I created a prompt where I tell ChatGPT to read the research [then I copy and paste the entire research into my prompt] and tell ChatGPT to answer the following 4 questions about the research in language lay people can understand. The 4 questions are; what was the aim of the research?, what was the method? What was the result? And how are the results important?
Basically it just gives me a more in-depth abstract. I find it suitable for the purpose I am using it for. I would not trust this method for scientific/academic research purposes.
Perhaps it's a bad sign that I don't have a use for ChatGPT, since it means I'm not commonly learning new technical things.... but I don't have much of a use case for it.
I'm not writing hundreds of business emails a day, I refuse to use AI for personal communication or creative writing (Aside from spelling and grammar checkers, non text stuff, and things like that).
Code review seems to be the only real application. I might use it to find a bug, but most of my bugs are like, 10 file multithreading legacy code type things.