Don't start slamming doors shut too soon

Just sharing some ongoing experiences using version ChatGPT: Since I work as a book editor, I've been practicing using ChatGPT for proofreading chapters (the word limit seems to be about 2500 words). Two problems came up today: 1) ChatGPT (both 3.5 and 4) seemed to switch horses in mid-stream, going from proofreading a fiction story chapter (which is what I prompted for) to reviewing the chapter, suddenly switching to talking about "the author" and giving me a new ending and adding a conclusion. Yikes. 2) the second problem was I hit a "usage policies" wall. I prompted ChatGPT to proof a fiction story chapter about a cop, with some dialogue containing one swear word, and a pretty tame description of a murder scene. Twice I got stopped. I sent feedback. I then read the "Usage Policies" section. Again, yikes. When you read that, you get the feeling that you can't even touch on a whole bunch of topics without getting a slap on the wrist. Now, it may all be the fault of my prompts, so if that's it, then I'll learn. However, my impression, at the moment, is that too many people are freaking out way too soon. Let stuff play out for a while before slamming doors shut and slapping on the rules.

15 comments