Carolyn Elefant has some good advice: “Check everything ChatGPT does.“

I have found two rules of thumb useful:

* Never trust ChatGPT advice on any important issue.

* Always ask ChatGPT for advice on every important issue.

I treat AI as a sort of super law clerk. I would always verify any law clerk work product before using it for anything important.

At the same time, AI frequently suggests ideas that would not have occurred to me, or comes up with improvements on my work.

My biggest problem is realizing when I’m dealing with something AI can help me with. Whenever I do, I ask.