@whitequark I’m also in the (3) camp. Multiple times now I’ve run across some terrible code during reviews and I’ve asked if AI was involved in the process. The answer is generally yes, and the reason given is that the developer didn’t understand the code so they used AI to help. I suspect when I get a No, it’s still really a yes.

I greatly resent the use of my time to gatekeep AI slop. In one extreme case, the code wouldn’t even compile. The function started, then mid code, not even in a comment, “rest of function goes here”.