• @mindbleach
    link
    36 months ago

    I don’t care that the draw-anything network was trained by throwing all available JPEGs at it. How else was it going to happen?

    I don’t care that the English literature engine read all the books in the library. Of course it did. How did you study English literature?

    Github’s “copilot” was alarming to me, not because it’s code and code is my thing, but because it tried acting like a bespoke and novel output with no possible complications. Like if Bing’s image thingy let you write “Shrek on a tricycle” and then pretend you owned Shrek. Or simply that DreamWorks Pictures couldn’t possibly object to you slapping that on a t-shirt for twenty bucks.

    Gig economy bullshit is exploitative and can be stopped.

    Rental in general is a target of leftists worldwide, and even mere liberals want it licensed.

    “Cash you can e-mail” was a genuinely great idea until money addicts turned it into wiggly-line gambling.

    But generative AI, even in this primitive state, is going to be a tool for all sorts of cool shit that would otherwise be infeasible bordering impossible. I do not give one solitary fuck that it looked at every page on the internet. We tried all the clever alternatives and it turns out they don’t work. Now there’s some four-gigabyte Python script that can turn any childish scribble into a photorealistic image… and I’m supposed to get mad at how it was made?

    Pass.

    • @SkySyrup
      link
      26 months ago

      exactly. The people who say this clearly have literally no idea how it actually works (I’ve seen people unironically saying it compresses all the images and stores them) and just jump off the next twitter post as reference

      • @mindbleach
        link
        26 months ago

        Or they’re mad it gave them what they asked for. ‘How could it show Sailor Marx eating a spaghetti sandwich, unless it was trained on exactly that subject?! It can’t just combine things!!!’

        This comment is about pornography.