Feels like a bit of a broad statement. I tend to request chunks of work/code that I place into my codebase. I still know what it's for and where it goes, I just care a lot less about the specific implementation details of the chunk, as long as it's typed and passed the smell test. To me it's just one higher rung on the abstraction latter.

But it goes to show how difficult it is to discuss this stuff, because depending on model, IDE, workflow, experience/seniority, intuition etc. LLM assisted developmental might look very different in each instance.