If you’re working with a massive complicated C++ repository, you have to take the time to collect the right context and describe the problem precisely enough. Then you should actually read the code to verify it even makes sense. And at that point, if you’re a principle level developer, you could just as easily do it yourself.

But the situation is very different if you’re coding slop in the first place (front end stuff, small repo simple code). The LLMs can churn that slop out at a rapid clip.