Will AI Replace Developers? A More Useful Question to Ask

"Will AI replace developers" gets asked a lot, and it's the wrong question — not because the answer is comfortable, but because it hides the more useful one underneath it.

The role that's shrinking is real

Writing well-specified, boilerplate-heavy code from a clear description is a task AI tools now do quickly and reasonably well. That's not a small category of work — a real share of junior and mid-level engineering time historically went into exactly that. Pretending that category isn't shrinking doesn't help anyone plan for what comes next.

The role that's growing is harder to automate

Deciding what shouldn't be built, judging which tradeoff a business actually needs, reviewing generated code critically enough to catch a subtly wrong assumption, and understanding why a system evolved the way it did — none of that has gotten meaningfully easier to automate. If anything, as more code gets generated faster, the judgment required to decide what's worth generating and whether it's correct becomes the scarcer skill, not a less important one.

The better question

Instead of "will AI replace developers," the more useful question for anyone actually planning a career or a team is: "which parts of this job are becoming commodity, and which parts are becoming more valuable as a result?" That reframes an anxious binary question into something you can actually act on — by deliberately investing in the judgment-heavy, architecture-level skills that are becoming relatively more valuable, rather than competing on typing speed against a tool that's already faster.

What this means for how we hire and grow engineers

Practically, this shows up as valuing engineers who can specify a problem clearly, review output critically, and reason about tradeoffs — over engineers whose main strength was fast, correct implementation of a well-defined spec. That's not a smaller skill set. It's a different one, and it's the one that was always the harder half of the job.