• ColeSloth@discuss.tchncs.de
    link
    fedilink
    arrow-up
    10
    arrow-down
    1
    ·
    1 day ago

    Since no one had AI anything a few years ago; all things that could possibly be done by AI are up 8,400% over the past two years.

    Oh look. Taking pictures has gone up 3,859,643,000,134% since 1827.

  • XiJinpingStanAccount@lemmy.ml
    link
    fedilink
    arrow-up
    10
    ·
    1 day ago

    Honestly that’s a really good point. Where the fuck did the training data come from? If AI can only make what its trained on. I hadn’t really thought about that in relation to this. How isn’t that a bigger talking point?

    • YourMomsTrashman@lemmy.world
      link
      fedilink
      arrow-up
      5
      ·
      6 hours ago

      On the one hand, diffusion models can merge concepts together pretty well (think of those bird-cat chimeras).

      On the other hand, yes, there’s CSAM in training data. A while ago researchers scanned and validated thousands of CSAM images in the training data of just one dataset. This one in particular the backbone of the popular Stable Diffusion.

        • anhopki1@lemmy.ml
          link
          fedilink
          English
          arrow-up
          1
          ·
          1 hour ago

          They probably wanted some ai generated CSAM for themselves too. I honestly wouldn’t be surprised.

          • XiJinpingStanAccount@lemmy.ml
            link
            fedilink
            arrow-up
            1
            ·
            3 minutes ago

            I mean its either gross negligence or active criminality. If PornHub had CSAM on it and just failed to take it down they’d get into serious trouble. An AI company including it in their data sets and then letting their models be used to make more of it is arguably much more blatant than that. Data sets don’t just appear out of thin air. Someone has to make them. We know for a fact they curate them too.

    • shalafi@lemmy.world
      link
      fedilink
      English
      arrow-up
      4
      ·
      22 hours ago

      If the AI knows what a naked adult looks like and knows what clothed child looks like, I don’t see why it couldn’t mash that up. I’d also assume it could train on medical reports as well.

      I’m also quite certain the model reached into plenty of dark areas of the internet and picked up CSAM.

    • Handles@leminal.space
      link
      fedilink
      English
      arrow-up
      4
      arrow-down
      4
      ·
      1 day ago

      Is there a bit of whataboutism at the end of that line of thinking? I think there is.

      • mfed1122@discuss.tchncs.de
        link
        fedilink
        English
        arrow-up
        5
        arrow-down
        1
        ·
        1 day ago

        No, it’s a statistically sound question, the answer to which is essential to have a meaningful understanding of the statistics. If AI image generation went up 100000% in the same time, that means a smaller proportion of users are making CSAM, which could tell us something about a change in audience for AI. On the other hand, if general image generation went up 10%, then this would mean many users have started making CSAM that weren’t before, or that the users who were making it are now making much more. And all of this really depends further on the ratio beforehand. A percent increase is just not a very meaningful way to report this important information.

        We should always encourage asking logical and scientific questions about things, even (in fact, especially) when the face-value interpretation aligns with our existing understanding. “Percentage increase” reports are famously capable of being misleading.

        Lastly, “whataboutism” is often used incorrectly, due to its unfortunate name, people think that any time someone’s question or issue might be phraseable as “what about XYZ”, that that’s whataboutism. But it really, really isn’t. Please read this https://en.wikipedia.org/wiki/Whataboutism , because I fear that the confusing terminology may be steering you wrongly.

      • Treczoks@lemmy.world
        link
        fedilink
        arrow-up
        5
        arrow-down
        1
        ·
        1 day ago

        No, I just would like to see this in a general perspektive. 8400% sounds rather extreme for any figure rising over just a few years, but I have no idea how big that is in perspective.

        Add to this that official numbers about CSAM are often bloated (there was a case years ago where the police claimed they had thousands of CC numbers to identify members of such a forum, in the end they got maybe a dozen culprits because most CC numbers turned out to be stolen).

  • Embargo@lemmy.zip
    link
    fedilink
    arrow-up
    21
    arrow-down
    1
    ·
    2 days ago

    Yeah… That’ll happen when the AI company’s country of origin deems it completely legal.

      • Stormy@thelemmy.club
        link
        fedilink
        arrow-up
        11
        arrow-down
        2
        ·
        1 day ago

        Yes. That’s what Trump and elons America does. They ruled it totally fine and legal to make. What else were they going to do charge elon with a crime for grok making it? No way this is America, men like assaulting children here, and that’s what the law reflects.

        • perturbedstarstuff@lemmy.world
          link
          fedilink
          arrow-up
          2
          ·
          1 hour ago

          There are a lot of events that I never think to make a note or save a source for, because I don’t realise at the time there will be people who never heard of it or deny that it ever happened.

          +1 to the pile.

  • MatSeFi@lemmy.liebeleu.de
    link
    fedilink
    arrow-up
    6
    arrow-down
    10
    ·
    2 days ago

    With the first case of AI related abuse a few years back, there was a more than 1242458842257963234564900521% increase compared to the years before !!!111oneeleven