Businesses are excited about AI performing entry-level work faster and at lower cost. But if companies no longer need juniors, a much more uncomfortable question emerges: where will tomorrow’s experienced professionals come from?
— We need someone with at least five years of experience.
— Where were they supposed to get those five years?
Silence.
Because five years earlier, companies decided that much of the work traditionally assigned to junior employees could be handled more cheaply by AI.
Research.
First-pass analysis.
Documentation.
Basic customer requests.
Reports.
Routine coding.
Data validation.
Drafts.
Junior roles began to disappear.
The economics initially looked excellent: lower costs, greater speed, higher productivity.
Then companies discovered a problem that another AI subscription could not solve:
You cannot hire experience if nobody is being given the opportunity to acquire it.
And this may become one of the most underestimated consequences of AI for the labor market.
We May Be Looking at the Wrong Part of the Problem
Most conversations about AI and employment revolve around one question:
“Which jobs will AI replace?”
But there may be a more important question:
“Which career ladders will AI break?”
Because a profession is not a single job.
It is a path.
Junior becomes middle.
Middle becomes senior.
Senior becomes architect, director, partner, chief engineer, CFO or CTO.
Between those stages is a mechanism we rarely think about.
Beginners are given simpler work.
By doing it, they gradually learn how to handle difficult work.
We call those tasks routine.
But for someone at the beginning of a career, they were never merely routine.
They were training.
A Junior Employee Does More Than Cheap Work
Consider a young financial analyst.
They are asked to collect data.
Check a spreadsheet.
Prepare the first draft of a report.
Compare performance.
Find discrepancies.
A senior reviews the result and says:
“No. Your calculation is correct, but your conclusion is wrong.”
Something important has just happened.
The junior has learned something that was never contained in the spreadsheet.
They begin to understand context.
A year later, they recognize certain problems independently.
Three years later, they understand why numbers can sometimes tell a misleading story.
Five years later, they are reviewing someone else's work.
Eventually, they become the person the organization trusts to make the decision.
Now imagine that AI performs the first three stages.
The data are collected automatically.
The spreadsheet is checked.
The draft is written.
The discrepancies are identified.
Excellent.
But now there is another question:
Where does the human learn?
We Can Automate a Task — and Accidentally Automate the Training
This is the real paradox.
A company looks at junior-level work and sees:
low complexity.
AI looks at the same work and sees:
a task it can perform.
But the labor market has quietly been using that work for another purpose:
creating future experts.
That function never appeared on a P&L statement.
Nobody wrote:
Preparing this report requires two hours of labor and generates 0.3 units of future senior expertise.
So when AI takes over a simple task, the economic model shows an obvious benefit.
Two hours saved.
What it may fail to show is what disappeared with those two hours:
two hours of professional practice.
Once, that means very little.
Repeated thousands of times over several years, it becomes a very different problem.
The Most Dangerous Automation May Be the One With an Excellent ROI Today
Imagine a department of ten people.
Previously:
two seniors;
three mid-level professionals;
five juniors.
AI arrives.
The company realizes that a substantial portion of the work performed by those five juniors can be automated.
A year later, the structure looks more efficient:
two seniors;
three mid-level professionals;
one junior;
AI.
Costs are lower.
Output is higher.
Management is satisfied.
Several years pass.
One senior leaves.
The other receives an offer from a competitor.
The company goes to the labor market.
Suddenly, experienced senior professionals are scarce.
And expensive.
Why?
Because other companies made roughly the same decision five years earlier.
Each company made a rational decision individually.
Collectively, they created a shortage.
This Would Not Be the First Time Efficiency Today Created Scarcity Tomorrow
Businesses are very good at optimizing what they can measure immediately.
Headcount.
Cost per employee.
Time per task.
Revenue per employee.
Tickets resolved.
Lines of code.
Hours billed.
What is much harder to measure is:
How much expertise does this organization create internally every year?
It is an unusual metric.
You will rarely find it on a CEO dashboard.
But in an economy where technology changes rapidly, it may become critical.
Because experience cannot simply be ordered for delivery.
IT Makes This Particularly Easy to See
Imagine a junior systems engineer.
They begin with relatively simple tasks.
Checking configurations.
Reading documentation.
Configuring equipment.
Making mistakes in a lab.
Asking a senior engineer questions.
Participating in deployments.
Sitting quietly on a troubleshooting call where they understand perhaps half of what is being discussed.
Several months later, they begin to understand why the senior asks certain questions.
Two years later, they solve some of those problems independently.
Five years later, they can walk into a server room and sometimes recognize the likely cause of a problem before opening their laptop.
Where did that knowledge come from?
Not only from certifications.
Not only from documentation.
Certainly not only from university.
It came from hundreds of small situations.
Mistakes.
Observations.
Questions.
Real projects.
And sentences like:
“Look — in situations like this, the problem is usually not where it first appears.”
That is experience.
AI May Know the Answer. The Junior Still Needs to Understand Why It Is the Answer.
There is an important distinction here.
AI can be an extraordinary educational tool.
A junior professional using AI may learn faster than one without it.
They can ask questions.
Request explanations.
Analyze code.
Compare architectures.
Simulate scenarios.
Study documentation.
Receive feedback almost instantly.
The problem does not begin when a junior uses AI.
It begins when the company decides:
“If AI can perform junior tasks, we no longer need juniors.”
Those are two completely different strategies.
The first can accelerate human development.
The second can eliminate the entry point into a profession.
And This Goes Far Beyond Programming
Consider law.
Junior lawyers read documents, identify clauses, compare contracts and prepare drafts.
AI can perform a significant portion of that work.
But it is through reviewing hundreds of contracts that lawyers begin noticing things they did not notice before.
Medicine.
Young doctors study cases, observe senior physicians and gradually develop clinical judgment.
Finance.
Young analysts work through hundreds of models and reports before they begin seeing the economics behind the numbers.
Cybersecurity.
Entry-level specialists investigate alerts, false positives, logs and incidents.
Sales.
Junior salespeople participate in dozens of conversations, hear objections, lose deals and observe experienced negotiators.
Engineering.
Young engineers begin with one small part of a project and, over time, learn to understand the entire system.
Across these professions, the same mechanism exists:
Simple work helps finance the development of complex expertise.
If we remove the first part, we need another way to produce the second.
And We Haven’t Fully Designed That Replacement Yet
One answer might be:
“We will simply train people.”
But training and experience are not the same thing.
You can complete a negotiation course.
That is very different from sitting across from a client, discussing a million-dollar contract, and experiencing ten seconds of silence after stating your price.
You can study incident response.
That is very different from making decisions while a company's real infrastructure is down.
You can read everything about project management.
That is very different from explaining to a customer why a project expected tomorrow will now be delayed by a month.
Some knowledge can be transferred.
Other knowledge has to be professionally lived.
Companies May Need to Redefine What a Junior Role Is For
Historically, companies often hired junior employees because someone had to perform simpler work.
That is an economic model built for the previous era.
In the next model, a company may hire juniors for a different reason:
because it will need seniors five years from now.
That turns an entry-level position from inexpensive labor into an investment in future expertise.
And the structure of the role changes with it.
Instead of giving junior employees whatever AI has not yet automated, companies may need to deliberately create experiences through which people develop judgment.
The Junior of the Future May Do Less — but See More
This could become a much more interesting model.
Previously, a junior analyst might spend four hours collecting data and ten minutes discussing the conclusions with a senior.
Now AI can collect the data in minutes.
What should happen to the four hours that were saved?
The least imaginative answer is:
remove the junior.
A better answer is:
give them four hours to understand why the senior reaches a particular conclusion.
Let them join the meeting.
Let them analyze a real case.
Let them compare alternatives.
Let them defend their own recommendation.
Let them make mistakes where those mistakes are still inexpensive.
Let them receive feedback.
In that model, AI does not destroy apprenticeship.
It can make apprenticeship significantly better.
We May Need a New Model of Professional Education
For centuries, many professions used a recognizable structure:
Apprentice.
Journeyman.
Master.
People did not simply study theory.
They worked alongside someone who knew more.
They watched.
Repeated.
Failed.
Received corrections.
Gradually, they were trusted with more difficult work.
Modern corporations preserved much of the same mechanism — they simply stopped calling it apprenticeship.
Junior → Middle → Senior.
AI forces us to ask a question we have largely avoided:
If the bottom of the pyramid is automated, how does a person reach the next level?
There Is Another Twist: Seniors Need Juniors Too
That may sound strange.
But teaching less-experienced professionals also develops senior employees.
When a senior has to explain:
“This is why we do it this way,”
they are forced to formalize their own knowledge.
When a junior asks:
“Why can't we do it differently?”
sometimes the answer turns out to be:
“Because that's how we've always done it.”
That is useful information.
Juniors bring questions.
Seniors bring context.
AI brings speed.
Perhaps the strongest future model is not:
Senior + AI.
But:
Junior + Senior + AI.
Three different forms of value.
For Armenia, This Question Matters Even More
For a small country, human capital is not an abstract HR issue.
It is an economic resource.
If companies stop hiring entry-level professionals because AI can perform entry-level tasks more cheaply, the consequences will not appear tomorrow.
For the first few years, everything may look excellent.
Productivity rises.
Teams become leaner.
Costs fall.
Then the market begins asking:
Where are the experienced engineers?
Where are the strong account managers?
Where are the architects?
Where are the cybersecurity experts?
Where are the project managers?
Where are the professionals who have already lived through real mistakes, difficult customers and complex projects?
And the answer may be uncomfortable:
We stopped developing them five years ago.
CEOs and HR Leaders Therefore Need a New Question
Not only:
“Which positions can we automate?”
But:
“Which capabilities will stop being created if we automate these positions?”
That is a much more mature question.
Before automating a process, organizations can ask:
Which tasks are genuinely routine?
Which of those tasks also function as training?
What does a junior actually learn by performing them?
Can AI take the mechanical part while leaving the human learning loop intact?
How will employees gain exposure to real cases?
Who will provide feedback?
How will we know in three years that they are ready to make more complex decisions?
And ultimately:
Where will our next senior come from?
Companies May Need to Start Measuring Another Kind of Debt
Technology organizations already understand technical debt.
A company makes convenient decisions today and pays for them later.
The AI era may create another form:
experience debt.
A company saves money today because it automates the work of junior employees.
At the same time, it gradually stops producing its own expertise.
For the first few years, that debt is almost invisible.
Then repayment begins.
A senior leaves.
There is no internal replacement.
The market is short of talent.
Salaries rise.
Recruitment takes months.
And the organization is forced to buy expertise at a premium that it could have spent years developing internally.
This Is Not an Argument Against AI
Quite the opposite.
It would make little sense to force people to manually perform work that machines can do faster, more accurately and more cheaply simply in the name of “experience.”
But businesses must then consciously replace the old learning mechanism with a new one.
If AI takes the routine work —
give people more context.
If AI creates the first draft —
teach juniors to challenge it.
If AI analyzes the data —
teach people to ask better questions about the analysis.
If AI writes the code —
young engineers must learn architecture, security and the consequences of that code.
If AI prepares the proposal —
young sales professionals need to understand the customer, economics and negotiation.
We should not preserve meaningless work.
We should preserve the mechanism that turns a beginner into an expert.
2031
Another interview.
— We need a senior specialist. At least five years of serious practical experience.
There are almost no suitable candidates.
The recruiter says:
— The market has a shortage.
The CEO looks surprised.
— How did that happen?
The answer may be remarkably simple:
Five years ago, everyone decided at the same time that juniors were no longer necessary.
Final Thought
AI really may eliminate a significant number of entry-level tasks.
In the short term, that can look like excellent optimization.
But businesses should remember one simple thing:
A senior professional is not created on the day you publish a senior vacancy.
They begin years earlier.
When someone first trusts a beginner with a small task.
When they make a mistake.
When a senior explains why.
When they observe.
When they try again.
When they receive slightly more responsibility.
And eventually, they become the person everyone else comes to for an answer.
AI can make that journey faster.
It can make it smarter.
It can remove enormous amounts of meaningless mechanical work from it.
But if we remove the journey itself, technology cannot give those lost years back.
So before a company says:
“AI can handle this junior role now,”
leadership should ask one more question:
“Then who will become our senior five years from now?”