• mo_ztt ✅
    link
    English
    21 year ago

    So I agree with him that LLMs can’t really understand what they’re doing, and that they can’t produce creative works like humans, and I tend to assume that’ll be true for a while.

    So that being said – a while back something very weird happened that I don’t think got enough attention at the time. When AlphaZero was learning to play chess, it arrived at this markedly different way of playing the game that is objectively better both than the way humans play chess, and the way traditional min-maxing engines play chess. agadmotor analyzed several of the games on his channel, and I looked up some of the games myself after watching his commentary, and it’s really remarkable how differently it plays chess than traditional human strategy. And, its way is clearly better. I improved my chess game significantly by watching it and trying to imitate some of the unusual features of its strategy. I would describe what it arrived at as “genius” – actually very different from the AI art or language generators, because it clearly understands the game at a deeper level than humans were able to do even after hundreds of years of study. That was really remarkable to me when it happened, and it was weird that no one else seemed to think it was.

    • @[email protected]OPM
      link
      fedilink
      English
      11 year ago

      Watch pinned post on this community called “sparks of AGI”, it should help you broaden your understanding on intelligence of LLM’s and potentially ai’s using different architecture.

      • mo_ztt ✅
        link
        English
        01 year ago

        How much time have you yourself spent trying to construct a system which applies LLMs to AGI problems? For me, it’s a few full work days’ worth of tinkering – not much, but enough that I feel like I have some real-world perspective on what’s involved. Please be careful about talking down to me about my need to broaden my understanding.

        So I did watch part of Bubeck’s talk quite a while ago when it first came out, because this is a deeply important topic to me. I don’t fully agree with it; so let me take an excerpt from his paper to explain what I disagree with:

        A question that might be lingering on many readers’ mind is whether GPT-4 truly understands all these concepts, or whether it just became much better than previous models at improvising on the fly, without any real or deep understanding. We hope that after reading this paper the question should almost flip, and that one might be left wondering how much more there is to true understanding than on-the-fly improvisation. Can one reasonably say that a system that passes exams for software engineering candidates (Figure 1.5) is not really intelligent? Perhaps the only real test of understanding is whether one can produce new knowledge, such as proving new mathematical theorems, a feat that currently remains out of reach for LLMs.

        To me, this is a pretty clear statement of the core of what Bubeck is saying, both in the paper and the talk: He goes through a very accurate list of the unbelievably impressive things that GPT-4 can do. Then, he says more or less that because it can do those things, it must be intelligent (or at least have the first sparks of real intelligence). To me, I simply don’t agree with that. Computerized systems could already do extremely impressive computational things; extending that into the domain of language is a huge leap forward, maybe towards AGI. But, if you’re going to say that because it can mimic the language of reasoning, it must be able to reason, because there’s in practice no difference between those things, then I don’t agree with that.

        Could we use LLMs as building blocks for a real AGI system? Yes, absolutely; like I say, I’ve spent a slight but nonzero amount of time actually experimenting with that myself. Are LLMs impressive? Fuck yes. Are they necessarily intelligent because they can do these impressive language-related tasks? To me, no. That seems like a non sequitur. To me, it’s still clear interacting even with GPT-4 that it doesn’t have real understanding of the underlying concepts, and these models are just getting better and better at moving symbols around. Again, I’d actually contrast that specifically against things like AlphaZero, where it does have a deep understanding of the underlying concepts, to the point that it can easily arrive at novel ideas on its own, beyond and superior to what it was programmed with.

        • @[email protected]OPM
          link
          fedilink
          English
          2
          edit-2
          1 year ago

          Please be careful about talking down to me about my need to broaden my understanding.

          Bruh, what? I just recommended you some videos becuse you wrote about how you were surprised about capabilities of AlphaZero. I’m not talking down to you, I was just trying to help you learn more and that’s it…

          Edit: And you overestimate human intelligence from what I can tell, we are not that special

          • mo_ztt ✅
            link
            English
            11 year ago

            Sorry if my message was unnecessarily abrasive or anything. I just was a little taken aback that you seemed to think I wasn’t aware of this particular video, or that I really needed to see it before I’d understand things. Like I said, I already watched this video, and several others, and also spent a not insignificant amount of time actually experimenting with the concepts in practice, which was hugely eye-opening in terms of what LLMs can and can’t do.

            To me, I was aiming to talk about AlphaZero and a particular example of where we already do have some AI systems that within their domain really can do this unbelievable thing he’s talking about when he says “Perhaps the only real test of understanding is whether one can produce new knowledge, such as proving new mathematical theorems, a feat that currently remains out of reach for LLMs.” I think that deserves a lot more notice than it’s been getting (now or back when it happened.)