AI coding comes with the tradeoff that you lose understanding of your own code. That is an acceptable tradeoff when getting it out to users matters more than understanding code. But free of capitalistic pressures, why would you accept AI slop?
But if your coworker is shipping 5x more PRs than you your manager couldn’t care less about understanding code
There's the "inline editing" paradigm - independent sessions, describe each micro-change individually. This is good for quick responses but requires you to re-describe the context every time. There's also the "use an agent as a fancy code editor" stuff - just list out all the functions and the changes you want to make in them. This is a better workflow IMO but the trap is that it doesn't feel "productive" compared to one-shotting it, irrespective of the actual quality.
So people gravitate towards one-shot agentic prompts without a human in the loop.