It always feels like some form of VR tech comes out with some sort of fanfare and with a promise it will take over the world, but it never does.

  • Chais@sh.itjust.works
    5·
    7 hours ago

    It’s quite bad at what we’re told it’s supposed to do (producing reliably correct responses), hallucinating up to 40% of the time.
    It’s also quite bad at not doing what it’s not supposed to. Meaning the “guardrails” that are supposed to prevent it from giving harmful information can usually be circumvented by rephrasing the prompt or some form of “social” engineering.
    And on top of all that we don’t actually understand how they work in a fundamental level. We don’t know how LLMs “reason” and there’s every reason to assume they don’t actually understand what they’re saying. Any attempt to have the LLM explain its reasoning is of course for naught, as the same logic applies. It just makes up something that approximately sounds like a suitable line of reasoning.
    Even for comparatively trivial networks, like the ones used for written number recognition, that we can visualise entirely, it’s difficult to tell how the conclusion is reached. Some neurons seem to detect certain patterns, others seem to be just noise.

    • Perspectivist@feddit.uk
      43·
      6 hours ago

      You seem to be focusing on LLMs specifically, which are just one subcategory of AI. Those terms aren’t synonymous.

      The main issue here seems to be mostly a failure to meet user expectations rather than the underlying technology failing at what it’s actually designed for. LLM stands for Large Language Model. It generates natural-sounding responses to prompts - and it does this exceptionally well.

      If people treat it like AGI - which it’s not - then of course it’ll let them down. That’s like cursing cruise control for driving you into a ditch. It’s actually kind of amazing that an LLM gets any answers right at all. That’s just a side effect of being trained on a ton of correct information - not what it’s designed to do. So it’s like cruise control that’s also a somewhat decent driver, people forget what it really is, start relying on it for steering, and then complain their “autopilot” failed when all they ever had was cruise control.

      I don’t follow AI company claims super closely so I can’t comment much on that. All I know is plenty of them have said reaching AGI is their end goal, but I haven’t heard anyone actually claim their LLM is generally intelligent.

      • Chais@sh.itjust.works
        4·
        6 hours ago

        I know they’re not synonymous. But at some point someone left the marketing monkeys in charge of communication.
        My point is that our current “AI” is inadequate at what we’re told is its purpose and should it ever become adequate (which the current architecture shows no sign of being capable) we’re in a lot of trouble because then we’ll have no way to control an intelligence vastly superior to our own.

        So our current position on that journey is bad and the stated destination is undesirable, so it would be in our net interest to stop walking.

      • Tar_Alcaran@sh.itjust.works
        4·
        6 hours ago

        If people treat it like AGI - which it’s not - then of course it’ll let them down.

        People treat it like the thing it’s being sold as. The LLM boosters are desperately trying to sell LLMs as coworkers and assistants and problemsolvers.