Comment on Japan determines copyright doesn't apply to LLM/ML training data

• camelbeard@lemmy.world ⁨5⁩ ⁨months⁩ ago

  If you read a book you can talk about it, quote it, draw characters from that book, write your own ending, etc.

  Isn’t that kind of the same? Let’s say some day we have an AI with near human intelligence, why can’t the AI be trained on copyright works, just like humans, all our school books are copyrighted works?

  source

  • Geobloke@lemmy.world ⁨5⁩ ⁨months⁩ ago

    Because you paid for that book?

    source

    • camelbeard@lemmy.world ⁨5⁩ ⁨months⁩ ago

      So if AI companies pay for a book or music (like a consumer) it’s no problem? Because I don’t think this is about paying for content, it’s that content holders refuse to work with AI companies.

      source

      • Mustard@lemmy.blahaj.zone ⁨5⁩ ⁨months⁩ ago

        Unironically yes, if AI companies paid for training data everyone would be much happier.

        I sincerely doubt that NOBODY is willing to sell data to them. It’s far more likely that they have not offered anyone a fair price yet, which makes sense because that would set a precedent.

        Even then, if people don’t want to sell them their copyrighted work then tough. You can’t compel people to take customers they don’t want.

        source

        • -> View More Comments
• General_Effort@lemmy.world ⁨5⁩ ⁨months⁩ ago

  Why do you think you can’t? What do you think was illegal?

  source
• lloram239@feddit.de ⁨5⁩ ⁨months⁩ ago

  You can. Distributing copies is illegal, not downloading them. That’s why torrents are bad and streaming sites are fine. (Some exceptions might apply depending on your country).

  source