> I tapped into Pangram. Pangram is a remarkably good, conservative model for detecting LLM-generated text. These detectors have a bad rep among techies, but th

> I tapped into Pangram. Pangram is a remarkably good, conservative model for detecting LLM-generated text. These detectors have a bad rep among techies, but the objections are often based on outdated assumptions Turing test is really in the rearview, huh? Humans need machines to detect if a machine wrote the text, because humans aren’t sure.