We Asked A.I. to Create the Joker. It Generated a Copyrighted Image.::Artists and researchers are exposing copyrighted material hidden within A.I. tools, raising fresh legal questions.

  • Jilanico@lemmy.world
    link
    fedilink
    English
    arrow-up
    1
    ·
    9 months ago

    I really do not believe midjourney is storing all the files on the Internet in a humongous database. They are just exposing the AI to them for training just like you expose your computer to them when you visit with your web browser. I’m happy to be wrong about this, but I’m just not convinced. Please try to keep your temper.

    • dragontamer@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      9 months ago

      I really do not believe midjourney is storing all the files on the Internet in a humongous database.

      Are you sure?

      The training weights can literally recreate images its been trained on. That makes them a humongous database, albeit with lossy compression. They aren’t replicated exactly, but they are replicated enough that I’m confident that these “Joker” images passes as copyright infringement before a jury (ie: is substantially similar).

      • Jilanico@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        9 months ago

        That makes them a humongous database

        I understand why you’re saying that, but I personally don’t agree. As AIs face legal challenges, maybe your opinion will be adopted, we’ll see.