I propose to consider the question, "Can machines think?"
A. M. Turing, Mind, October 1950
The AI passages here were written by an 8-billion-parameter open-source model (Qwen3-8B-Base) that was fine-tuned (LoRA) on the public blog writing of a well-known ML author, then trained further with reinforcement learning. The human passages are real excerpts from that author's public posts.
Half the passages are his; half are the model's. Rate each one. Your score appears at the end; the running average across all visitors is on the results page.
The passages are on different subjects and share no text, so there is nothing to compare against: each one has to be judged on its own. Nothing is revealed until you submit all ten.