Why is AI advancing so quickly?

Shit is moving insanely quick. Like it’s been 2 years and now we’re beginning to get into pseudo sentient AI (autoGPT). Did Moores Law just get thrown out the window after he died? What’s goin on here.

ChatGPT Wizard Shirt $21.68

Beware Cat Shirt $21.68

ChatGPT Wizard Shirt $21.68

  1. 12 months ago
    Anonymous

    why did you mention Moores Law? he is unrelated to your topic of discussion. are you a dunning kruger that names people who you don't know about?

    • 12 months ago
      Anonymous

      I know moores law is transistors, not software. Shit comparison. Regardless, software doesn’t magically advance on its own, blows my mind.

      • 12 months ago
        Anonymous

        cant moores law be for software too ?
        I've seen lots of people do it even I thought it was like that too.

        • 12 months ago
          Anonymous

          so far software has been written by puny limited humans. we can't evolve on human lifespans. we somewhat improve, but compared to the concept of AGI that's basically nothing.
          What happens next depends on the ability of the AI to improve itself, if we provide it with the needed resources

        • 12 months ago
          Anonymous

          It's an unnecessary analogy. Just say it like it is instead of what it seems like.

  2. 12 months ago
    Anonymous

    insane we're seeing new developments on a daily basis

  3. 12 months ago
    Anonymous

    Advances are made easier by previous advances. Technology follows an exponential curve

  4. 12 months ago
    Anonymous

    okay please tell me, how do I make money from this? WHAT DO I INVEST IN??? CRYPTO???

    • 12 months ago
      Anonymous

      you can't. i can.

  5. 12 months ago
    Anonymous

    It's moving me more into the sticking my head in the sand and hoping everything works out for the best camp. Spending the last few days listening to these doomsday predictions only ramp up is not good for my mental health.
    Originally I thought I need to get in on this and take advantage of it, buy now I'm about to stop using it altogether (in whatever capacity I can).

    • 12 months ago
      Anonymous

      I’m not a huge believer in it being capable of doomsday bullshit, since there’s hypothetically always a way to limit what an AI can and can’t do. Sure they can code, which means they can also pentest, attempt buffer overflows, find exploits, but we also have the tech to sandbox them. I’m just waiting for advanced ai waifus then I’m clocking out.

      • 12 months ago
        Anonymous

        how does one "clock out" of being turned into paperclips

        • 12 months ago
          Anonymous

          I don’t follow.

      • 12 months ago
        Anonymous

        >but we also have the tech to sandbox them.
        Do we?

  6. 12 months ago
    Anonymous

    Because Hinton was right.

  7. 12 months ago
    Anonymous

    jews hidden technology.. now kys

  8. 12 months ago
    Anonymous

    >Shit is moving insanely quick.
    Only the things you as a consumer are being shown are moving quickly. For years nobody was investing any serious time and effort (and more importantly: letting AI devs us computational power), with the obvious result being nothing shiny to show off to the consumer.
    When the hype train got rolling everybody and their mother jumped aboard and dedicated vast amounts of computing power to their AI projects in hopes that they're the ones to come out on top.
    A combination of vastly increased resources for AI projects and a need to show off to the public at large to attract investors makes it seem like AI projects are developing much more rapidly than they really are.

    • 12 months ago
      Anonymous

      I disagree. A big part of mainstream AI being difficult is the computational power required to run it. Now we’ve gotten to the point where it’s affordable enough to power data centers, consumer hardware for running it is attainable. Maybe I’m stupid but that’s how I kind of see it?

      • 12 months ago
        Anonymous

        You're repeating what he said.

  9. 12 months ago
    Anonymous

    >Like it’s been 2 years and now we’re beginning to get into pseudo sentient AI (autoGPT)
    LMFAO no. Until people are easily able to train uncucked models at least as good as GPT4 that can run on 12GB VRAM or less it's an absolute nothingburger.
    Remember how enormous of an innovation self-attention was? We're about 7 or 8 equally-sized innovations away from being remotely close to true AGI. And with the way the world is going there simply won't be enough competent whites/chinks to get us there.

    • 12 months ago
      Anonymous

      >at least as good as GPT4 that can run on 12GB VRAM or less
      what do when GPT5 comes out? you gonna have a nice day for not having that locally on 2MB VRAM?
      it has to be usable and absolutely fine if it's one step behind commercial offers.

      • 12 months ago
        Anonymous

        >you gonna have a nice day for not having that locally on 2MB VRAM?
        No, I'm going to laugh at AGItrannies who won't be able to afford OAI's $6 gorillion/month API keys and watch as that dogshit corp crumbles due to business not wanting to give them their data either.
        Letting zoomers on the internet was a mistake.

  10. 12 months ago
    Anonymous

    it's less about the process node and more about the architecture, although the node is obviously important
    if you dedicate most of your silicon to matrix multiplications it's going to be good at matrix multiplications, e.g nvidia's h100, or the various other 'tensor processors'
    the reason they didn't do this before is that they're lazy and could count on new process nodes to do the heavy lifting for them

Leave a Reply to Anonymous Cancel reply

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