Using AI agents “hurts the ego,” he admits. There’s something inherently diminishing about going from creator to director, from craftsman to critic.
Yet he finds the trade-off worth it because coding with AI is “more fun” — the drudgery removed, leaving only the creative architecture and problem-solving.
This reframing is crucial. The future of software engineering may not be about writing better code, but about asking better questions. The engineer becomes a product manager, systems architect, and quality assurance specialist rolled into one — directing AI labor rather than performing it.
Those without these foundational abilities may find themselves unable to direct AI agents effectively, widening the gap between elite performers and average developers. The 10x engineer becomes a 100x engineer. The average engineer risks becoming obsolete — not replaced by AI, but by other engineers wielding AI more effectively.
We’re witnessing what might be looked back on as the moment software engineering fundamentally changed when the profession shifted from manual craft to AI orchestration.
https://blog.stackademic.com/the-end-of-manual-coding-d801dd05c3f3
Steve Jobs had a great outlook: It’s about taste. AI is the greatest amalgamation, and that mediocratic generalization inherently has no taste because it reflects and embodies every taste.
The thing to remember is that senior software engineers are not paid to write code. They are paid to build well designed, maintainable systems that meet the application need. This can be done faster with an llm. So to not use it is malpractice. But not carefully reviewing the code is worse.