A rather militant approach to testing for the most part, even if I don't know how to do something off the top of my head I know what I want my code to be doing so in my opinion having decent testing in place limits the risk of a ChatGPT misinterpretation causing harm. I've not done away with documentation wholesale either, if I'm in doubt I'll revert to the usual way of doing things.

I use them in conjunction with each-other as well, I can use ChatGPT as a 'really fuzzy find' for things like 'what's the name of x function in a library that sounds a bit like y whose name I've forgotten' then go and look at the docs in a way search engines are generally quite bad at.