Nobody's Hiring Juniors Anymore

If you manage a team, you’ve probably had some version of this conversation. Next year’s budget has room for one hire, but the work you’d planned to give a junior now looks like something a software licence could cover.

What worries me is how long it could take to see the cost. In 5 years, that person might have been ready to lead a team. If you need someone with that experience then, today’s saving could look very different.

In an earlier post about AI and jobs, I argued that AI often removes the next hire before it replaces an existing employee. This post is about how people get the experience a business will need if it stops hiring beginners.

It’s how every senior you’ll want to hire learned the job.

The Senior You Need in 5 Years Has to Start Somewhere

Hiring a beginner costs more than their salary. Someone has to explain the work, check it and make time for mistakes. In return, the business has a chance to develop someone it can trust with more responsibility.

That return is uncertain.

People leave, roles change, and some hires don’t work out. But when a tool makes the junior’s tasks cheaper to complete, it’s easy to remove the role without deciding how the team will develop its next experienced person. That’s the part of the hiring decision I’d want to see in the budget too.

It’s like waiting for a train that’s never going to come.

Young Workers Are Losing Ground in AI-Exposed Jobs

Stanford’s payroll study found a widening gap among workers aged 22–25:

• Experienced workers showed no comparable gap. The shortfall was concentrated among people starting their careers.
• Fewer hires appear to be the main reason. Stanford found the adjustment came mainly through reduced hiring of young workers, rather than more departures.

That doesn’t prove AI caused the decline. Stanford says so, and New York Fed research on job postings found little evidence of a distinct AI effect. But fewer people getting a start is reason enough to ask how they’ll get the experience employers want.

What Gets Lost When the Work Gets Automated

Imagine a junior analyst preparing a monthly report. They pull the numbers, notice that two systems disagree, and ask which one to trust. Their manager explains that one counts a sale when the contract is signed and the other when the invoice is paid. The analyst changes the report, but they also learn something about how the business works.

Next month, they know to check. Later, they recognise the same problem in a different set of numbers. Eventually, they’re the person who can explain why a result looks wrong before anyone has worked out where the error is.

That’s an illustrative example, but it’s the kind of learning I mean. The report is the visible output. The questions, corrections and repeated decisions are how someone becomes capable of taking responsibility for it.

AI can help with that process. It can explain a calculation, suggest checks or give a beginner another way into a problem. But if it produces the report and sends it straight to a senior for approval, the beginner never encounters the disagreement between the systems. The work gets done without anyone new learning why it was difficult.

It’s like a staircase with the bottom steps sawn off.

That doesn’t mean preserving every tedious task. Copying numbers between spreadsheets for three years is no guarantee of expertise either. The useful part is having to understand a problem, attempt it, receive feedback and try again with more responsibility. If automation removes the old way of doing that, someone has to design a new one.

You Can Hire Experience Until Everyone Else Needs It Too

A company can reasonably decide to recruit experienced people instead of training beginners. It may need skills now, lack the capacity to supervise someone, or have too little suitable work to support a junior role.

Many employers are already making that call. Job postings on Indeed show the shift:

But recruiting experience depends on other employers helping people acquire it. If enough businesses reduce their intake, the pool they all expect to hire from could become smaller. A saving on training today can turn into a longer search, a higher salary or an unfilled role later.

It’s like every company waiting for another one to train the people it wants to hire.

Think of the experience someone starting today could build over 5 years. Whether a shortage develops will depend on how many companies keep hiring, how jobs change and what other routes into the work emerge.

For a manager, I’d start closer to home. Where did your strongest people learn? How many joined before they could do the job independently? If you’ve relied on that progression before, what makes you confident you can stop funding it now?

Give the Junior Work They Can Learn From

Simply making a beginner check AI output doesn’t solve this. If they don’t understand the work yet, they may approve an answer because it sounds plausible. Telling them they own the result doesn’t give them the judgement to assess it. Owning what you forward is the right rule, but it assumes the person forwarding can tell what’s wrong.

I’d design the role around a gradual increase in responsibility, with time from a senior built into the plan.

• Give them a bounded problem to work through themselves. For the analyst, that might mean calculating one part of the report from the source data before comparing it with the model’s answer. The point is to practise the reasoning they’ll need when the answers disagree.
• Make the checks specific. Ask them to trace a figure to its source, test an assumption or explain what would change the conclusion. “Review this” is too vague for someone who is still learning what can go wrong.
• Have a senior review their reasoning. Discuss what they checked, what they missed and why it mattered. Budget for those conversations. They are part of the job.
• Expand responsibility as they demonstrate it. Start with work where mistakes can be caught before they reach a customer. Give them more complex decisions as they show they understand both the task and its consequences.

AI belongs in that role too. The junior can use it to explore an unfamiliar subject, compare approaches and handle routine steps. They should also get chances to explain a decision without leaning on the model’s explanation, and to follow the work through far enough to see whether it helped.

If you’re early in your career, those are the opportunities I’d look for: access to the underlying work, someone who will challenge your reasoning, and responsibility that grows as you learn. Being the person who passes a polished draft along is a much thinner education.

For me, the question to a client considering a cancelled hire is still: where do you expect your next senior to come from? “We’ll recruit them” can be a reasonable answer. I’d want to know what it depends on, what it might cost, and what happens if the search takes longer than expected.

The tools will keep changing what beginners do. We still need to give them work through which they can become experienced.

If you’re working out how to use AI in your team, including how roles and responsibilities should change, that’s the work I do.