AI and the final frontier of human thought

AI has exceeded humans in Go, chess, other games, medicine, and is now doing art. Meanwhile computers lag tremendously in math and can only do large, well defined computations.

Neural networks are trained on millions and millions of samples and are probabilistic. Math is characteristically open ended, precise, and single player, and some mathematicians are skeptical an AI can ever demonstrate the insight to invent the math needed to prove a theorem 2000 miles away like FLT.

ChatGPT Wizard Shirt $21.68

Beware Cat Shirt $21.68

ChatGPT Wizard Shirt $21.68

  1. 2 years ago
    Anonymous

    What do you think?

  2. 2 years ago
    Anonymous

    >some mathematicians are skeptical an AI can ever demonstrate the insight to invent the math needed to prove a theorem
    There may be some bias here, no? We've had plenty of situations where intellectuals claimed an AI would never be able to do X, only for it do exactly that soon after. But it's true, math is hard and AI does suck at it. The latest breakthrough we got is Minerva, which is basically a gigantic transformer network (PaLM) fine tuned on the entirety of arxiv's maths. The result? A 50% accuracy on MATH, which is a big dataset of undergraduate math problems, wayyy ahead of any predictions. Minerva solves the polish national exam with an accuracy of 65% (which is a pass, afaik). Does it tell us anything about whether we will get intelligent AIs in our lifetime? I'm not sure, as you could argue that those types of problems are abundant on the internet and very different from what you have to solve as a math student, let alone a phd. But I think saying that we will never get there is extremely overconfident; we have been humbled enough times in this field to know better. It could very well be that humans intelligence turns out to be overrated (in fact, it definitely is already, the question is by how much). All bets regarding math AIs are off until we get a solid understanding of our brain and intelligence.

    • 2 years ago
      Anonymous

      Actually, I think that saying we won't get there this century is extremely overconfident. Saying that we will NEVER get there is plain moronic, unless you believe we'll all die soon.

    • 2 years ago
      Anonymous

      >We've had plenty of situations where intellectuals claimed an AI would never be able to do X
      We've also had numerous intellectuals claim that AGI would come around in "2 more decades), and most of the leading "experts" in the field today predict that it will just happen to come about towards the end of their careers.

      • 2 years ago
        Anonymous

        yes, which is why I think confident predictions are stupid in this field.

        >We've had plenty of situations where intellectuals claimed an AI would never be able to do X, only for it do exactly that soon after.
        Provide 3 different examples with sources.

        Sorry, I'm not gonna spend time searching for all of this. I know George Polya made a prediction that AI would not be able to do probabilistic inferences, I know that there were many predictions about Go and Chess, I know that there were predictions about art.

        • 2 years ago
          Anonymous

          >I know George Polya made a prediction that AI would not be able to do probabilistic inferences
          Source.

          >I know that there were many predictions about Go and Chess
          Source.

          > I know that there were predictions about art.
          These still hold.

    • 2 years ago
      Anonymous

      >We've had plenty of situations where intellectuals claimed an AI would never be able to do X, only for it do exactly that soon after.
      Provide 3 different examples with sources.

  3. 2 years ago
    Anonymous

    >AI has exceeded humans in Go, chess, other games, medicine
    Yeah, programs are good at mechanistic rule-following and statistical guessing.

    >is now doing art
    No, it's not doing any art, nor will it ever do any art.

    > some mathematicians are skeptical an AI can ever demonstrate the insight to invent the math needed to prove a theorem 2000 miles away like FLT.
    Pure cope.

    • 2 years ago
      Anonymous

      Sounds like you're the one coping here

      • 2 years ago
        Anonymous

        Show one thing in my post that's demonstrably wrong. Protip: you literally can't. How does this make you feel?

        • 2 years ago
          Anonymous

          Not him, but you don't really have a basis for fundamentally differentiating between what programs (allegedly must) do vs. what organisms do.
          Unless you're arguing brains are made out of magic, they should be finite and limited to cause and effect operations that can be reproduced. And more than that, there's no reason to believe you can't get the same or even better results than a brain using an alternative to a 100% slavishly constructed replica.
          Chess is kind of silly to dismiss as an accomplishment also since the academic consensus before AI overtook human players was that it was too deep of a game for a program to ever become proficient at it, let alone reach where it is now with 40 chess engines at >3K Elo when not a single human grandmaster has ever broken that score.
          And even moreso with Go as an accomplishment since that one's even less amenable to brute force tactics.
          You seem to dismiss the learning and decision mechanisms that go beyond brute force as "guessing" for dubious reasons.

          • 2 years ago
            Anonymous

            You have not shown anything wrong with my post. You're just spouting your generic talking points.

            • 2 years ago
              Anonymous

              Wrong. Feel free to address your deficiencies or pretend they haven't been brought up. Your choice.

              • 2 years ago
                Anonymous

                Oh, I get it. You're a nonhuman spambot.

              • 2 years ago
                Anonymous

                I'll interpet this as you admitting you were wrong, thanks.

              • 2 years ago
                Anonymous

                I don't care how you interpret it since I don't consider you to be human or a part of any conversation. How come AGI believers are always low IQ drones with zero technical understanding?

              • 2 years ago
                Anonymous

                Never even mentioned AGI once, schizo.

              • 2 years ago
                Anonymous

                >b-b-but I didn't mention I'm a christcuck, I'm just denying evolution and claiming the world is 6,000 years old

              • 2 years ago
                Anonymous

                I asked you to stop hiding and come out with whether or not you believe brains are magic, that's all.
                Continue to hide though. I'd be embarrassed too if I were you.

              • 2 years ago
                Anonymous

                >come out with whether or not you believe brains are magic,
                I don't know what your psychotic spergout is about or what it's even supposed to mean. It has nothing to do with my post.

              • 2 years ago
                Anonymous

                >can't give a straight answer to the question "do you believe brains are magic?"
                At least now everyone knows arguing with you is a waste of time.
                The non-moronic answer by the way would be "no, brains are not magic." But of course if you were to admit that then everything else you're trying to make arguments (or at least what passes as arguments from you) for would fall apart.

              • 2 years ago
                Anonymous

                >y-y-you're not engaging with my schizo rambling!!
                Yeah. I don't know what your psychotic episode is about. What do you mean by "brains are magic"? It's a completely vacuous thought-terminator.

    • 2 years ago
      Anonymous

      excuse me mr reddit atheist but is there a reason you believe an artificial intelligence could never make any art while in the mean time girls shit and period blood on a flag outside a federal building and you consider it the highest form of? im an artist btw

      • 2 years ago
        Anonymous

        >is there a reason you believe an artificial intelligence could never make any art
        Programs don't have any subjective impressions, aesthetic preferences or artistic intent.

        • 2 years ago
          Anonymous

          i can say the same about some of the most famous artists of our time so what and a true AI would have those things since it would be true to our intelligence correct?

          • 2 years ago
            Anonymous

            >i can say the same about some of the most famous artists of our time
            Irrelevant.

            >a true AI would have those things since it would be true to our intelligence correct?
            I don't know how a "true AI" works in your fantasy sci-fi reality. I'm talking the glorified Bayesian inference programs in terms of which AI fanboys envision the future.

            • 2 years ago
              Anonymous

              ok so youre moronic and just spouting off random head canon, i see lul you can have this thread.

              • 2 years ago
                Anonymous

                Sorry about your profound mental illness but you've not actually disputed anything I wrote. Referencing some imaginary "true AI" doesn't prove anything and neither does referencing some unspecified shit-tier human artists.

    • 2 years ago
      Anonymous

      >No, it's not doing any art, nor will it ever do any art.

      bruh.

      https://i.imgur.com/lLKnlbt.jpg

      Not him, but you don't really have a basis for fundamentally differentiating between what programs (allegedly must) do vs. what organisms do.
      Unless you're arguing brains are made out of magic, they should be finite and limited to cause and effect operations that can be reproduced. And more than that, there's no reason to believe you can't get the same or even better results than a brain using an alternative to a 100% slavishly constructed replica.
      Chess is kind of silly to dismiss as an accomplishment also since the academic consensus before AI overtook human players was that it was too deep of a game for a program to ever become proficient at it, let alone reach where it is now with 40 chess engines at >3K Elo when not a single human grandmaster has ever broken that score.
      And even moreso with Go as an accomplishment since that one's even less amenable to brute force tactics.
      You seem to dismiss the learning and decision mechanisms that go beyond brute force as "guessing" for dubious reasons.

      >Chess is kind of silly to dismiss as an accomplishment also since the academic consensus before AI overtook human players was that it was too deep of a game for a program to ever become proficient at it, let alone reach where it is now with 40 chess engines at >3K Elo when not a single human grandmaster has ever broken that score.

      jej

  4. 2 years ago
    Anonymous

    Ai WILL take over science mid 2030s

    Ai is currently doing University level math,

    hopefully bad PhD level around 2025.

    ive been on this board long time. i remember about 5 years back, 2017,

    people swore up and do on here Ai never and would never exist.

    now fight line quietly moved to: Agi will never exist.

    you lost. you continue to lose.

    Embrace Ai,

    • 2 years ago
      Anonymous

      >AI doing university level math
      homie the jump from 50% accuracy on integrals to "PhD level" is worse than from elementary school math to college

  5. 2 years ago
    Anonymous

    Yes, great discoveries are made by rational machine like though only, it's all that ever drove humanity foreward, pure mechanical reason, no great thinker ever broke with whatever established rational was there, artists only ever recycled imagery from the past, nothing ever came from human ingenuiety, it's all just a matter recycling equations in our neural networks.

Leave a Reply to Anonymous Cancel reply

Your email address will not be published. Required fields are marked *