• hperrin@lemmy.world
      link
      fedilink
      arrow-up
      29
      arrow-down
      2
      ·
      5 months ago

      It might turn into dumb skynet though. Like a version of skynet that does malicious things, but not because it’s trying to hurt people, just because it’s really stupid and we put it in charge of things.

      • CanadaPlus@lemmy.sdf.org
        link
        fedilink
        arrow-up
        4
        ·
        edit-2
        5 months ago

        We can’t even get them to not be racist under light adversarial conditions. Billions of dollars have probably been spent on that problem to no avail.

        LLMs like ChatGPT have kind of just turned the problem of getting knowledge into a computer, into the problem of getting it back out in a controlled way. It’s still hard and failure-prone but now nobody knows how it works inside.

    • Wirlocke@lemmy.blahaj.zone
      link
      fedilink
      arrow-up
      3
      ·
      5 months ago

      I’ve begun to think of LLMs as compression algorithms for patterns. It can take an existing pattern and apply it on unusual subjects. Like take the pattern of a limerick and apply it to the patterns of Danny Devito, that’s the upper limit of their creativity. So rather than storing information, it stores these patterns making it seem more dynamic.

      The way I see it, human creativity is the combination of patterns but in a chaotic non-analytic way. We make leaps of logic that without precise knowledge of our brains can’t be exactly replicated. Meanwhile LLM’s just do the basic combination of patterns that result in the most generic realization of any idea.

      However the well dries up as soon as we stop training them. They’ll store the basics of any field but fail to replicate new developments or conclusions until trained.

      • Citizen@lemmy.ml
        link
        fedilink
        arrow-up
        3
        ·
        5 months ago

        However the well dries up as soon as we stop training them. They’ll store the basics of any field but fail to replicate new developments or conclusions until trained.

        Exactly this is the reason we should prevent any further data collection by these bastards…

        Don’t feed the beast!