An idle and I'm sure unoriginal thought on AI codegen: most (human-generated, in-industry) code already sucks. The selection bias on fedi means people _here_ are intentional and careful and horrified by the slop. And folks who've spent their careers at marquee tech firms may have a similar perspective. But tech is big and most tech is broken most of the time. I'm not sure the bots are going to change the amount of broken code meaningfully. (This is not at all an endorsement, to be clear)