… and neither does the author (or so I believe - I made them both up).

On the other hand, AI is definitely good at creative writing.

  • @[email protected]
    link
    fedilink
    English
    13 months ago

    Of course you don’t. You’re one of the non-creatives who thinks that “prompt engineering” makes you a creative, undoubtedly.

    Sure, that’s exactly what I believe … Wow I’m so called out. I use it as a tool to do boring menial tasks so that I can spend my time on more meaningful things, like spending time with my family, making some dinner, spend time on the parts of my work I enjoy and automate the boring tedious parts, like writing boilerplate code that’s slightly different based on context.

    But the first “L” in “LLM” says it all. The very definition of degenerative AI requires the wholesale dismemberment of human culture to work and, indeed

    Can you elaborate on how and the mechanisms by which this is happening as you see? Why do you see it that way? Do you not see any circumstances in which it could be useful? Like legitimately useful? Like have you not written a stupid tedious email to someone you didn’t like that you couldn’t be bothered to put more than 2 seconds to prompt it to some one or thing else to deal with it for you?

    there’s already a problem: the LLM purveyors have hit a brick wall. They’ve run out of material to siphon away from us and are now stuck with only finding “new” ways to remix what they’ve already butchered in the hopes that we think the stench from the increasingly rotten corpse won’t be noticeable.

    This is true it’s starting to eat its own tail. That also doesn’t mean all new models are using new data. It could also be using better architectures on the same data. But yes using ai generated data to train new ai is bad and you’ll end up creating nerfed less useful model that will probably hallucinate more. Doesn’t mean the tech isn’t useful cause you’ve not seen it used for anything good.

    • Like have you not written a stupid tedious email to someone you didn’t like that you couldn’t be bothered to put more than 2 seconds to prompt it to some one or thing else to deal with it for you?

      No, I haven’t. I call out bullshit in my job instead of acquiescing to it. I’m not sure when I last wrote an email at work at all, not to mention a stupid, tedious one.

      If there’s a part of your job that can be done by degenerative AI, change how your job works. If your boss won’t let you change the bullshit, change your job. I’ve been doing this since I was 15. It’s not that hard.

      Can you elaborate on how and the mechanisms by which this is happening as you see?

      Here, this may help you grasp it.

      Why do you see it that way?

      Because I looked into how it works and spotted the bit where it needs a huge volume of input data. That input data is going to be indiscriminately vacuumed up because it’s not feasible to check each piece for permission. (Or do you naively believe that if I put a disclaimer on, say, a blog saying “this material is specifically not permitted to be used as training material for AI projects” means that it won’t be Hoovered in with everything else?)

      And here’s some cool little factoid for you if you don’t believe that it’s being vacuumed up indiscriminately: Meta announced a new AI siphonbot and gave the information needed to block it. Two weeks after they started using it. And this is generally positive behaviour. Most of the AI bot-crawlers have been found out by sleuthing, not by an announcement. Even AI research teams at universities aren’t doing the basics of ethical conduct: getting consent.

      Do you not see any circumstances in which it could be useful?

      Yes. It’s very useful for non-creatives to pretend they’re actually creative when they send a machine to stitch together the corpse of human culture in entertaining new shapes rendered from rotting flesh. Personally, though, I can live without masterpieces like “Sonic the hedgehog gives birth to Borat” or whatever idiotic shit these keyboard monkeys think is art.

      Doesn’t mean the tech isn’t useful cause you’ve not seen it used for anything good.

      There is no use sufficiently good to justify the dismemberment and destruction of human culture. Sorry.

      • @[email protected]
        link
        fedilink
        English
        13 months ago

        Sounds like all your problems are with capitalism and not LLMs but you can’t see that.

        And good for you that you’re in a position to not deal with bullshit in your work. Not everyone has that luxury.

        Get some empathy for people in different circumstances as you. You sound like a child.

        Also there’s a fuck ton of useful training data with permissive licenses. Also, fuck copyright law. It’s been weaponized by capitalists to control our lives. Especially cause the artists barely gets theirs.

        We’re never gonna see eye to eye so don’t bother. Peace and love. Have a good day.

        • Sounds like all your problems are with capitalism and not LLMs but you can’t see that.

          Show me an anarchist use of LLMs that respects consent. I’ll wait. Indeed, since there are no such examples and thus this is an unfair challenge, I’ll loosen it: Just describe such an LLM: one that people will explicitly opt-in to instead of having to keep track of every two-bit, LLM-pumping moron that pops up so they can opt-out.

          That’s the foundation of the dismantling of the corpse of human creation, after all: the lack of consent mechanism. If you can conceive of a feasible way to provide said consent, then your system is just the looting of the corpse of human creativity.

          Get some empathy for people in different circumstances as you. You sound like a child.

          And you sound like a techbrodude (read: child) throwing a tantrum at people pointing out the absence of clothing on your emperor.

          We’re never gonna see eye to eye …

          This is true. Because you believe in idiotic bullshit and I don’t.