Generative AI Has a Visual Plagiarism Problem - eviltoast

Experiments with Midjourney and DALL-E 3 show a copyright minefield

  • TheFriar@lemm.ee
    link
    fedilink
    English
    arrow-up
    10
    arrow-down
    2
    ·
    10 months ago

    Artists: they’re stealing our work and building a company on it!

    This outlet, years later: breaking news! Generative AI is hurting Disney’s bottom line by making Star Wars characters!

    • Cyber Yuki@lemmy.world
      link
      fedilink
      English
      arrow-up
      6
      arrow-down
      1
      ·
      10 months ago

      It’s only a problem when a corporation’s greed hurts another corporation’s greed. Until then it’s just business as usual.

  • Peanut@sopuli.xyz
    link
    fedilink
    English
    arrow-up
    2
    arrow-down
    1
    ·
    edit-2
    10 months ago

    And you violate copyright when you think about copywritten things alone at night.

    I violate copyright when i draw Mario and don’t sell it to anybody.

    Or these are dumb stretches of what copyright is and how it should be applied.

    the reasoning in this article is dumb and all over the place.

    Seems like gary marcus being gary marcus.

    Already seen openAI calling out some of the bullshit specifically noted in this. That doesn’t matter though, damage is done and people WANT to believe ai is terrible in every way.

    Everyone is just deadfast determined to climb onto the gary marcus unreasonable AI hate train no matter what.

    • test113@lemmy.world
      link
      fedilink
      English
      arrow-up
      2
      arrow-down
      1
      ·
      10 months ago

      The funny thing is, copyright doesn’t even matter; at least half of the world’s market couldn’t care less about copyright, especially if it’s from the “west.” They certainly won’t suddenly start respecting copyright law. They will use and develop AI without the restriction of copyright. All this talk about copyright and the law, and all the copyright suits against AI and tech firms, will be fruitless since we either forget copyright like we used to know it, or we get left behind in development because we need to respect the copyright of everything and make contracts with every big outlet, etc. Big tech knows that, so they walk this gray zone walk to still train AI on copyrighted material but somehow proclaim they are not copyright-dependent.

      I’m not saying this is a good development, just that I think we need to reassess how we treat copyright on a fundamental level under the current development structure of AI.

      We need to slow down the development of AI and hinder monopolization of the market. My guess is it’s too late, but we can still hope that maybe this time it will be different.

      • Peanut@sopuli.xyz
        link
        fedilink
        English
        arrow-up
        3
        ·
        edit-2
        10 months ago

        I definitely agree that copyright is a good half century in need of an update. Disney company and other contemporaries should never have been allowed the dominance and extension of copywrite that allows what feels like ownership of most global artistic output. They don’t need AI, they have the money and interns to create whatever boardroom adjusted art they need to continue their dominance.

        Honestly I think the faster AI happens, the more likely it is that we find a way out of the social and economical hierarchical structure that feels one step from anarcho-capitalistic aristocracy.

        I just hope we can find the change without riots.

        • test113@lemmy.world
          link
          fedilink
          English
          arrow-up
          3
          arrow-down
          1
          ·
          10 months ago

          Well, I hope you are right. xd

          It just seems to me like a monopolization of the market by the big tech corps, which won’t be beneficial to the majority, but at least a few billionaires will get richer.

          I was recently invited to the Google research center where they presented their new AI assistant features, which should be coming this year. It was weird; it was at the same time more capable than I thought and more restrictive than one would assume. It’s like not even Google knows exactly what to do with it, or what it should be able to do, or what exactly it is capable of. I also once got to try an “uncensored” / “unrestricted” information model, which was actually a bit scary but far more useful than any of the current “restricted” chatbots. I’m sure AI will change things up, but how, when, and why I don’t know, and the more I find out, the more unsure I am about predictions, besides the one that big corps will try to monopolize the market.