Raw LLM Responses

It does feel like there is either a lot of bots posting these sorts of things or very junior people that can't gauge quality of the code exaggerating their expertise. In my org Claude 4 opus could not get anything through code review except the most trivia tickets, unless I spent more time trying to make it generate stuff appropriate for our code base than writing it myself. I'm considered an ai bro at my company too. I do think it's useful for boilerplate and random functions with a very narrow scope, but full on vibe coding a ticket? Even with all this planning stuff just doesn't seem good enough currently.