Comment on Authors Are Furious After Finding Their Works on List of Books Used To Train AI

• PsychedSy@sh.itjust.works ⁨1⁩ ⁨year⁩ ago

  AI isn’t either. It’s selling statistical data about the books.

  source

  • mojo@lemm.ee ⁨1⁩ ⁨year⁩ ago

    It literally shares passages verbatim

    source

    • BetaDoggo_@lemmy.world ⁨1⁩ ⁨year⁩ ago

      So does any site that quotes the book. Just being trained on a work doesn’t give the model the ability to cite it word for word. For most of the books in this set you wouldn’t even be able to get a single accurate quote out of most models. The models gain the ability to cite passages from training on other sources citing these same passages.

      source
    • lloram239@feddit.de ⁨1⁩ ⁨year⁩ ago

      It shares popular quotes from books, it can’t reproduce arbitrary content from a book. The content needs to be heavily duplicated in the training data to stick around, and even than half of it might still end up being made up on the spot.

      Also request for copyrighted content will be blocked by ChatGPT and just receive the stock “I can’d do that” response anyway.

      If you have some damning examples that show the opposite, show them.

      source

      • BURN@lemmy.world ⁨1⁩ ⁨year⁩ ago

        Being blocked by ChatGPT just means that the interaction layer you see doesn’t show the output, not that the output wasn’t generated.

        Everything you see that’s public facing and interfacing with an AI is an extreme filtering layer for what is output. There’s tons of checks that happen to ensure that they don’t output illegal content or any of a million other undesirable things.

        source
      • mojo@lemm.ee ⁨1⁩ ⁨year⁩ ago

        I’m too lazy and care too little but you can basically get it to roleplay as a book expert or something and to “remind” you of certain passages. It gets around the filter pretty easily, that’s how jailbreaks work.

        source
    • Piecemakers3Dprints@lemmy.world ⁨1⁩ ⁨year⁩ ago

      That claim is disingenuous at best, and misinformed otherwise.

      source
    • PsychedSy@sh.itjust.works ⁨1⁩ ⁨year⁩ ago

      That’s maybe an issue. I mirror speech a lot, though. How large are the passages?

      source
  • LtLiana@startrek.website ⁨1⁩ ⁨year⁩ ago

    “I’m not reselling your book, I am selling a machine that holds a mathematical formula that partly represents your entire book word for word and can reprint it on command!”

    source

    • FaceDeer@kbin.social ⁨1⁩ ⁨year⁩ ago

      LLMs can't reprint their entire training data on demand. They rarely even remember quotes.

      source

      • kromem@lemmy.world ⁨1⁩ ⁨year⁩ ago

        Don’t bother shouting into the AI misinformation void.

        People aren’t going to put down their pitchforks and torches to brush up on basic ML principles and it’s just going to frustrate you engaging.

        It’s going to be a non-issue within 24 months anyways.

        No matter how the OpenAI court cases land, the writing is on the wall that the next generations of models are going to be built on the backs of synthetic data, which is inherently without copyright.

        At best rulings against OpenAI mean a secondary market emerges in China for repackaging copyrighted data into synthetic data of equivalent value to help buffer SotA synthetic data in avoiding model collapse.

        It’s not even going to end up amounting to a minor speedbump to progress by the time the court cases are finalized.

        Let the armchair activists rant and rage and tire themselves out worrying about a fabricated version of reality, and just focus more on staying informed about actual reality for yourself when all this passes.

        It will be years before people eventually drop the bias against AI we self-instilled from shortsighted Sci-Fi over the past few decades, and until then the average person online will be irrationally upset about something related to the tech. Might as well run themselves ragged over the misinformed “it just remixes copyright” in the meantime.

        source

        • -> View More Comments
    • PsychedSy@sh.itjust.works ⁨1⁩ ⁨year⁩ ago

      I mean, yeah? They were running to a concrete description. That is not valid. My brain has most of Terry Pratchett’s works.

      source