• istanbullu@lemmy.ml
    link
    fedilink
    English
    arrow-up
    15
    arrow-down
    2
    ·
    1 year ago

    I don’t buy into this “AI is dangerous” hype. Humans are dangerous.

    • funkless_eck@sh.itjust.works
      link
      fedilink
      English
      arrow-up
      10
      ·
      1 year ago

      “ooh it’s more advanced but don’t worry- it’s not conscious”

      is as much a marketing tactic as “how it feels to chew 5 gum” or buzzfeedesque “top 10 celebrity mistakes - number 3 will blow your mind”

      it’s a tech product that runs a series of complicated loops against a large series of texts and returns the closest comparison, as it stands it’s never going to be dangerous in and of itself.

      • kromem@lemmy.world
        link
        fedilink
        English
        arrow-up
        3
        ·
        1 year ago

        it’s a tech product that runs a series of complicated loops against a large series of texts and returns the closest comparison, as it stands it’s never going to be dangerous in and of itself.

        That’s not how it works. I really don’t get what’s with people these days being so willing to be confidently incorrect. It’s like after the pandemic people just decided that if everyone else was spewing BS from their “gut feelings,” well gosh darnit they could too!

        It uses gradient descent on a large series of texts to build a neural network capable of predicting those texts as accurately as possible.

        How that network actually operates ends up a black box, especially for larger models.

        But research over the past year and a half in simpler toy models has found that there’s a rather extensive degree of abstraction. For example, a small GPT trained only on legal Othello or Chess moves ends up building a virtual representation of the board and tracks “my pieces” and “opponent pieces” on it, despite never being fed anything that directly describes the board or the concept of ‘mine’ vs ‘other’. In fact, in the Chess model, the research found there was even a single vector in the neural network that could be flipped to have the model play well or play like shit regardless of the surrounding moves fed in.

        It’s fairly different from what you seem to think it is. Though I suspect that’s not going to matter to you in the least, as I’ve come to find that explaining transformers to people spouting misinformation about them online has about the same result as a few years ago explaining vaccine research to people spouting misinformation about that.

      • Thorny_Insight@lemm.ee
        link
        fedilink
        English
        arrow-up
        2
        ·
        1 year ago

        Generative AI and LLMs is not what people mean when they’re talking about the dangers of AI. What we worry about doesn’t exist yet.

        • funkless_eck@sh.itjust.works
          link
          fedilink
          English
          arrow-up
          2
          ·
          1 year ago

          I dont think AI sentience as danger is going to be an issue in our lifetimes - we’re 123 years in January since the first well known story featuring this trope (Karel Čapek’s Rossumovi Univerzáiní Robotī)

          We are a long way off from being able to copy virtual perception, action and unified agency of even basic organisms right now.

          Therefore all claims about the “dangers” of AI are only dangers of humans using the tool (akin to the dangers of driving a car vs the dangers of cars attacking their owners without human interaction) and thus are just marketing hyperbole

          in my opinion of course

          • Thorny_Insight@lemm.ee
            link
            fedilink
            English
            arrow-up
            1
            ·
            1 year ago

            Well yeah perhaps, but isn’t that kind of like knowing that an asteroid is heading towards earth and feeling no urgency about it? There’s non-zero chance that we’ll create AGI withing the next couple years. The chances may be low but consequences have the potential to literally end humanity - or worse.

        • hikaru755@feddit.de
          link
          fedilink
          English
          arrow-up
          2
          ·
          1 year ago

          I mean… It might be. Just depends on how much potential there still is to get models up to higher reasoning capabilities, and I don’t think anyone really knows that yet

          • Thorny_Insight@lemm.ee
            link
            fedilink
            English
            arrow-up
            3
            ·
            1 year ago

            Yeah maybe. I just personally don’t think LLMs are actually intelligent. They’re just capable of faking intelligence but at the same time making errors that perfectly indicate that it’s basically just bluffing. I’d be more worried about an AI that knows less things but demonstrates higer capability for logic and reasoning.

    • Thorny_Insight@lemm.ee
      link
      fedilink
      English
      arrow-up
      7
      ·
      edit-2
      1 year ago

      AI can be dangerous. The point is not that it’s likely but that in the very unlikely event of it going rogue it can at worst have civilication ending consequences.

      Imagine how easy it is to trick a child as an adult. The difference in intelligence between a human and superintelligent AGI would be orders of magnitude greater that that.

      • conciselyverbose@sh.itjust.works
        link
        fedilink
        English
        arrow-up
        1
        ·
        1 year ago

        An actual AI (that modern tools don’t even vaguely resemble) could maybe theoretically be dangerous.

        An LLM cannot be dangerous. There’s no path to anything resembling intelligence or agency.

    • kromem@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      Exactly. People try to scare into regulatory capture talking about paperclip maximizers when meanwhile it’s humans and our corporations that are literally making excess shit to the point of human extinction.

      To say nothing for how often theorizing around ‘superintelligence’ imagines the stupidest tendencies of humanity being passed on to it while denying our smartest tendencies as “uniquely human” despite existing models largely already rejecting the projected features and modeling the ‘unique’ ones like empathy.

    • xcjs@programming.dev
      link
      fedilink
      English
      arrow-up
      4
      ·
      edit-2
      1 year ago

      I was reflecting on this myself the other day. For all my criticisms of Zuckerberg/Meta (which are very valid), they really didn’t have to release anything concerning LLaMA. They’re practically the only reason we have viable open source weights/models and an engine.

  • kromem@lemmy.world
    link
    fedilink
    English
    arrow-up
    3
    ·
    edit-2
    1 year ago

    It’s not as good as it seems at the surface.

    It is a model squarely in the “fancy autocomplete” category along with GPT-3 and fails miserably at variations of logic puzzles in ways other contemporary models do not.

    It seems that the larger training data set allows for better modeling around the fancy autocomplete parts, but even other similarly sized models like Mistral appear to have developed better underlying critical thinking capacities when you scratch below the surface that are absent here.

    I don’t think it’s a coincidence that Meta’s lead AI researcher is one of the loudest voices criticizing the views around emergent capabilities. There seems to be a degree of self-fulfilling prophecy going on. A lot of useful learnings in the creation of Llama 3, but once other models (i.e. Mistral) also start using extended training my guess is that any apparent advantages to Llama 3 right now are going to go out the window.

  • BetaDoggo_@lemmy.world
    link
    fedilink
    English
    arrow-up
    1
    ·
    1 year ago

    The 8B is incredible for it’s size and they’ve managed to do sane refusal training this time for the official instruct.