AI Will Change Every IT Job — and We'll Need More Developers, Not Fewer
AI makes building software easier for everyone. That's exactly why the world will need more developers in a few years, not fewer. Here's why.
Every few months someone announces the end of programming. Meanwhile, the world is producing more code than at any point in history — and someone has to own all of it.
Here’s my bet: AI will make development much simpler. And in a few years, we’ll need more developers than we have today.
Building got cheap. Owning didn’t.
Look at what one year of AI-assisted development looks like on GitHub alone. In the Octoverse 2025 report, developers pushed nearly 1 billion commits (+25% year over year), created over 121 million new repositories, and 36 million new developers joined the platform.
That’s an illustration, not exact data. But the direction is real: the amount of software being generated right now is enormous. Most of it is written quickly, reviewed lightly, and deployed somewhere it will live for years.
Every one of those lines is a future bug report, security patch, dependency upgrade, or migration. Writing code got cheap. Owning code is as expensive as it ever was.
AI code works — until it doesn’t
Here’s a function an AI will happily write for you. It looks fine, the types check, and it passes a quick test.
function daysUntil(deadline: string): number {
const diff = new Date(deadline).getTime() - Date.now();
return Math.floor(diff / 86_400_000);
}
console.log(daysUntil("2026-12-24"));The bug: new Date("2026-12-24") is parsed as midnight UTC. For a user in New York, that’s still December 23rd. Your countdown is off by a day for half the planet.
function daysUntil(deadline: string): number {
const [year, month, day] = deadline.split("-").map(Number);
const target = new Date(year, month - 1, day);
const today = new Date();
today.setHours(0, 0, 0, 0);
return Math.round((target.getTime() - today.getTime()) / 86_400_000);
}Now multiply that by a billion commits. Somebody has to know why the second version is right. That somebody is a developer.
How every IT job changes
AI won’t remove IT jobs. It will change what each one spends its day on. The boring, repetitive part shrinks. The judgment part grows.
| Role | AI simplifies | Still needs a human |
|---|---|---|
| Developer | Boilerplate, CRUD, tests, refactors | Architecture, trade-offs, owning production |
| QA | Writing test cases, regression runs | Knowing what can actually break |
| DevOps | Config files, pipelines, scripts | Incidents, cost, reliability decisions |
| Security | Scanning, first-pass triage | Threat models, a growing attack surface |
| Product & design | Prototypes, mockups, specs | Deciding what's worth building |
| Support | Answering common questions | Hard cases and angry customers |
Notice the pattern. The right column doesn’t shrink when there’s more software. It grows.
Why cheaper software means more developers
This has happened before. Compilers, frameworks, cloud, open source — every time building got easier, we didn’t need fewer developers. We built far more software, and needed more people to run it.
Economists call it the Jevons paradox: when something gets cheaper to use, people use much more of it. Software is the textbook case.
Even with AI in the picture, the US Bureau of Labor Statistics still projects 10% growth in software developer jobs from 2025 to 2035 — faster than the average for all occupations.
And there’s a timing effect. Today, lots of quick AI-built apps are still young. In two or three years they’ll hit real users, real scale and real security audits. That’s when companies will look for people who can understand, untangle and rewrite them.
What changes for you
The job is moving up a level. Less typing, more deciding. To stay valuable, focus on the things AI is worst at:
- Reading code fast. You’ll read far more than you write. Get good at it.
- Understanding systems. How data flows, where state lives, what happens when a service is down.
- Saying no. Knowing which generated code should never be merged.
- Owning production. Monitoring, incidents, migrations — the unglamorous work AI can’t take responsibility for.
- Knowing the business. The best developers will be the ones who understand why the software exists.
The takeaway
AI won’t end the developer job. It will end the version of it that was mostly typing.
The world is generating software faster than it can understand it. In a few years, that pile of code will need people who can maintain, secure and rewrite it. Start practicing those skills now: next time AI writes something for you, don’t just run it — explain to yourself why it’s correct before you merge it.
Sources:
- GitHub Octoverse 2025 — commits, repositories and new developer numbers
- U.S. Bureau of Labor Statistics — Software Developers, Quality Assurance Analysts, and Testers — employment projections 2025–2035
- Jevons paradox (Wikipedia) — why cheaper resources lead to higher total use