It's limited by the trends of human art. The art and text AI that we have are based on pattern processing. They output what is expected based on what we feed it. They aren't able to come up with entirely new styles or philosophies. They don't even have a cognitive ability to have any philosophy. An AI describing a tree or depicting an image of a tree doesn't have an understanding of what a tree is, they are not aware of the world, they can only replicate human words and images.

A breakthrough needs to happen for them to be capable of anything more, but that's going to be its own can of worms.