CMV: There is no reason AI companies should not be able to train their models on copyrighted materials.
In principle, I don’t see why AI companies shouldn’t be able to train models on copyrighted material, provided the AI does not ultimately produce outputs that infringe anyone’s copyright.
Don’t humans learn the same way? Everything someone creates has been influenced by things they’ve watched, read, or heard. As long as the final product doesn’t violate anyone’s copyright, then there is no issue. I don’t need permission from Dreamworks to watch Shrek, get inspired by it and create my own movie, video, or drawing that uses some aspects from Shrek.
My understanding is that AI is learning patterns, sort of like humans do. It learns the general structure of language, music, writing, and art rather than reproducing the specific works it was trained on.
I notice a lot of people online are super against this, why? What am I missing? If AIs are learning general patterns and not reproducing the works they were trained on, why should it be treated different from human learning/inspiration?