As models get better and better at doing the stuff users are asking it do to, the more and more they trust it and grant it increased privileges.
Back when AI was new, I needed to check every single output because it was 50/50 on the code being complete junk. Now 99% it's correct first time and 1% close but needs some tweaking. I'm lazier and lazier WRT granting permissions and just trusting it to go nuts.
Back when AI was new, I needed to check every single output because it was 50/50 on the code being complete junk. Now 99% it's correct first time and 1% close but needs some tweaking. I'm lazier and lazier WRT granting permissions and just trusting it to go nuts.