The GAN Is Not the End Sort of Product, Is It?

The GAN has little to do with what the image represents, any sort of meaning encoded into it. This is visual information reduced to pixel values and weighted as statistical elements that have likelihood. It's analyzing all of those human faces in order to say this cluster is usually in the center. And so you and I would call that a nose. The GAN would say this is a sequence of pixels that appear in the center of the image. If this sequence is here, very often to the top left, top right are this other sort of identical sequence of pixels, which we would call eyes.