Raw LLM Responses

I don't think the music standard is a "double" standard so much as it is a different company holding themselves to a different standard. the only time AI directly copies an art style is the case of "overfitting" which at that point I agree there is a copyright issue, and seems to be the message of the quote early on which you reference, but over generalize the application to all AI generated art, not just the overfitting examples. from a legal perspective it will be interesting to see where it goes, the law is so bad at keeping up with these sorts of things, and it is unclear either in music, art, or code how this will settle in a legal sense. being familiar with the training of datasets I have trouble seeing how we can make a clear legal distinction between AI learning and human learning, unless it also pushes for human learning to require "opt in" under the protection. the reality for all of us is the economy needs to shift for a world with AI, and people do things they love, not necessarily in a need to turn a profit. no job is safe at the rate of progress we have seen in these systems lately. ChatGPT helps to emphasize this with a major leap in general application. (also trained on a lot of scraped data) and GitHub's copilot has similar lawsuits underway that it is unclear how the laws will settle on the matter but may establish precedent that applies to visual art generation as well.