[國際] AI挑戰160年數學難題 研究員稱找到黎曼猜想證明
黎曼猜想(Riemann hypothesis)常被視為數學界最重要的未解難題,如今可能成為人工智慧(AI)征服的最新目標。 紐約時報(The New York Times)報導,任職於Anthropic的數學家阿爾波格(Levent Alpoge)20日在個人網站發布一份4頁說明,表示他與公司旗下AI系統Claude合作,已完成黎曼猜想的證明。他沒有提出傳統形式的書面論證,而是公布可供電腦檢驗的程式碼,聲稱可用來驗證結果。 這項說法尚未經外部數學專家正式審查,相關證明仍有待學界進一步檢視。 若這項宣稱獲得證實,將是AI今年又一項驚人進展。數月來,數學家一再驚訝地看著機器解決多年來難倒人類天才的數學問題;以黎曼猜想而言,這道難題已懸而未解超過一個半世紀。 康乃爾大學(Cornell University)數學教授史托加茲(Steven…
A mathematician working for the AI company Anthropic claims to have found a proof for the Riemann Hypothesis, considered one of the most important unsolved problems in mathematics. The New York Times reported that Levent Alpoge, a researcher at Anthropic, released a four-page explanation on his personal website on the 20th, stating that he collaborated with Anthropic's AI system Claude to complete the proof.
Instead of presenting a traditional written argument, Alpoge published code that can be tested by computers, claiming it can verify the results. However, this claim has not yet been officially reviewed by external mathematicians, and the proof is still awaiting further scrutiny from the academic community. If validated, it would mark another remarkable achievement for AI, as machines have recently solved numerous mathematical problems that stumped human experts.
The Riemann Hypothesis, proposed by Bernhard Riemann in 1859, deals with prime numbers and their distribution along the number line. As numbers get larger, primes become increasingly scarce, with about 25% of numbers between 1 and 100 being prime, compared to 7.8% between 1 and 100 million. Riemann proposed a formula to estimate the number of primes below any given number.
Other researchers are examining Alpoge's code, written in Lean, a language that converts mathematical statements into verifiable forms. If the Lean code is error-free, it suggests the proof is logically sound. While the mathematical community reviews Alpoge's proof, other mathematicians have independently submitted their own attempts.
Stanisloao Masullo, an independent mathematician in Rome, revealed a draft proof after 12 years of work, admitting it still has many shortcomings. Masullo acknowledges that these gaps demonstrate the proof is still a human endeavor. Brian Conrad, a Stanford University mathematician and co-director of Dr. Masullo's thesis advisor, states that Masullo's approach differs significantly from Alpoge's and Claude's methods, and his work appears credible with no obvious flaws. Conrad emphasizes that Masullo's proof requires further refinement, verification, and improvement.
Rewritten by urgent.news; may contain errors — check the original.