Its hard to explain but if theres a distinct disconnect between high fidelity renders and the compositional/subject matter choices. If a human is rendering at the level AI usually samples from, they tend to have a level of panache to their artistic choices that the non artist prompter lacks.