Even granting your girlfriend’s point, it’s still true that AIs’ image generation capabilities are tilted more towards landscapes (and other types of images) and away from hands, compared with humans, right? I mean, by the time that any human artist can create landscape images that look anywhere nearly as good as the ones in my example, they would certainly not be drawing hands as bad as the ones in my example (i.e., completely deformed, with wrong number of fingers and so on).
By the time a human artist can create landscape images which look nearly as good as those examples to humans, yeah, I’d expect they at least get the number of fingers on a hand consistently right (which is also a “how good it looks to humans” thing). But that’s still reifying “how good it looks to humans” as the metric.
Even granting your girlfriend’s point, it’s still true that AIs’ image generation capabilities are tilted more towards landscapes (and other types of images) and away from hands, compared with humans, right? I mean, by the time that any human artist can create landscape images that look anywhere nearly as good as the ones in my example, they would certainly not be drawing hands as bad as the ones in my example (i.e., completely deformed, with wrong number of fingers and so on).
By the time a human artist can create landscape images which look nearly as good as those examples to humans, yeah, I’d expect they at least get the number of fingers on a hand consistently right (which is also a “how good it looks to humans” thing). But that’s still reifying “how good it looks to humans” as the metric.