Raw LLM Responses

That anecdote about your friend using AI to cheat on her astronomy homework is sadder from a teacher's perspective. You assumed that the teacher wouldn't care because it was only a low-stakes assignment in an introductory class and that they were probably going to use AI to grade it anyway. As a teacher who knows a lot of other teachers, I think it's likely that teacher cares deeply about astronomy and painstakingly reads through each submission with care. It's also likely that seeing students pick up those foundational concepts and think through them in assignments like that is part of what gives the teacher's life meaning--it's how they know their life's work is having a positive effect in the world. So not only was your friend forgoing the possibility of learning something interesting about astronomy, she was contributing to turning what had been a fulfilling vocation for her teacher into a soul-killing exercise of sifting through AI slop that pretends to say various things but really only communicates one thing: the student who submitted it couldn't care less about the class. (And while she thought taking out the m-dashes etc. would make the ChatGPT output undetectable as such, remember that teachers read hundreds of responses to the same assignment, so they pretty quickly learn to identify the signs of AI "writing" in general and even the specific things it says when asked to respond to that specific assignment.)