Is anyone surprised that a diffusion model trained on petabytes of human created artifacts could properly mimic human created artifacts?
I'd still argue that there's this AI'ish veneer that's just slimy enough to pickup and tell.
The real benchmark in human creativity is originality / novelty. These look like they came off of every half-decent designer's Pinterest board.