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

  • FluffyPotato
    link
    fedilink
    English
    arrow-up
    8
    arrow-down
    5
    ·
    7 months ago

    The copyrighted material has no business being used for AI training anyways. Like why train AI to write books or make art when you could feed it documentation and teach it data entry instead like stuff that would actually work well for this and wouldn’t require copyrighted works.

    • ryathal@sh.itjust.works
      link
      fedilink
      English
      arrow-up
      1
      ·
      7 months ago

      Scanning documents, recognizing text and populating a form is already a feature for document scanning. It’s not labeled as ai, because it’s already commercially viable.

      • FluffyPotato
        link
        fedilink
        English
        arrow-up
        1
        ·
        7 months ago

        I mean feed it documentation for software so if you need to do something obscure in like the Outlook server you can ask for a step by step guide. They need to fix the hallucination thing for that though, nonsense responses to technical questions are far too common for me to trust it currently.