|
Knowledge Management |
Junior employees need systematic access to the organization's expert knowledge and previous cases |
|---|---|
|
Role Design |
Junior jobs should be designed around what employees need to learn, not merely around whichever tasks AI cannot yet perform |
|
Learning In The Flow Of Work |
Learning should be integrated directly into real work tasks rather than taking place primarily on courses |
|
Managerial Coaching |
Senior employees should have an explicit role as coaches, rather than simply being the productive staff who can now complete junior work faster with AI |
The First Job May No Longer Exist
Conceptual illustration
7 MINUTES READ
AI is being sold as a technology that frees employees from routine work.
But routine work has also served another purpose: it is where young people learn their profession
By Mark Sinclair Fleeton
“What happens when companies no longer need junior staff?”
The Job Advertisement That Never Gets Posted
You have just graduated, and you have done everything right along the way. You chose the right degree, the right student jobs, the right internship, and you earned good grades. Now it is time to look for your first proper job. Traditionally, someone with your profile would have been at the front of the queue for the best jobs. But the world may no longer work that way. Companies do not necessarily have the same need for the junior positions you are looking for.
It is not that companies are in crisis – quite the opposite, in fact. The difference is that today an experienced employee can much more easily carry out tasks that might previously have required one senior employee and two junior colleagues. AI has made senior employees far more efficient or will, at any rate, in the very near future. But where does that leave you, a recent graduate without the senior employee’s experience? Who will hire junior staff when senior employees can simply do the work themselves?
AI Does Not Necessarily Take the Job – It Takes the Junior Tasks
Of course, companies differ in how they treat junior employees and which tasks they assign them. But the basic idea is that newcomers learn by taking on fundamental, more routine tasks. A junior lawyer might review documents and identify relevant rulings. A programmer is assigned smaller coding tasks; a consultant conducts desk research and prepares PowerPoint presentations; an auditor reviews supporting documents; and a marketing employee might draft initial copy and campaigns.
What many of these tasks have in common is that they are precisely the kind of work generative AI and AI agents are very good at automating and accelerating. In July 2026, the European Parliamentary Research Service published a briefing on AI and the labour market, The Debate on AI and Jobs. Its conclusion was that generative AI disproportionately affects entry-level jobs.
A study published in June 2026 by Switzerland’s most widely used job portal, jobs.ch, found that the number of advertised entry-level positions in 2025 was 32% below the 2019–2022 average. The study also found that junior positions declined by 16% in AI-exposed fields, while senior positions increased by 26%.
Junior Work Was Never Just Cheap Labour
The problem with this development is that junior positions were never merely a source of cheap labour for companies. They were also where you learned your profession, gaining the experience that would eventually make you a senior employee. In other words, if companies stop offering junior positions, they effectively break the pipeline that produces senior staff. We are automating the tasks that turned beginners into experts.
The next question is whether it is possible to skip the junior stage. Could a 25-year-old lawyer, for example, move straight into complex advisory work if AI has done the groundwork?
Lavor Bojinov and Edward McFowland III of Harvard Business School examined this question in Gen AI Boosts Productivity, But Can’t Turn Novices Into Experts. Their conclusion was that AI can make beginners significantly better at their work, but it does not make them experts – as the title rather gives away. AI narrows the immediate performance gap because a novice can produce a better result. But when a task requires genuine expertise, a substantial difference remains.
Part of the problem is that we learn a great deal from our mistakes. If AI simply supplies the correct answers, we never encounter the productive struggle of working through a problem. Stanford researchers described this issue in programming education in Productive Struggle: The Future of Human Learning in the Age of AI. When a student spends hours trying to understand why a program does not work, something happens beyond eventually solving the task. The student builds mental models of how the system works, why it fails, and how different solutions fit together.
Is It Already Happening?
There are clear signs of a weakening entry-level market in several AI-exposed knowledge professions. But finding firm evidence that AI is the cause is more difficult. The decline in software jobs is substantial, but it probably began before generative AI arrived. In law, what we are seeing so far is more a change in tasks than the outright disappearance of jobs. Consulting lies somewhere between the two. The graduate recruitment market is weakening, and employers explicitly say that AI is reducing demand, but economic conditions and cost-cutting play a larger role.
A UK analysis published in June 2026 by the Department for Science, Innovation & Technology together with LinkedIn offers an interesting snapshot. The overall UK hiring rate in April was 14% lower than a year earlier. Entry-level hiring fell by roughly the same amount as the wider labour market. The analysis shows that entry-level positions for accountants, graphic designers, data analysts, legal assistants, software engineers and product managers experienced the steepest declines. What these roles have in common is that much of their work involves finding, processing, organising, analysing or producing information – tasks at which generative AI has become markedly better. But the report is also very clear that there is no proof that AI caused the trend.
In other words, changes are taking place in precisely the professions whose tasks AI is best equipped to perform. But we still do not know for certain how much AI has to do with those changes.
The New Catch-22: You Need Experience to Gain Experience
This can put recent graduates in a difficult position, because employers may now expect more from them simply because AI is available. Where employers previously did not expect new graduates to know everything, but rather to learn on the job, they now expect them to deliver from day one using the tools at their disposal.
This may create an entirely new vulnerable group. Young people without experience are competing not only with other young people but also with experienced employees equipped with AI.
In July 2026, the US job platform ZipRecruiter published More Jobs, Higher Bar: The 2026 AI Employer Report, based on a survey of more than 1,000 American employers. It found that 38% had transferred basic data and processing tasks from entry-level employees to AI. Another 31% said explicitly that AI had led them to raise experience requirements for entry-level positions, while 57% had increased their overall productivity expectations because of AI.
This means that early-career candidates need a stronger CV to secure their first job, even as some of the tasks through which they would previously have gained that experience are being automated.
PwC calls this "seniorization" in its 2026 AI Jobs Barometer. It analyzed 2.4 million US entry-level jobs and found that the entry-level roles most exposed to AI were seven times more likely to require skills traditionally associated with senior employees, such as judgement, leadership, strategic decision-making and collaboration.
What Happens to the Apprenticeship Principle?
Skilled trades have apprenticeships; doctors undergo clinical training; lawyers have trainee solicitors; and researchers have PhD and postdoctoral programmes. Many knowledge professions, by contrast, have depended on a more informal model in which people are hired as juniors and learn through their work. If that model breaks down, we will need new institutional approaches to learning.
These could include graduate programmes, longer internships, formal apprenticeship models for graduate professions, rotation programmes, or requirements for companies to retain junior tasks as learning opportunities even when AI can perform them faster.
The consultancy McKinsey proposes four elements in Building Expertise in the Age of AI: Who Trains the Next Generation?
Another company in the same industry, Gartner (yes, the one behind the Hype Cycle), proposes changing the content of junior roles and giving new employees access to more complex tasks earlier. Gartner points out that traditional junior jobs allowed employees to build skills, relationships and an understanding of their organizations through relatively low-risk work. If those tasks disappear, companies must create other situations in which beginners can try, fail and learn without catastrophic consequences.
The Employee Who Never Became Senior
Let us return to the recent graduate at the beginning of the article, facing a difficult situation. The problem is not necessarily that the profession they trained for is disappearing. There will still be strong demand for experienced lawyers, programmers, analysts and journalists in the future. The problem is how to become experienced in a labour market that no longer needs the inexperienced.
We are currently spending a great deal of time discussing how many jobs technology creates or destroys. But perhaps it is just as important to examine where on the career ladder those jobs are disappearing.
If it is mainly the bottom rungs that are being automated, we face not only an employment problem but also a generational one.
About the author
Is a Danish journalist, author and editor-in-chief of AI Portalen and co-founder of FutureAble. He works with artificial intelligence, technology and society with a particular focus on, among other things, the labor market, democracy, digital sovereignty, regulation and AI skills. He also teaches and gives lectures on AI and society.
