Definitely the former. Although the generated images can be interesting or amusing, I lean toward @peepeep’s other topic title: Post your dam AI art here and keep it out of the other threads, it's GROSS 🤮. I’m not saying that the AI tools aren’t useful or that they can’t produce intriguing imagery, but I maintain my previously stated belief that they tend to make mistakes that an actual human artist likely wouldn’t, especially when it comes to detailed elements of a composition.
It’s not terrible, and I can imagine people mistaking it for an actual photograph at a glance, but, again, where the AI fails is apparent in the details. For instance, look at the area of the shirt near the collar (the blue shirt that’s off-shoulder and draped on the back). The stripes aren’t continuous in the way that actual fabric would appear when photographed, and the AI has difficulty determining where the edge of the fabric should be versus where the fabric folds over itself, so there’s some unnecessary noise in the image there that appears as blur or artifact. It’s places like that—like where the AI seems to be unable to make a “decision” and commit—where there’s still a lot of room for improvement with these kinds of tools.
Part of it is gestalt. Part of it is experience with a variety of anatomic variants and knowledge of what does and doesn’t fall within the range of “normal”.