Make the AI smarter. I asked for a cartoon version of Vincent Van Gogh painting in a field and I got a lot of grassy fields and only 1 had a person in it and it definitely didn't look like Van Gogh. I asked for Alma Thomas in an art studio teaching students about art and none of the teachers looked like Alma and there were even some that were men.
I tried name art, where each letter of a name was represented by an object shaped like that letter. Did not work at all. Even with me spelling the name out and put into quotations, "Adam" was Adn when the image generated. All of the letters were plain letters, no objects. Instead objects were in the background surrounding the misspelled name.
This would work for plain, simple backgrounds but it doesn't seem to work for anything more than that. This needs to be way smarter.