[amlib]

So you are telling me you can always properly infer what the "black boxes" the AI auto-completed for you does without reading the documentation?

I bet you are most likely just blindly trusting the AI response and moving on. Sure, the code structure might checkout and the calls it completed are sometimes fairly generic/predictable, but there will be plenty of situations where the behavior is just different enough or the black boxes are something you have no idea about what it even does and you are too lazy to check the docs and commit the code anyway.