Raw LLM Responses

I'm a computer scientist, and it saddens me to see the toxicity from users toward artists, derived from the existence of these AI models. There's no justification for that, and there will never be. The way current AIs work is definitely too close to the source and that needs to be taken into account for evaluating the ethics of those models, before we go out of our ways to monetize and capitalize on these models. I do hope that the laws will catch up quickly with this, and people will be prevented from continuing to capitalize on models trained using copyrighted data obtained without a license. As you mentioned the models *could* be trained exclusively on open-source and public domain images (I don't like the term copyright-free, because in most cases, it's not true -- having a permissive license doesn't cancel out copyright, only public domain does, and public domain doesn't exist in all countries), although this will very definitely result in much lower quality models with lower quality output -- not because permissive licensed art is inferior, but because there's less quantity of it. That said, I do believe that with enough evolution of the technologies used for AI art, in the future we could get to a point where the process used by AI will be unique and original. But this will require that the models learn about shapes and colors, instead of learning how to predict what an image looks like based on a prompt and random noise. I do not believe that creativity is a special thing that only fleshy brains can have, but rather an emergent property of the complex abstraction that our brain does, which is orders of magnitude more nuanced than current AI models (regardless of their size) can hope for.