from a technical standpoint, can you explain in a very layman term to a newbie like me on why AI is so bad in doing fingers but good enough to do faces? is it the lack of data of fingers compared to the vast amounts of faces ?
Hands are hard to draw, even for humans, lot of people take years to learn hand anatomy, so of course it'll be hard for computers to do it, when it doesnt even know what hands are.
Each finger has 3 joints, for 4 fingers, plus thumb with 2 joints, then also have to take into account curvature of fingertips in the direction they bend, and make it match the wrist joint direction. Theres a reason most people use 3D reference for it.
Definitely has to do with less reference, as you said, but also hard for AI to determine the location of all those factors above, in relation to each other, and to the main body figure. (AI also has issues with edge detection right now, which is why fingers usually look like blobs, and eyes will go through hair)
Faces imo are pretty generic in AI art, hence most of the time theyre decent, but without enough detail in prompts and variation in data, u can get really unique faces (though then you gotta replicate that unique face if u wanna use it in multiple scenes, as u stated before)
AI will figure out hands in a few years, hopefully at that point we'll be able to make stuff in unique artstyles, rather than steal from existing artists' artwork (personally as an artist I don't see AI art as a horrible thing, its really just a tool, will take a long time to replace artists, if ever, and even then, it would be used by technical artists to create the art, wont just run on its own)