• Buelldozer@lemmy.today
    link
    fedilink
    English
    arrow-up
    0
    ·
    5 months ago

    The question I’d like to ask them is WHY they want to get involved in Content Moderation. They make a toolset, nothing more, so why do they care what someone is using the tools for? What could they possibly get out of this that makes it worth the time or expense?

    • Jordan117@lemmy.world
      link
      fedilink
      English
      arrow-up
      0
      ·
      5 months ago

      I imagine it’s because of the generative AI stuff. If they’re using their servers to generate, they’re going to be responsible for what it puts out, even if it’s just responding to user prompts.

        • Halcyon@discuss.tchncs.de
          link
          fedilink
          English
          arrow-up
          0
          ·
          5 months ago

          Yep, and with access to the work files they not only can use final images for AI training but they have access to the complete background information like the different layers of an image.

      • thatsnothowyoudoit@lemmy.ca
        link
        fedilink
        English
        arrow-up
        0
        ·
        5 months ago

        As someone who’s used their tooling and the generative tooling… I have to admit trying to push its limits for giggles. It is VERY conservative already so I don’t see why they’d need additional moderation privileges.

        This is an awful change.

        • jacksilver@lemmy.world
          link
          fedilink
          English
          arrow-up
          0
          ·
          5 months ago

          I tried using their generative tools a while back and they were pretty terrible. Curious what your experience has been.

          • thatsnothowyoudoit@lemmy.ca
            link
            fedilink
            English
            arrow-up
            0
            ·
            edit-2
            5 months ago

            The illustrator tools are terrible. But removing and replacing backgrounds in Photoshop has been spectacular with one caveat - they are less great if you give it any instruction. If you use the generative fills with prompts the results are not at all great. However, if you leave the prompt blank it does a bang-up job matching the existing background set / scene.

            Equally impressive has been generating parts of photos that are missing when extending the canvas size.

            It tends to work best with photos that are “inside” (interiors) with strong geometric cues - but it has expertly matched lighting, backgrounds and their level of focus (or lack thereof).

            • jacksilver@lemmy.world
              link
              fedilink
              English
              arrow-up
              0
              ·
              5 months ago

              Thanks for the insight, I was using it to create something new from a prompt, so my bad experience seems to align with yours.