• Eggymatrix@sh.itjust.works
        link
        fedilink
        arrow-up
        5
        ·
        14 hours ago

        Same could be said with people hyping something.

        This tech might be good for tripping on shrooms, but as with other applications it is being forced on, lacks the planning and stability to be useful yet.

        One year is overéy optimistic. Maybe 5 - 10.

        • ☆ Yσɠƚԋσʂ ☆@lemmy.ml
          link
          fedilink
          arrow-up
          1
          arrow-down
          1
          ·
          12 hours ago

          The rate of improvement every few months has been absolutely incredible. Given we already have open source stuff like this, 5-10 years seems rather pessimistic. But even that isn’t really that long of a time frame in the grand scheme of things.

          • msage@programming.dev
            link
            fedilink
            arrow-up
            1
            ·
            3 hours ago

            I mean, it’s not?

            Demos of marginal improvements pale in comparison of the real missing pieces.

            I’ll keep saying it: we need new designes of models, not billions of investment into the current ones.

            • ☆ Yσɠƚԋσʂ ☆@lemmy.ml
              link
              fedilink
              arrow-up
              1
              arrow-down
              1
              ·
              2 hours ago

              There’s been stunning progress happening every year here, and people are in fact designing new models all the time. The whole thing with stuff like Genie is a result of new models in fact. The whole idea of world models is a significant step forward from LLMs.