9 votes

Using Anthropic AI, mathematician claims to have solved Riemann hypothesis

7 comments

  1. [5]
    tibpoe
    Link
    It seems like the Generalized Riemann Hypothesis was simultaneously being proven by a human mathematician without any machine assistance and was in the final stages of editing.

    It seems like the Generalized Riemann Hypothesis was simultaneously being proven by a human mathematician without any machine assistance and was in the final stages of editing.

    16 votes
    1. [4]
      teaearlgraycold
      Link Parent
      The plagiarism machine is at it again!

      The plagiarism machine is at it again!

      1 vote
      1. [3]
        R3qn65
        Link Parent
        That was just published, after anthropic's claimed proof.

        That was just published, after anthropic's claimed proof.

        7 votes
        1. [2]
          teaearlgraycold
          Link Parent
          Just a couple of months ago when OpenAI said they solved Navier-Stokes they'd been tipped off to the solution by a mathematician who had solved it in August.

          Just a couple of months ago when OpenAI said they solved Navier-Stokes they'd been tipped off to the solution by a mathematician who had solved it in August.

          3 votes
          1. R3qn65
            Link Parent
            That's what the initial rumours were, yeah, but additional reporting suggests that those rumours were overblown or flatly untrue. No shade to you for thinking that as things have been changing...

            That's what the initial rumours were, yeah, but additional reporting suggests that those rumours were overblown or flatly untrue. No shade to you for thinking that as things have been changing very rapidly.

            When you go to the details, it is a different approach,” says Gonzalo Cao-Labora, a mathematician at EPFL, the Swiss Federal Institute of Technology in Lausanne. Scientific American spoke with several groups of experts working through the OpenAI paper. All said that it’s increasingly clear that the company’s proof does not follow the Spanish program.

            Https://www.scientificamerican.com/article/openais-blockbuster-navier-stokes-proof-isnt-what-we-thought-it-was/

            7 votes
  2. Adys
    Link
    We stand on the shoulders of the giants that taught rocks how to think.

    We stand on the shoulders of the giants that taught rocks how to think.

    5 votes
  3. skybrian
    Link
    From the article: [...] [...] [...] [...]

    From the article:

    Another domino of mathematics appears to have fallen after a mathematician at artificial intelligence company Anthropic claimed an AI-powered proof to the Riemann hypothesis — the latest remarkable discovery in a tide of findings generated by AI that have left human mathematicians reeling about being bested by machines.

    [...]

    The Riemann hypothesis concerns the mysterious, unpredictable structure of the prime numbers (numbers that can be divided only by one and themselves). It was viewed as one of the field’s deepest mysteries and is one of the seven Millennium Prize problems for which an accepted solution comes with a $1 million prize.

    [...]

    For mathematicians, the allure of the Riemann hypothesis, and others like it, is not the prize money but its ability to motivate useful and profound new discoveries. Hundreds of other questions of interest have been felled in recent months by powerful AI tools, and the field is still reckoning on what to do without these North Stars for direction.

    “It is natural to feel that this is an end, in my case a loss of a principal guiding light,” wrote Levent Alpöge, the Anthropic mathematician who claimed the solution. “But what I saw during the course of progress so amazed me that I believe it is one dawn across science of something far greater.”

    In a lengthy statement accompanying the solution, Alpöge said he had seen Claude, Anthropic’s name for its AI models, rapidly outpace him as they neared the conclusion of the argument. Alpöge said he had helped guide the AI to an initial idea, and then “Claude began to leave me in awe.”

    Anthropic did not immediately respond to a request for comment.

    Now, mathematicians will need to pick through the proof to understand it; some may use AI tools to help. Alpöge released a certificate of 17 million lines of code that he said formally verified the correctness of the argument, but did not have a typed-up paper, which would have run at least 250,000 pages.

    “It being obviously inappropriate to attempt to post some massive unreadable travesty of ‘slop,’” he wrote, alluding to the fact that any write-up would have been principally generated by AI.

    [...]

    Alpöge said he had consulted Anthropic’s leadership about announcing the problem, and that they ultimately decided releasing it was the right thing to do.

    “This problem has been the sacred object in mathematics for me since my childhood, and I am very touchy about the treatment of great intellectual goals, though seriously imperfect,” he said. “I just could not live if I waited even longer.”

    [...]

    Now, the mathematics community will have to absorb the discovery. In the hours after Alpöge’s claim began circulating online, mathematicians rushed to online forums to begin discussing his work.

    “If it turns out to be correct, it will be a memorable moment,” in mathematics, said Cornell professor Steven Strogatz in a message. “A very skilled AI jockey riding a very fast AI horse.”

    2 votes