Every proprietor of an AI interface makes it in his own image, more or less.
I was playing with Grok a bit last night and generated an image of a character in my books, but I quickly found that I just don’t have that much of a use for it. It’s still way too hard to get it to produce precise results, and my mind’s eye is still better at filling in certain details. In one case I was absolutely floored by the results (though it still got certain details wrong), but the rest were just meh.