A new research paper laid out ways in which AI developers should try and avoid showing LLMs have been trained on copyrighted material.

OpenAI now tries to hide that ChatGPT was trained on copyrighted books, including J.K. Rowling’s Harry Potter series::A new research paper laid out ways in which AI developers should try and avoid showing LLMs have been trained on copyrighted material.

  • Hildegarde@lemmy.worldEnglish
    102·
    1 year ago

    I defend the idea of copyright. The first copyright law was in 1710, to protect authors from the printing press. Without copyright, whoever owned the printing press would sell copies of books with no obligation to pay the author. When copying art is trivial, the artist needs copyright protection in order to make a living creating art.

    There are major problems with modern copyrights. Like all things in capitalism it has been subverted to benefit the rich, but the core idea behind copyright is sound.

    These lawsuits are not to stop the development if generative AI. These lawsuits are to stop the unlicensed use of copyrighted works as AI training data.

    There are AI models that are only trained with licensed data. This doesn’t stop the development of AI.

    Artists should have the right to choose whether their work is used as training data. And they should be compensated fairly for it. That will be the case if these lawsuits succeed.