The Riemann hypothesis is one of the most difficult problems in mathematics, having stood for 167 years. It may have just been settled with the help of AI

  • Rimu@piefed.social
    link
    fedilink
    English
    arrow-up
    8
    ·
    1 day ago

    They’re using AI to generate proofs that can be machine-verified to be true yet are too complicated and voluminous for humans to keep up with. 17 million lines of Lean code, in this case.

    If we ever get a warp drive, we might not understand how it works.

    • CombatWombat@feddit.onlineOP
      link
      fedilink
      English
      arrow-up
      1
      ·
      24 hours ago

      A lot of the big Lean proofs I’ve seen end up proving there are gaps or bugs in the Lean compiler, rather than what they set out to prove, so it’ll be interesting to see on review which case this is.

  • CombatWombat@feddit.onlineOP
    link
    fedilink
    English
    arrow-up
    4
    ·
    1 day ago

    I am no friend of the llm, but this’ll probably be in the news for a bit

    “I believe I have, working with Claude, discovered a proof of the Riemann hypothesis,” writes Alpöge. “It is natural to feel that this is an end, in my case a loss of a principal guiding light, 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 August, Anthropic announced previous work on the Riemann hypothesis, but that stopped short of a full proof.

    • Valmond@lemmy.dbzer0.com
      link
      fedilink
      arrow-up
      5
      arrow-down
      1
      ·
      1 day ago

      Who knows, but people believe anything. Makes me remember scientists and researchers “breaking RSA” by not understanding that it’s based on complexity and going for small keys, which are solvable in an excel spreadsheet…

  • sqw
    link
    fedilink
    arrow-up
    2
    arrow-down
    1
    ·
    1 day ago

    “sycophancy engine claims to have done what was asked of it”