Go to your favorite image search engine and try to replicate anything AI has generated. It's impossible. The results you get from image search are nowhere near as specific as what the AI produces. It's not even close. The only time image search can compete is when you are highly specific, e.g. "painting of the Mono Lisa", both AI and image search will produce very similar results. But for a generic prompt, image search will come up with nothing that gets even close to the query, while AI can produce a highly specific image.

DreamBooth really should have destroyed all doubt about this point point, as the AI can generated highly specific images of subjects that aren't even in the original training set.