• CanadaPlus
    link
    fedilink
    English
    arrow-up
    1
    ·
    edit-2
    2 years ago

    It’s not exactly a huge breakthrough from a maths perspective, but this will make it easier to explain to people why ChatGPT isn’t just smoke and mirrors.

    • beebarfbadger@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      2 years ago

      It feels like instead of letting the infinite monkeys randomly type random combinations on typewriters, we now breed the monkeys that have shown combinations that were closer to Shakespeare than the statistical average was before.

      • CanadaPlus
        link
        fedilink
        English
        arrow-up
        1
        ·
        edit-2
        2 years ago

        Aren’t we monkeys selectively bred to correctly solve problems?

        The fact that we have no idea how ChatGPT internally works remains, but by every empirical measure we can come up with it shows intelligence. This story specifically shows very straightforwardly that it’s not just copying our work, either.