Raw LLM Responses

Diffusion models do generate each piece of art from a blank slate. They start with an image that's pure noise, and gradually 'de-noise' it until it fits the prompt given, just using the machine learning they've undergone. This is radically different than starting with a specific artist's work and doing some photoshop magic on it. Which is to say, the process is actually creative and not purely derivative. And the fact that humans have to put effort into processing an image when they study references--while an AI does it much more easily--really shouldn't matter. If I'm planting seeds by hand, I might put a tremendous amount of effort into the process, but we might get the exact same result, or even a better result, from an automated tractor. Should we stop using tractors because they remove the human touch from planting crops? But that said, the amount of work people put into their art will always be impressive, it's something to think about that goes along with the art. Much in the same way that even while we have perfect chess-playing computers, we still want to see actual humans playing the game, it's just cool to see how far people can take their abilities. The issue here should just be the data, not the process. And even on that note, I don't really believe in intellectual property, since IP will always be abused to favor corporations. The only thing that should matter is guaranteeing the basic amenities of life to artists (and all people in general). What stands in the way of doing that is capitalism, not any one technology.