A ladder without its first rung: How automation is depriving the labor market of its future professionals

Part One.

Artificial intelligence is taking over more than the routine work of entry-level employees. Along with it, the space in which they became professionals is disappearing.

Image created by the author using AI

To get your first job, you need experience. To get experience, you first need a job. This is a familiar catch-22.

In the past, this formula was treated as a sign of a poorly written job posting. A company was looking for an entry-level employee but wanted someone who could already work independently, make the right decisions, and avoid taking up the time of senior colleagues.

Now this contradiction is gradually becoming a structural feature of the labor market.

Employers are still advertising entry-level positions, but the tasks once entrusted to people with little experience are disappearing from the jobs themselves. Algorithms sort, draft, check, and complete them.

The work that remains requires analysis, an understanding of context, and responsibility for the outcome - in other words, qualities that usually develop not before a first job, but because of it.

The Work No One Considered Training

An entry-level employee was “rarely” hired to make strategic decisions.

In technical support, they answered common questions, studied the documentation, and escalated difficult cases to the next tier. In software development, they fixed minor bugs and wrote isolated pieces of code. In analytics, they cleaned data and prepared simple reports. In marketing, they collected statistics, checked advertisements, and drafted initial versions of copy.

This work could be repetitive, poorly paid, and not especially prestigious. Admittedly, those conditions alone were enough to discourage plenty of applicants. But the work also served a second purpose - one that never appeared in company reports.

By performing simple tasks, a person learned how a real system worked. They discovered which instructions failed in practice, how users described the same problem in different words, why apparently clean data turned out to contain errors, and when a small change in the code affected the entire product.

At first, their decisions were checked. Then they were trusted with more complicated cases. Mistakes remained relatively inexpensive, while responsibility increased gradually.

That was how experience was built.

It consisted of more than knowledge or the number of years listed on a résumé. Experience emerged from countless small encounters with reality in which a person learned to notice what the instructions did not mention.

The Second Outcome of Simple Work

When a junior analyst prepared a report, the company received two outcomes at once.

The first was the report itself. The second was an employee who understood the data, the product, and the company a little better after preparing it.

The same was true of a resolved support request, a corrected bug, a reviewed contract, or a finished advertisement. Every completed task produced something useful while also helping to shape a future professional.

Automation preserves the first outcome for the company, but it may deprive it of the second.

Within seconds, AI can prepare a summary, sort support requests, draft an email, find a common coding error, or transform a spreadsheet. From a business perspective, the savings seem obvious: why assign slow, repetitive work to a person when software can do it faster?

But an algorithm that completes a thousand entry-level tasks does not become a future department head, engineer, or support specialist. It simply completes the next thousand.

Every task completed by a beginner left behind a slightly more experienced person. Automation leaves behind only the completed task.

Simple Tasks Are Disappearing from Entry-Level Jobs

The first changes are already visible.

In 2026, the Strada Institute surveyed nearly 1,500 executives and hiring professionals. Forty-one percent said that AI had reduced the number of basic tasks through which entry-level employees acquired skills. Almost as many - 42 percent - had begun assigning newcomers more analytical work requiring independent judgment. When hiring recent graduates, employers valued candidates who already had experience performing similar work above all others. The Strada Institute published the survey results.

This creates a troubling sequence.

First, a company automates the tasks through which a person could have started learning. Then the remaining work comes to involve more analysis and responsibility. Finally, the employer concludes that a recent graduate is not ready for the role and chooses someone who has already performed it elsewhere.

The entry-level position formally survives, but its lower half disappears.

PwC calls this process the “seniorization” of entry-level work. According to the company, junior positions in fields most affected by AI are seven times more likely to require leadership, strategic thinking, and other skills traditionally associated with more experienced employees. The number of these “seniorized” entry-level positions has grown by 35 percent since 2019. These findings appeared in the 2026 AI Jobs Barometer.

The employer is still looking for a beginner - only now it needs a beginner who can reason like a professional.

The First Consequences Are Already Visible

Researchers at the Stanford Digital Economy Lab studied data from the largest payroll-processing company in the United States. Following the spread of generative AI, employment among workers aged 22 to 25 in occupations most exposed to automation fell by 16 percent relative to other groups — even after the researchers accounted for changes within individual companies.

The decline was particularly pronounced where AI was used to automate work rather than assist people. Employment among more experienced workers in the same occupations remained stable or continued to grow. The study was published by the Stanford Digital Economy Lab.

This does not prove that artificial intelligence single-handedly destroyed the entry-level labor market - which, to be fair, was hardly thriving to begin with.

Hiring was also affected by high interest rates, layoffs following a period of rapid growth in the technology sector, the relocation of jobs to other countries, and broader economic uncertainty. The Federal Reserve Bank of New York found no distinct, sudden decline in job postings that could confidently be attributed to the arrival of ChatGPT. Demand in many occupations had begun falling earlier, while the number of junior and senior vacancies often declined at the same time. The analysis was published by the New York Fed.

It would therefore be premature to claim that all aspiring professionals are being left without work.

But the transformation of the jobs themselves is already visible: there are fewer simple tasks, while expectations of beginners are rising.

Knowledge Is Not the Same as Experience

It may seem that education can solve this problem.

Give people more knowledge in advance, teach them how to use AI, and show them typical workplace situations, and they will arrive at a company better prepared.

This can certainly narrow part of the gap. Simulations, training environments, and artificial intelligence can explain mistakes, create scenarios, and allow people to practise individual actions safely.

But professional experience is not built from correct answers alone.

It develops when information is incomplete, instructions contradict one another, a user behaves unexpectedly, and a decision has consequences for a real product and real people. In these circumstances, a professional learns not merely how to perform an action, but how to recognise when the standard action should not be performed at all.

This part of professional development is difficult to move entirely outside the workplace.

Society may therefore end up with large numbers of people who know how to use the tools, while simultaneously facing a shortage of people whom companies are willing to trust with real responsibility.

So Where, Exactly, Does the First Rung Disappear?

Automating entry-level work does not necessarily have to eliminate entry-level employees.

AI can make a newcomer more useful: it can help them understand documentation faster, explain unfamiliar code, suggest possible solutions, and provide immediate feedback. Some employers already expect this to allow them to hire more graduates rather than fewer.

But that outcome does not occur automatically.

If a company uses AI solely to reduce costs, it removes simple tasks and cuts the number of entry-level positions. If it uses AI for training, it must still allocate mentors’ time, give people access to real situations, and transfer responsibility gradually.

The first path delivers savings immediately.

The second produces a professional several years later.

An individual employer has an understandable incentive to choose the first option and later hire a fully trained professional from someone else.

The problem is that one day, everyone may make the same choice.

The next question is how far this process can go. That is the subject of Part Two: *The Intern Who Never Was*.

Artificial intelligence is not merely raising the bar for a first job. It is removing part of the road people once travelled to reach that bar.