@Ironwood @volpeon
My understanding is that the typical artist gains experience of drawing things from life, drawing things from picture reference, looking through the work of other artists, and sometimes even taking thorough notes on other people's art styles, and experimenting.

A neural network is a simplified mathematical recreation of a biological 'mind' (at the level of calculating the outputs of each synapse) that learns through experiences as well, and is capable of intuiting new content based upon it's learned experiences of different concepts.

Further: no artist creates in a void, an artist is a consumer of variety of other works, sometimes creating remixes or transformative works of content they like (game characters, character in a movie, specific type of monster or werecreature, etc).

What deviates so much to make the situation different of a neural network intuiting new works from it's experiences, versus a human artist doing the same? Maybe in the situation of someone using a very small training set maybe, but in that situation you wouldn't get really useful nor coherent results.