I trained an AI in PyTorch to scroll through the Canvas of Babel, a collection of every single possible image ever made or that ever will be made, and save any images it deems normal-looking. It zooms through thousands of pictures a second and it's a project I've been writing for about a week. Read about it or try your luck and run it yourself on the GitHub page: https://github.com/DvorakDwarf/Witness-of-Babel