Over the past year, I handed all repetitive tasks to AI: CRUD, unit testing, refactoring, initial documentation. Now, I spend nearly 40% less time writing code by hand, but twice as much time on "thinking through requirements" and "picking apart AI outputs."

AI produces code quickly, but edge cases often break, and business context frequently disappears. At this point, if you don’t make the call, no one will back you up.
A possibly counterintuitive view: software engineering isn’t dying, it’s polarizing. Entry-level coding jobs will shrink, but system design, complex decomposition, and cross-domain integration will become more valuable.
The gap between those who use AI well and those who don’t may become as significant as the previous gap between those who could write code and those who couldn’t.
But what truly widens the gap isn’t fancy prompts; it’s whether you understand the business, can judge trade-offs, and dare to interrupt the AI when it’s talking nonsense.
A CS degree won’t become outdated; in fact, it’s more important than ever to have a solid foundation. AI is an amplifier, not a wish-granting machine.
People without systematic thinking will just accumulate technical debt faster with AI.
In the future, teams might consist of one veteran plus a bunch of agents. But that one person must be a product, architecture, and quality gatekeeper all at once. The threshold will only get higher.
How much has AI changed your work? Has it really made things easier, or just turned 10 low-level bugs into 20? Let’s talk.