Popular iPad design app Procreate is coming out against generative AI, and has vowed never to introduce generative AI features into its products. The company said on its website that although machine learning is a “compelling technology with a lot of merit,” the current path that generative AI is on is wrong for its platform.

Procreate goes on to say that it’s not chasing a technology that is a threat to human creativity, even though this may make the company “seem at risk of being left behind.”

Procreate CEO James Cuda released an even stronger statement against the technology in a video posted to X on Monday.

  • net00@lemm.ee
    link
    fedilink
    English
    arrow-up
    52
    arrow-down
    24
    ·
    edit-2
    4 months ago

    Built on a foundation of theft

    Sums up all AI

    EDIT: meant all gen AI

    • Mirodir@discuss.tchncs.de
      link
      fedilink
      English
      arrow-up
      29
      arrow-down
      6
      ·
      4 months ago

      Does it? I worked on training a classifier and a generative model on freely available galaxy images taken by Hubble and labelled in a citizen science approach. Where’s the theft?

      • unsignedbit@lemm.ee
        link
        fedilink
        English
        arrow-up
        1
        arrow-down
        11
        ·
        4 months ago

        Hard to say. Training models is generative; training a model from scratch is costly. Your input may not infringe copyright but the input before or after may have.

        • Mirodir@discuss.tchncs.de
          link
          fedilink
          English
          arrow-up
          8
          arrow-down
          1
          ·
          4 months ago

          I trained the generative models all from scratch. Pretrained models are not that helpful when it’s important to accurately capture very domain specific features.

          One of the classifiers I tried was based on zoobot with a custom head. Assuming the publications around zoobot are truthful, it was trained exclusively on similar data from a multitude of different sky surveys.

    • BumpingFuglies@lemmy.zip
      link
      fedilink
      English
      arrow-up
      16
      arrow-down
      6
      ·
      4 months ago

      Can you explain how you came to that conclusion?

      The way I understand it, generative AI training is more like a single person analyzing art at impossibly fast speeds, then using said art as inspiration to create new art at impossibly fast speeds.

      • Clasm@ttrpg.network
        link
        fedilink
        English
        arrow-up
        17
        arrow-down
        16
        ·
        4 months ago

        The art isn’t being made btw so much as being copy and pasted in a way that might convince you it was new.

        Since the AI cannot create a new style or genre on its own, without source material that already exists to train it, and that source material is often scraped up off of databases, often against the will and intent of the original creators, it is seen as theft.

        Especially if the artists were in no way compensated.

        • FatCrab@lemmy.one
          link
          fedilink
          English
          arrow-up
          10
          arrow-down
          4
          ·
          4 months ago

          This is absolutely wrong about how something like SD generates outputs. Relationships between atomic parts of an image are encoded into the model from across all training inputs. There is no copying and pasting. Now whether you think extracting these relationships from images you can otherwise access constitutes some sort of theft is one thing, but characterizing generative models as copying and pasting scraped image pieces is just utterly incorrect.

          • Clasm@ttrpg.network
            link
            fedilink
            English
            arrow-up
            2
            arrow-down
            1
            ·
            4 months ago

            While, yes it is not copy and paste in the literal sense, it does still have the capacity to outright copy the style of an artist’s work that was used to train it.

            If teaching another artist’s work is already frowned upon when trying to pass the trace off as one’s own work, then there’s little difference when a computer does it more convincingly.

            Maybe a bit off tangent here, since I’m not even sure if this is strictly possible, but if a generative system was only trained off of, say, only Picasso’s work, would you be able to pass the outputs off as Picasso pieces? Or would they be considered the work of the person writing a prompt or built the AI? What if the artist wasn’t Picasso but someone still alive, would they get a cut of the profits?

            • FatCrab@lemmy.one
              link
              fedilink
              English
              arrow-up
              1
              arrow-down
              1
              ·
              4 months ago

              The outputs would be considered no one’s outputs as no copyright is afforded to AI general content.

        • paw@feddit.org
          link
          fedilink
          English
          arrow-up
          12
          arrow-down
          7
          ·
          4 months ago

          To add to your excellent comment:

          It does not ask if it can copy the art nor does it attribute its generated art with: “this art was inspired by …”

          I can understand why creators unhappy with this situation.

          • ReCursing@lemmings.world
            link
            fedilink
            English
            arrow-up
            2
            arrow-down
            8
            ·
            4 months ago

            Do you go into a gallery and scream “THIS ART WAS INSPIRED BY PICASSO. WHY DOESN’T IT SAY THAT! tHIS IS THEFT!” - no, I suspect you don’t because that would be stupid. That’s what you sound like here

      • gap_betweenus@lemmy.world
        link
        fedilink
        English
        arrow-up
        5
        arrow-down
        5
        ·
        4 months ago

        With this logic photography is a painting, painted at an impossible high speed - but for some reasons we make a difference between something humans make and machines make.

        • ReCursing@lemmings.world
          link
          fedilink
          English
          arrow-up
          4
          arrow-down
          10
          ·
          4 months ago

          Amusingly, every argument against ai art was made against photography over a hundred years ago, and I bet you own a camera - possibly even on the device you wrote your stupid comment on!

          • gap_betweenus@lemmy.world
            link
            fedilink
            English
            arrow-up
            3
            ·
            4 months ago

            Sure, I even do photography professionally form time to time - I just don’t consider it to be a painting.

            • pirat@lemmy.world
              link
              fedilink
              English
              arrow-up
              2
              arrow-down
              1
              ·
              edit-2
              4 months ago

              But art, right?

              (edit for clarity: at least in some cases)

              • gap_betweenus@lemmy.world
                link
                fedilink
                English
                arrow-up
                3
                ·
                4 months ago

                Photography can be art as well as AI generated images can be art as well. AI is a tool and people can create art with it. But also what is art is completely subjective to the viewer.

    • ribhu@lemmy.world
      link
      fedilink
      English
      arrow-up
      8
      arrow-down
      3
      ·
      4 months ago

      That’s a blanket statement. While I understand the sentiment, what about the thousands of “AIs” trained on private, proprietary data for personal or private use by organizations that own the said data. It’s the not the technology but the lack of regulation and misaligned incentives.