• Phoenixz@lemmy.ca
    link
    fedilink
    English
    arrow-up
    31
    ·
    3 days ago

    It’s technically possible because AI doesn’t exist. The LLM’s we have do exist and these have no idea what it’s doing.

    It’s a database that can parse human language and put pixels together from requests. It has no such concept as child pornography, it’s just putting symbols together in a way it learned before that happen to form a child pornography picture

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

      smh, back in my day we just cut out pictures of the faces of woman we wanted to see naked, and glued them ontop of (insert goon magazine of choice)

    • prac@lemmy.world
      link
      fedilink
      English
      arrow-up
      5
      ·
      3 days ago

      This is a lot of words to basically say the developers didn’t bother to block illegal content. It doesn’t need to ‘understand’ morality for the humans running it to be responsible for what it produces.

      • 🔍🦘🛎@lemmy.world
        link
        fedilink
        English
        arrow-up
        27
        ·
        edit-2
        2 days ago

        Neither of you are wrong. LLMs are wild uncaged animals. You’re asking why we didn’t make a cage, and they’re saying we don’t even know how to make one yet.

        So, why are we letting the dangerous feral beast roam around unchecked?

          • HasturInYellow@lemmy.world
            link
            fedilink
            English
            arrow-up
            4
            ·
            2 days ago

            We as a society have failed to implement those consequences. When the government refused, we should have taken up the mantle ourselves. It should be a mark of great virtue to have the head of a CEO mounted over your fireplace.

            • muusemuuse@sh.itjust.works
              link
              fedilink
              English
              arrow-up
              1
              ·
              edit-2
              2 days ago

              Okay, I’ll take Zuckerberg over the TV if I can place used dildos in his mouth from time to time. Elon, on the other hand, might frighten the cat.

      • Honytawk@feddit.nl
        link
        fedilink
        English
        arrow-up
        3
        ·
        2 days ago

        Yeah, how hard is it to block certain keywords from being added to the prompt?

        We’ve had lists like that since the 90’s. Hardly new technology. Even prevents prompt hacking if you’re clever about it.

      • Phoenixz@lemmy.ca
        link
        fedilink
        English
        arrow-up
        1
        ·
        2 days ago

        Eh, no?

        It’s really REALLY hard to know what content is, and to identify actual child porn even remotely accidentally, even with AI

      • Phoenixz@lemmy.ca
        link
        fedilink
        English
        arrow-up
        5
        ·
        2 days ago

        Not AI as the common people think it is, I guess I should have cleared that up.

        AI as we currently have it is little more than a specialized database