GenAI tools ‘could not exist’ if firms are made to pay copyright | Computer Weekly
www.computerweekly.com
Artificial intelligence firm Anthropic hits out at copyright lawsuit filed by music publishing corporations, claiming the content ingested into its models falls under ‘fair use’ and that any licensing regime created to manage its use of copyrighted material in training data would be too complex and costly to work in practice

GenAI tools ‘could not exist’ if firms are made to pay copyright::undefined

  • JackGreenEarth@lemm.eeEnglish
    93·
    10 months ago

    I see way too many people advocating for copyright. I understand in this case it benefits big companies rather than consumers, but if you disagree with copyright, as I do, you should be consistent.

    • Optional@lemmy.worldEnglish
      93·
      10 months ago

      Copyright law should benefit humans, not machines, not corporations. And no, corporations are not people. Anthony Kennedy can get bent.

      • JackGreenEarth@lemm.eeEnglish
        64·
        10 months ago

        I hate the MAGMA companies as much as anyone, but AI such as LLMs, especially the open source stuff Facebook and Stable diffusion is making, is beneficial to us all.

      • LainTrain@lemmy.dbzer0.comEnglish
        53·
        10 months ago

        Abolishing copyright in the way that allows for the existence of Gen AI benefits people far more than it does corpos

    • General_Effort@lemmy.worldEnglish
      62·
      10 months ago

      You don’t have to be against copyright, as such. Fair Use is part of copyright law. It exists to prevent copyrights from being abused against the interests of the general public.

      • JackGreenEarth@lemm.eeEnglish
        42·
        10 months ago

        But I am against any copyright beyond forcing attribution to the original creator.

          • JackGreenEarth@lemm.eeEnglish
            31·
            10 months ago

            AI creators, at least the open source ones, are usually pretty open about where they got the training data for their model

          • Zoboomafoo@slrpnk.netEnglish
            31·
            10 months ago

            Here’s your works cited for any generative AI:

            Humanity. “The Entire Publicly Accessible Internet .” The World Wide Web, , 1 Jan. 1983, WWW.org.

          • assassin_aragorn@lemmy.worldEnglish
            11·
            10 months ago

            At the very least, every AI should be able to spit out a comprehensive list of all the material it used for training. And it should be capable of removing any specific item and regenerating its algorithm.

            This is a fundamental requirement of the technology itself to function. What happens if one the training materials has a retraction? Or if the authors admit they used AI to generate it? You need to purge that knowledge to keep the AI healthy and accurate.