Raw LLM Responses

Let me start by saying I'm a professional artist with decades of experience, and I too am represented in the training data. My problem here is that much of what you've said is demonstrably factually incorrect: AI does _not_ take data directly from the data set - there are between two and five stages of removal - and the existing models are absolutely not able to create a perfect replica - though, if you "cheat" to influence the results by using an image prompt, you can get awfully close. That, though, is like saying a photo copier can create a perfect replica... A human has to put the effort into making it happen, first. Likewise, if someone uses an artist's name in a prompt and generates images that can pass as that artists work, and does something fraudulent with those images, it's the person doing those things whose done something wrong. You can create a copy or a forgery of someone's art in Photoshop, but we've stopped wailling about Photoshop being the end of art some decades ago. This is the same thing. Whether you understand or agree with the process, the closest functional analogy IS someone learning art by looking at art we've publicly displayed. The model this trained contains NO copyrighted art... Only algorithms the training models derived from patterns in that art. My bottom line is that putting forward objections that rely on either innocent false assumptions or intentional false assertions may rally angry supporters to the flag in the short term, but they're only going to help the corporations in the long term, once it's publicly proven that those assertions were false.