> Humans are really very similar to LLMs... So true. I was actually once accused on Reddit of using ChatGPT to write a comment that I genuinely wrote myself, b

> Humans are really very similar to LLMs... So true. I was actually once accused on Reddit of using ChatGPT to write a comment that I genuinely wrote myself, by a person who (if I recall correctly) didn’t like the substance of what I was saying. It was a long detailed response with multiple paragraphs and proper usage of words like “However”. Yes, a LLM may write similarly to people who have lots of experience reading and writing formal written English, including academic and legal writing. At least my example was purely a harmless Reddit accusation. I really feel bad for the people who write essays in school that the teacher incorrectly rejects and reports as AI-written because the teacher misunderstands the limitations of AI detectors and the meaning of their output. (Of course, this is a separate question from students who genuinely try to pass off AI-written work as their own.)