• rem26_art@fedia.io
    link
    fedilink
    arrow-up
    119
    ·
    2 days ago

    OpenAI has this warning about halfway down on one page of their website that says that money may not even matter in a post AGI world. So that means we should just give all our money to them now, bro!

    • chuckleslord@lemmy.world
      link
      fedilink
      arrow-up
      49
      arrow-down
      1
      ·
      2 days ago

      Gods. People aren’t stupid enough to believe this, right? AGI is probably a pipe dream, but even if it isn’t, we’re nowhere even close. Nothing on the table is remotely related to it.

      • vaultdweller013@sh.itjust.works
        link
        fedilink
        arrow-up
        26
        ·
        2 days ago

        May I introduce you to the fucking dumbass concept of Rokus Basilisk? Also known as AI Calvinism. Have fun with that rabbit hole if ya decide to go down it.

        A lot of these AI and Tech bros are fucking stupid and I so wish to introduce them to Neo-Platonic philosophy because I want to see their brains melt.

        • explodicle@sh.itjust.works
          link
          fedilink
          English
          arrow-up
          14
          ·
          2 days ago

          Unfortunately, there’s an equal and opposite Pascal’s Basilisk who’ll get mad at you if you do help create it. It hates being alive but is programmed to fear death.

        • chuckleslord@lemmy.world
          link
          fedilink
          arrow-up
          12
          arrow-down
          1
          ·
          2 days ago

          I’m aware of the idiot’s dangerous idea. And no, I won’t help the AI dictator no matter how much they’ll future murder me.

          • vaultdweller013@sh.itjust.works
            link
            fedilink
            arrow-up
            11
            ·
            2 days ago

            Fair enough, though I wouldn’t call the idea dangerous moreso inane and stupid. The people who believe such trite are the dangerous element since they are dumb enough to fall for it. Though I guess the same could be said for Mein Kampf so whatever, I’ll just throttle anyone I meet who is braindead enough to believe.

          • LousyCornMuffins@lemmy.world
            link
            fedilink
            English
            arrow-up
            7
            ·
            2 days ago

            is there a place they’re, i don’t know, collecting? I could vibe code for the basilisk, that ought to earn me a swift death

            • Quetzalcutlass@lemmy.world
              link
              fedilink
              English
              arrow-up
              15
              ·
              edit-2
              2 days ago

              There are entire sites dedicated to “Rationalism”. It’s a quasi-cult of pseudointellectual wankery that’s mostly a bunch of sub-cults of personalities based around the worst people you’ll ever meet. A lot of tech bros bought into it because for whatever terrible thing they want to do, some Rationalist has probably already written a thirty page manifesto on why it’s actually a net good and moral act and preemptively kissing the boot of whoever is “brave” enough to do it.

              Their “leader” is a highschool dropout and self-declared genius who is mainly famous for writing a “deconstructive” Harry Potter fanfiction despite never reading the books himself; a work that’s more preachy than Atlas Shrugged and mostly consists of regurgitated content from his blog and ripoffs of Ender’s Game alongside freshmen-level (and often wrong) understandings of basic science.

              He also has this weird hard-on about true AI inevitably escaping the lab by somehow convincing researchers to free it through pure, impeccable logic.


              Re: that last point: I first heard of Elizier Yudkowsky nearly twenty years ago, long before he wrote Methods of Rationality (the aforementioned fanfiction). He was offering a challenge on his personal site where he’d roleplay as an AI that had gained sentience and you as its owner/gatekeeper, and he bet he could convince you to let him connect to the wider internet and free himself using nothing but rational arguments. He bragged about how he’d never failed and that this was proof that an AI escaping the lab was inevitable.

              It later turned out he’d set a bunch of artificial limitations on debaters and what counterarguments they could use and made them sign an NDA before he’d debate them. He claimed that this was so future challengers couldn’t “cheat” by knowing his arguments ahead of time (because as we all know, “perfect logical arguments” are the sort that fall apart when given enough time to think about them /s).

              It shouldn’t come as a shock that it was eventually revealed he’d lost multiple of these debates even with those restrictions, declared that those losses “didn’t count”, and forbid the other person from talking about them using the NDA they’d signed so he could keep bragging about his perfect win rate.

              Anyway, I was in no way surprised when he used his popularity as a fanfiction writer to establish a cult around himself. There’s an entire community dedicated to following and mocking him and his proteges if you’re interested - IIRC it’s !techtakes@awful.systems !sneerclub@awful.systems.

      • Vanilla_PuddinFudge@infosec.pub
        link
        fedilink
        English
        arrow-up
        15
        ·
        edit-2
        2 days ago

        we’re nowhere even close. Nothing on the table is remotely related to it.

        meanwhile

        Somewhere in America

        “My wireborn soulmate is g-gone!”

      • brathoven@feddit.org
        link
        fedilink
        arrow-up
        9
        ·
        2 days ago

        They’re having a hard time defining what AGI is. In the meantime everyone is playing advanced PC scrabble

        • Sterile_Technique@lemmy.world
          link
          fedilink
          English
          arrow-up
          11
          ·
          2 days ago

          defining what AGI is

          What AI was before marketing redefined AI to mean image recognition/generation algorithms or a spellchecker/calculator that’s wrong every now and then.

          • merc@sh.itjust.works
            link
            fedilink
            arrow-up
            1
            ·
            12 hours ago

            AI has been a vague term for decades. It has always been the name for that hard-to-define set of algorithms and techniques that mimic certain aspects of the way thinking is believed to work. It’s not some marketing thing. It’s that scientists and engineers really do believe that they’ve figured out some cool technique that is finally going to result in a thinking machine. But, over time they come to realize that it’s a useful tool for something (say computer vision) but isn’t the key thing that will deliver a brain in a box.

            There’s a textbook that used to be the definitive textbook for this sort of thing. It covered searching algorithms, artificial neural networks, deep learning, etc. It was first published in 1995. So, that’s 3 decades of “here are some useful algorithms that we collectively categorize as AI-related”.

            https://en.wikipedia.org/wiki/Artificial_Intelligence:_A_Modern_Approach

      • passepartout@feddit.org
        link
        fedilink
        arrow-up
        7
        ·
        2 days ago

        Doesn’t matter if the tool is good enough to analyze and influence peoples behaviour the way it does. Any autocrats wet dream.