I keep hearing the same question in meetings, in podcasts, and in chats with friends. If AI can do the work of a junior developer, why should we hire junior developers at all?
I understand why people ask it. The tools today can write a CRUD endpoint, fix a simple bug, or write tests faster than a person in their first year on the job. From a pure cost view, in a single quarter, the math seems easy. Pay for a tool, skip the salary.
But I think this math is wrong. Not a little wrong. Dangerously wrong. Let me explain why.
A team with no pipeline has an expiry date
Every senior developer I know was a junior once. That sounds obvious, but companies act like seniors appear from nowhere, fully formed, ready to design systems and mentor people. They do not. They grow. And they grow inside teams, working on real problems, getting feedback, breaking things in safe places.
If the whole industry stops hiring juniors today, in ten years we will have a serious problem. Seniors retire. Seniors burn out. Seniors move to management or open a bakery. Who replaces them? Nobody, because we cut the pipeline to save a few salaries in 2025.
A team with no pipeline is a team with an expiry date. You may not see the date today, but it is printed there, and it is closer than you think.
Juniors plus AI learn faster than I ever did
Here is the part that makes me optimistic. When I started, more than two decades ago in Brazil, learning was slow. I waited for books, for a senior with free time, for a forum answer that maybe came in three days. Today a junior has an assistant available all day. They can ask why a piece of code works, what a stack trace means, what trade offs exist between two designs. They get an answer in seconds.
This means a motivated junior with AI can learn in one year what took me three. The tools that some people use as an argument against hiring juniors are actually the best argument for hiring them. The cost of growing a person dropped. The speed of growing a person went up. That is a good deal.
Barbara Liskov spent her career showing that the fundamentals of our field can be taught. Abstraction, modularity, how to think about what a piece of code promises to do. These are not magic talents. They are skills, and skills are teachable. If the fundamentals are teachable, then juniors are an investment with a known path, not a gamble.
What the tools cannot teach
There is one thing AI does not give a junior, and that is judgment. Knowing when the stakeholder asked for X but needs Y. Knowing when a quick fix is fine and when it will cost us a weekend in six months. Knowing how to say no in a kind way.
Judgment comes from being in the room. From shipping something, watching it fail, and fixing it with the team. From seeing a senior handle a hard conversation with a client. You cannot download that. You have to live it. And a junior can only live it if someone hires them.
The business case, not the charity case
I want to be clear. I am not saying hire juniors as charity. I am saying hire them because it is good business.
Short term productivity is one metric. Long term team health is another, and it is the one that decides if your company is still shipping in five years. A team of only seniors plus AI looks very productive on a dashboard. It is also fragile. Salaries are high, replacement is hard, and knowledge sits in few heads. When one of those heads leaves, you lose more than a person. You lose continuity.
Juniors bring fresh energy, they ask the dumb questions that expose the dumb parts of our systems, and they cost less while they grow. With AI helping them, the gap between what they cost and what they deliver closes faster than ever before.
Martin Fowler has written for years about how software is a team activity, not a typing activity. I believe that more today than ever. AI changed the typing part. It did not change the team part.
So what should change
The role changes, yes. A junior in 2025 should not spend the day writing boilerplate. The AI does that. A junior should spend the day reviewing what the AI wrote, asking why, learning the domain, talking to users, and slowly building the judgment I talked about. That is a better junior job than the one I had. Honestly, I am a little jealous.
So my answer to the question is simple. Yes, hire juniors. Hire them on purpose, with a plan, with mentors, with real work. Stop thinking of them as cheap labor for boring tasks, because AI took the boring tasks. Start thinking of them as the only insurance policy your team has against time.
The companies that understand this will have strong teams in 2030. The ones that do not will be posting senior job ads that nobody can fill, and wondering where everyone went.
Pax et bonum.