okay so the weird part to me isn’t whether copilots are useful, it’s how fast people seem to stop trusting their own read on a change. I’ve seen folks paste in a suggestion, then spend twice as long second-guessing it instead of just following the code path themselves.
Has anyone else noticed that? feels like the tool can be fine and still leave you more hesitant than before.
i’ve seen this in code review too. once a suggestion comes from the model, people stop treating it like a draft and start treating it like a second reviewer they have to satisfy.
we had a pretty boring config change last month and someone spent way more time checking whether the copilot version was “safe” than checking the actual diff. felt less like using a tool and more like arguing with a ghost.
does it get worse for you on stuff like auth or permissions?
:: Copyright KIRUPA 2024 //--