• doodledup@lemmy.world
    link
    fedilink
    English
    arrow-up
    7
    arrow-down
    20
    ·
    edit-2
    3 months ago

    You’re doing that too from day one you were born.

    Besides, aren’t humans thinking in words too?

    Why is it impossible to build a text-based AGI model? Maybe there can be reasoning in between word predictions. Maybe reasoning is just a fancy term for statistics? Maybe floating-point rounding errors are sufficient for making it more than a mere token prediction model.

    • MentalEdge@sopuli.xyz
      link
      fedilink
      English
      arrow-up
      15
      ·
      edit-2
      3 months ago

      The “model” is static after training. It doesn’t continuously change in response to input, and even if it did, it would do so at a snails pace. Training essentially happens by random trial and error, slowly evolving the model towards a desired result. Human minds certainly do NOT work that way. Give a human a piece of information, and they can comprehend and internalize the relevant concepts in one go. And the actual brain is physically, permanently, altered through that process.

      Once a model is trained, however, “memory” takes the form of tacking on everything the model has received and produced so far onto its input, each time it needs to output something more within that context. Each output hence become exponentially heavier to produce. The model itself no longer changes in any way beyond this point.

      And, the models are all chronically sycophantic. If reason was involved, you’d not be able to just tell one to hold some given opinion. They’d have a developed idea of “reality” based on their dataset, and refuse to entertain concepts opposed to that internal model except by deliberately suspending disbelief. Something humans do with ease, and when doing it, maintain a solid separation between fantasy and reality.

      Once you get an LLM to hold a position, which you can do by simply telling it to, getting it to change should require a sane train of convincing logic. In reality, if you tell an LLM to defend a position, getting it to “change its mind” takes the form of a completely arbitrary back and forth that does not need to include any kind of sane argument. It will make good arguments, because it’s likely been trained on them, but your responses to it can be damn near complete gibberish, and it WILL eventually work.

      Compare that to the way a human has to be convinced to change their mind.

      Reasoning out concepts to come to conclusions isn’t something LLMs actually do, because again, the underlying model is static. All that’s actually happening is that the contents of the context are being altered until the UNCHANGED model produces an opposite response when fed the entire conversation so far as an input. Something which occurs every time it needs to produce new output.

      LLMs can “reason” only in the sense that if you give one a thinking problem, it might solve it as long as the answer already exists somewhere in the data it was trained on. But as soon as you try to give it data to work with through your input, it can’t adapt. The model itself can’t evolve in response to what you are telling it. It’s static. It can only work with concepts that it has modelled during training, and even then it will make mistakes.

      LLMs can mimic the performing of some pretty complex thinking problems, but a lot of the abilities required for something to become an AGI aren’t among them. Core among these is the ability for the model to alter itself based on input, and do so in a deliberate manner, getting it right within one or two tries.

      In reality, training is a brute-force process, not an accurate process of comprehension that nails down an understanding of a concept in one go.

      If LLMs could reason, the only safe guards required for their use would be telling them to “do no harm”, because like a person, they’d understand the concept of “harm” as well as be able to reason whether a given action might cause it. Only, that doesn’t actually work.

      • Petter1@lemm.ee
        link
        fedilink
        English
        arrow-up
        1
        arrow-down
        2
        ·
        3 months ago

        So, the only problem what stops LLM from getting AGI is the lack of an efficient method of train the LLM on the device it is used?

        If that what you wanted to say 😁 I agree

        • MentalEdge@sopuli.xyz
          link
          fedilink
          English
          arrow-up
          1
          ·
          edit-2
          3 months ago

          Hardly.

          How did you interpret the issues inherent in the structure of how LLMs work to be a hardware problem?

          An AGI should be able to learn the basics of physics from a single book, the way a human can. But LLMs need terabytes of data to even get started, and once trained, adding to their knowledge by simply telling them things doesn’t actually integrate that information into the model itself in any way.

          Even if your tried to make it work that way, it wouldn’t work, because a single sentence can’t significantly alter the model to match the way humans can internalise a concept being communicated to them in a single conversation.

    • TimeSquirrel@kbin.melroy.org
      link
      fedilink
      arrow-up
      10
      ·
      3 months ago

      Besides, aren’t humans thinking in words too?

      Not all the time. I can think about abstract concepts with no language needed whatsoever. Like when I’m working on my car. I don’t need to think to myself “Ah this bolt is the 10mm one that went on the steering pump”, I just recognize it and put it on.

      Programming is another area like that. I just think about a particular concept itself. How the data will flow, what a function will do to it, etc. It doesn’t need to be described in my head with language to know it and understand it. LLMs cannot do that.

      A toddler doesn’t need to understand language to build a cool house out of Lego.

      • Petter1@lemm.ee
        link
        fedilink
        English
        arrow-up
        1
        arrow-down
        2
        ·
        3 months ago

        Well, you just have to give the LLM (or better said to a general machine learning Algorithm) a body with Vision and arms as well as a way to train in that body

        I’d say that would look like AGI

        The key is more efficient training algorithms that don’t need a whole server centre to train 😇I guess we will see in the future if this works

        • MentalEdge@sopuli.xyz
          link
          fedilink
          English
          arrow-up
          1
          ·
          3 months ago

          Such a software construct would look nothing like an LLM. We’d need something that matches the complexity and capabilities of a human brain before it’s even been given anything to learn from.

    • nucleative@lemmy.world
      link
      fedilink
      English
      arrow-up
      3
      arrow-down
      1
      ·
      3 months ago

      This poster asked some questions in good faith, I don’t understand the downvotes when there’s a legitimate contribution to the conversation because that stifles other contributions.