If AI "slop" scares you, your PR reviews are broken

At least with a small contributor's team, the problem with AI PRs is that the (most of) feedback is useless.

In hand-written PR, I can write a comment about better way to do something, and my teammate will learn and use better methods next time. And moreover, they will generalize, understand the principles of efficient code (as practiced in our organization), and eventually have much better initial PRs. Sure, when a new person joins, initially the reviews take longer than writing the code myself, but this is an investment in the future.

In AI Slop PR, you can waste your time on most beautiful and clear explanation, but they will be ignored and next PR will have exactly the same problem. If you get lucky, the author will maybe remember some of the comments, but don't count on it, most likely they won't. They will paste the comment to AI, AI will apply the fix, and the author will move on, without learning anything.

This is incredibly demotivating, and wastes everyone's time. And the worst part, if you got unlucky and the person is very lazy, they will now be able to waste _tremendous_ amount of your time. Because before AI, at least there was a balance - a person making bad PRs used to likely spend more time fixing them than me reviewing them. And with AI, it's just the other way around.

("But wait!" you can can say. "I am not like that, I carefully review my code and pay attention to comments". In this case great, but you are not making "AI slop" then, you are making regular PRs. The "AI slop" rules do not apply to you.)

N/A