Sorry, OT: but I couldn't resist...
I asked a few persons around how could one identify GPT-generated text?
I recently heard a radio interview with a college professor who was changing his assignments (especially the standard 30-page paper) in the face of ChatGPT. His (semi-humorous) metric for detecting AI-generated content was the lack of spelling or grammar errors...
Rich
PS He did have a serious plan: ask each student to give a 15-minute presentation on the topic that would have been the subject of that 30 page paper. It takes the same time (or less!) to assess the work, and gives a personal connection with the students...