I wonder if it might be useful to "generalize" this advice at a higher level -- something like "ask ChatGPT a dumb question, you'll get a dumb answer." It's still a computer running code, designed to respond a certain way to certain prompts, which for me observing and learning what different kinds of prompts yield. I'll often say the equivalent of "let me re-phrase my question" when I realize, relative to what I wanted, that I "mis-prompted" it.