A year ago, the loudest AI skeptics in software engineering were junior developers worried about job security. Now the anxiety has shifted upstream. Senior engineers with a decade or more of experience are hitting an identity crisis: they built their careers on craft, and suddenly the tools want to do the crafting for them.
The tension is real and specific. A senior engineer who spent years mastering debugging, advocating for better tooling, and mentoring others now watches an AI assistant produce functional code in seconds. The code works. It passes tests. It is also, frequently, code they would never have written that way. The question is no longer "does AI help?" but "what am I if AI does the part I was good at?"
This is not a technical problem. It is a values problem. Engineers who identify with craftsmanship feel a genuine loss when the craft gets automated. Engineers who identify with shipping product see AI as the strongest tool they have ever had. Most people are both, which is why the feeling is more like vertigo than resistance.
The practical reality is that AI coding tools like Cursor, Claude Code, and GitHub Copilot are not replacing senior engineers. They are replacing the mechanical parts of the job while making the judgment parts more important. Deciding what to build, how to architect it, what trade-offs to accept, and when the AI's output is subtly wrong are all senior-engineer skills that matter more now, not less.
The engineers who will struggle most are those who mistake the tool for the job. The ones who will do best are those who treat AI like they treated every other powerful tool in their career: learn it, use it where it is strong, override it where it is weak, and keep their own judgment sharp.