You know you should be suspicious of results you get when you are looking around the world using an AI chatbot. You know this! And yet - they always seem so fully confident, like they are definitely sure of their answer - and that it's going to make your life easier.
Sigh.
I like using AI chatbots. They do make some things easier, and that's great. I'm just dubious about their writing skills (not great!) and their research skills (aspirational!).
And I found a suggestion that can be helpful. (With the research part; the writing part is still not great.) From the newsletter "The Comms Stack" came a good suggestion to help you get the best information from a chatbot:
- At the end of your next AI draft, paste this: "List every fact, number, name, and quote this draft relies on, with the source for each. Flag anything you can't source."
- Count how many come back flagged. My guess is it'll be more than you'd like.
- Make it a standing rule — in your custom instructions, your project settings, wherever your AI keeps its memory — so you never have to remember to ask.