Of course they don't work. The whole point of LLMs is that they're indistinguishable from human-written text. People need to understand that AI isn't magic, it can't tell that something wasn't written by a human without a distinctive clue.