I can't think of a kinder way to put this ... so you know the saying "there are no stupid questions"? Yeah, that has never been true, and this is a great example of why.
Well, I guess there are indeed stupid questions, but I don't think this is one.
(Perhaps I'm using the term "derivated" incorrectly - I'm not a native speaker.)
What I mean is:
Generative AIs produce things by some type of command (currently simple prompts) and for the production process use a model trained with a lot of data. The result is new in a sense that it didn't exist before in this same way.
Humans produce things not necessarily by commands, but by various motivations, e.g. they need to produce something for an employer, or they just want to do it themselves for joy.
For the production process humans use their experience, not only from works of art from the same category/modality (e.g. images), but general (multi-modal) world "knowledge", including their own emotions. This all (or parts of it) come together resulting in a work of art, which, same as with an AI, didn't exist before in this way.
So in both cases the art is derived from (multimodal) input.
The interesting question is, if the human brain (what else?) is able to actually create something new that is not somehow a (partly randomized) product of these inputs.
For example can a human create an actual new style of art that hadn't been done before (like Dalí, Picasso, Gaudí etc.)?
And if yes, can an AI also do this?
If yes, what then is it that humans can do, but AIs can not?
Or asked differently: which special capability do humans have in creating things, which we're more or less sure that AIs don't also have and never will have?
And don't take the current state of generative AIs as given forever. That's only the beginning. Feeding them more data, especially in different modalities, will result in a lot more sophisticated models and results.
As I've said before, I'm not an expert when it comes to human creativity and art, and I'm sure smart and experienced people have been thinking about stuff like this for a longer time. And I'm genuinely interested in what they think about it.