... Csathy argues that AI’s reliance on copyrighted content necessitates fair compensation for creators. He said the use of AI to generate content raises concerns about the loss of control and the potential devaluation of creative works. ...
Playing devil’s advocate, Tercek laid out some of the defense that Big Tech will use to justify training on copyrighted work as “fair use.”
It’s an interesting argument, and goes like this:
“The AI reads all the books, or looks at all the images, or listening to all the music and then that model begins to build parameters. A big question about this is, is it fair use? Is it okay to look or read or listen? There is no law that prohibits reading a book. There’s no law that says, You can’t learn by looking at a picture.” ...
It’s not replicating any of Van Gogh’s paintings but we have millions of incredibly precise, measurements about his paintings. Those are facts and facts can’t be copyrighted. What we can recreate in LLM is a factual representation of all those different values that go into creating that work.” ...
“What LLMs are doing is transforming those fixed works into something that is participatory, that billions of people can interact with to build new creative things,” he says. ...
See the full story here: https://amplify.nabshow.com/articles/ic-ai-copyright-law-robert-tercek-peter-csathy/?utm_source=substack&utm_medium=email