A junior analyst prepares a first market assessment. A new recruiter reviews dozens of profiles. An HR officer drafts a policy, checks it against the company’s practices and asks why an exception exists. These tasks take time. They are also how people learn what matters.
Now imagine those first drafts, reviews and summaries are completed by AI in seconds. The productivity gain is obvious. The less obvious question is what happens to the employee who would have done that work.
As an HR consultant, I see this as a workforce design issue, not merely a technology decision. Companies may be able to automate the first rung of the career ladder before they have worked out how anyone will climb to the second.
The warning sign is already visible
In an August 2026 update, researchers at Stanford’s Digital Economy Lab reported that employment among 22-to-25-year-olds in occupations highly exposed to AI was 19% below where it would have been had it kept pace with less-exposed peers. The gap was driven primarily by reduced hiring rather than a wave of layoffs. The researchers also found no evidence of widespread, economy-wide job displacement. These are findings about the US labor market, not a forecast for every industry or country. Read the Stanford research.
At the same time, Gartner’s Q3 2026 CHRO Report frames the design of an AI-augmented workforce and the redefinition of jobs as central leadership questions. The connection to early-career development is my inference: if the tasks inside jobs change, so must the way people gain experience.
Why the “easy work” was never just easy work
Some routine tasks deserve to be automated. Yet a junior employee’s first assignment often does more than produce a deliverable. It lets them see imperfect information, notice patterns, make mistakes, receive feedback and build a mental model of the business.
A recruiter who screens applications learns what a role actually requires when they compare profiles with hiring outcomes. An HR officer who works through an employee case learns why a policy exists and where a standard answer falls short. A junior analyst who checks the underlying data develops a feel for what looks implausible.
AI can produce an answer to each of these tasks. Producing the answer does not automatically transfer that understanding to the person asked to approve it.
“We can automate a task in a day. Rebuilding the learning it provided takes deliberate design.”
The hidden cost appears later
The immediate business case for automation is usually measured in hours saved. The cost of a thinner development pipeline may appear years later, when a company needs people who can challenge a recommendation, handle an exception or lead through uncertainty.
Consider a team that stops hiring junior HR professionals because an AI assistant can draft contracts, answer routine questions and summarize employee feedback. Three years later, who has spent enough time with real employee situations to advise a founder through a sensitive restructuring? A tool can help prepare the work. Experience is still needed to understand the consequences.
This does not mean every junior task must survive. It means that before removing one, leaders should identify the capability it helped build and decide how that capability will now be taught.
Design an apprenticeship for the AI era
A better junior role may include less repetitive production and more guided practice. For example:
- Start with a human hypothesis. Ask the employee to state what they expect to find before requesting an AI analysis.
- Make verification visible. Have them check sources, assumptions, calculations and missing context, then explain what they changed.
- Give them real exceptions. Routine cases are useful, but unusual cases build judgment. Review them with a senior colleague rather than leaving the junior employee alone with an AI answer.
- Protect coaching time. If AI saves managers an hour, reserve some of that hour to review decisions and explain tradeoffs.
- Assess understanding. Ask an employee to defend a recommendation without the tool open. A polished deliverable is not proof that learning occurred.
These are design choices, not reasons to slow adoption. They make AI use part of a development system rather than a substitute for one.
A new question for workforce planning
HR and business leaders already ask which tasks AI can perform, how much capacity it creates and which roles should change. They should add a longer-term question: Where will our future experts get their first 10,000 real decisions?
The answer will differ by profession. It may involve supervised client work, simulations, rotations, case reviews or junior employees directing AI tools while being accountable for the reasoning. What matters is that someone owns the learning design.
The companies that benefit most from AI will not simply be those that remove the largest number of tasks. They will be the ones that increase productivity while still developing people who can exercise judgment when the answer is uncertain, the stakes are high and the tool sounds more confident than it should.
“Before eliminating an entry-level responsibility, ask: What was this work teaching—and where will people learn that now?”