems on graphs - the framework of the unique games conjecture - and focuses on Khot’s 2-to-1 variant, which addresses cases where the best solution satisfies every constraint yet approximations remain hard. Minzer, Fei and Wang proved a closely related, slightly weaker 4-to-1 version by building a bridge from the graph-coloring formulation to error-correcting-code constructions. After years of partial advances (including a 2018 milestone) and several failed attempts, they combined a new code with previous pieces to achieve the result in April 2026. That 4-to-1 theorem already yields many of the conjecture’s important consequences, notably new hardness results for classic graph-coloring problems.

Read on quantamagazine.org0 comments on Hacker News

Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.

More in Science

The daily digest

Today's best Hacker News stories, summarized and screenshotted, one email a day.