I haven't experienced much help in the research department. That might become a wild goose chase if you're not careful.

For creative writing (this isn't limited to stories!) or slightly more contextual boilerplate, I love it. I think of it like a slight upgrade to what I already do, to allow me to get to the steps that really matter, not this revolutionary new world people are imagining it to be.

But how do you trust the output of the summarization? Do you trust yourself to identify when the model goes off the rails and starts making things up?

But I also don’t trust what humans write. All the information I get gets a critical eye. ML generated summaries or whatever make ML kinds of mistakes and it’s easy learning what to look for. Just like the kinds of mistakes journalists, random internet people, authors, scientists, etc. make their own brands of mistakes you learn to look out for.

I dispute that it's easy to discern when AI is making a mistake. ChatGPT has given me very plausible incorrect answers, and had I not been fact checking I would have been none the wiser. I could choose not to trust anything ChatGPT puts out, but if I'm verifying literally everything then it becomes less and less useful as a tool.