• over_clox@lemmy.world
    link
    fedilink
    arrow-up
    3
    arrow-down
    19
    ·
    5 months ago

    Yes, I’ve tried similar silly things. I’ve asked AI to render an image of Mr. Bean hugging Pennywise the clown. And it delivered, something randomly silly looking, but still not far off base.

    But when it comes to inappropriate material, well the AI shouldn’t be able to generate any such thing in the first place, unless the developers have allowed it to train from inappropriate sources…

    • FaceDeer@fedia.io
      link
      fedilink
      arrow-up
      19
      arrow-down
      1
      ·
      5 months ago

      The trainers didn’t train the image generator on images of Mr. Bean hugging Pennywise, and yet it’s able to generate images of Mr. Bean hugging Pennywise. Yet you insist that it can’t generate inappropriate images without having been specifically trained on inappropriate images? Why is that suddenly different?

        • FaceDeer@fedia.io
          link
          fedilink
          arrow-up
          9
          arrow-down
          1
          ·
          5 months ago

          3,226 suspected images out of 5.8 billion. About 0.00006%. And probably mislabeled to boot, or it would have been caught earlier. I doubt it had any significant impact on the model’s capabilities.

      • over_clox@lemmy.world
        link
        fedilink
        arrow-up
        2
        arrow-down
        13
        ·
        5 months ago

        Who is responsible then? Cuz the devs basically gotta let the AI go to town on many websites and documents for any sort of training set.

        So you mean to say, you can’t blame the developers, because they just made a tool (one that scrapes data from everywhere possible), can’t blame the tool (don’t mind that AI is scraping all your data), and can’t blame the end users, because some dirty minded people search or post inappropriate things…?

        So where’s the blame go?

        • FaceDeer@fedia.io
          link
          fedilink
          arrow-up
          14
          arrow-down
          1
          ·
          5 months ago

          First, you need to figure out exactly what it is that the “blame” is for.

          If the problem is the abuse of children, well, none of that actually happened in this case so there’s no blame to begin with.

          If the problem is possession of CSAM, then that’s on the guy who generated them since they didn’t exist at any point before then. The trainers wouldn’t have needed to have any of that in the training set so if you want to blame them you’re going to need to do a completely separate investigation into that, the ability of the AI to generate images like that doesn’t prove anything.

          If the problem is the creation of CSAM, then again, it’s the guy who generated them.

          If it’s the provision of general-purpose art tools that were later used to create CSAM, then sure, the AI trainers are in trouble. As are the camera makers and the pencil makers, as I mentioned sarcastically in my first comment.

          • over_clox@lemmy.world
            link
            fedilink
            arrow-up
            2
            arrow-down
            15
            ·
            5 months ago

            You obviously don’t understand squat about AI.

            AI only knows what has gone through it’s training data, both from the developers and the end users.

            Hell, back in 2003 I wrote an adaptive AI for optical character recognition (OCR). I designed it for English, but also with a crude ability to learn.

            I could have taught that thing hieroglyphics if I wanted to. But AI will never generate things that it’s never seen before.

            Funny that AI has an easier time rendering inappropriate material than it does human hands…

            • FaceDeer@fedia.io
              link
              fedilink
              arrow-up
              13
              arrow-down
              1
              ·
              5 months ago

              You obviously don’t understand squat about AI.

              Ha.

              AI only knows what has gone through it’s training data, both from the developers and the end users.

              Yes, and as I’ve said repeatedly, it’s able to synthesize novel images from the things it has learned.

              If you train an AI with pictures of green cars and pictures of red apples, it’ll be able to figure out how to generate images of red cars and green apples for you.

              • over_clox@lemmy.world
                link
                fedilink
                arrow-up
                2
                arrow-down
                10
                ·
                5 months ago

                Exactly. And if you ask it for the opposite of an older MILF, then how does it know what younger ladies look like?

                • FaceDeer@fedia.io
                  link
                  fedilink
                  arrow-up
                  13
                  arrow-down
                  1
                  ·
                  5 months ago

                  It’s possible to legally photograph young people. Completely ordinary legal photographs of young people exist, from which an AI can learn the concept of what a young person looks like.

                  • over_clox@lemmy.world
                    link
                    fedilink
                    arrow-up
                    2
                    arrow-down
                    12
                    ·
                    5 months ago

                    The only example I can think of with what you said is just a couple brief innocent scenes from The Blue Lagoon.

                    Short of that, I don’t know (nor care for any references to) any other legal public images or video of anything as such.

                    I dunno, I’m just bumfuzzled how AI, whether public or private, could have sufficient information to generate such things these days.

                • ricecake@sh.itjust.works
                  link
                  fedilink
                  arrow-up
                  13
                  arrow-down
                  1
                  ·
                  5 months ago

                  Is an image of a child inappropriate? Fully clothed, nothing going on.

                  Is the image of an adult engaging in sexual activity inappropriate?

                  Based on those two concepts, it can generate inappropriate child sexual imagery.

                  You may have done OCR work a while ago, but that is not the same type of machine learning that goes into typical generative AI systems in the modern world. It very much seems as though you are profoundly misunderstanding how this technology operates if you think it can’t generate a novel combination of previously trained concepts without a prior example.

                  • over_clox@lemmy.world
                    link
                    fedilink
                    arrow-up
                    2
                    arrow-down
                    11
                    ·
                    5 months ago

                    I’m referring to the inappropriate photography and videos out there. Please learn to read.