Origami world.

To be honest, I was thinking about lots dots - no easily inferred shapes, but pretty obvious for a human. If the state-of-the-art software using neural nets and what not is capable of reading even *letters* made up of many dots (circles) in different colors, shapes, and textures (and with many "false" dots in colors that blend in the background for the human eye), then the programmers are much more advanced than anything I read about.

Computer generates a random 3D object out of large polygons with fairly sharp angles of contact, subject to various limits on the way in which the object is generated. Displays 2D image of 3D object.

Human infers 3D object from 2D image, infers unseen portions of the image from rules by which the 3D image is generated -- for example that the object must make sense mechanically -- that it should be stable resting on a plane.

I don't know how well people would manage to do that. One requirement was to be easy for a human to answer the puzzle. Mark