Employers may be removing the very work through which college and university graduates once learned, proved themselves and became experienced professionals
A partner at an Ottawa firm recently did something she had waited 20 years to do: She eliminated the worst job she ever had. The document review that consumed her first two years, repetitive, unglamorous, occasionally soul-destroying, now runs through an AI system before most of the office has finished its first coffee. The productivity gain is real, and nobody mourns the drudgery. Then the firm’s newest hire appeared at her door asking what he should work on, and she suddenly recognized that the worst job she ever had was also where she’d learned much of what she now knows.
Entry-level work was never only work. It was also an informal apprenticeship. Every profession has had its version of the place where beginners take on limited responsibilities, make recoverable mistakes and learn while someone experienced watches: the new analyst’s first research assignment, the junior accountant’s routine reconciliations, the beginning lawyer’s document and case review, the new programmer’s small debugging tickets: None of it was glamorous, and all of it was teaching more than the task itself: teaching how the organization worked, what “good work” looked like, how seasoned people exercised judgement, and what happened when something went wrong. AI is now removing many of these routine tasks, which leaves a question most organizations have not yet answered: After AI takes over the beginner’s work, how does the beginner stop being a beginner?
The informal apprenticeship
What made the old arrangement work was that it built trust in both directions over time. The beginner learned through repetition, correction and watching, absorbing standards that weren’t necessarily written down. The employer learned, too: Watching a new person handle small, low-risk work was how an organization decided who was ready for more. Nobody needed certainty on day one; there was time to build trust. Remove the task, however, and you may remove the learning within it as well as the watching, which was how readiness used to be recognized.
Some employers will redesign junior roles thoughtfully, using AI to accelerate learning. Others will simply stop hiring at the bottom and demand that every arrival perform like a five-year veteran. This creates the contradiction that’s increasingly evident in job postings: Employers want experienced applicants even as they eliminate the work through which experience was once acquired.
This is not simply a graduate skills gap
It is tempting to file all this under the familiar heading of a “skills gap.” A skills gap says the graduate lacks something, and the colleges or universities must supply it. But a person can carry genuine capability and still be denied the first opportunity to show it. An employer can lament inexperience while offering no position in which experience could possibly be built. No additional course can solve that problem by itself, because the missing ingredient is not always knowledge. It is an opportunity to begin. Readiness is produced on both sides of the transition, and each side now has redesign work it cannot delegate to the other. The question employers must answer is: What are we prepared to teach after the person arrives?
What faculty can change
The reality is that if fewer workplaces provide entry-level learning opportunities, colleges and universities will need to create more of them before graduation. The good news is that faculty control exactly the practices this requires. The shift is from judging the answer to examining the student’s decisions. A student who used AI should be able to give an account of the collaboration — the prompts, the doubts, the overrides — and defend the result out loud, because explaining several consequential decisions can reveal more than another polished submission.
Assignments can carry the uncertainty students will encounter at work: requirements that shift mid-project, evidence that conflicts, a client who rejects the first proposal, a crucial fact that is missing. These experiences can reproduce part of the learning once provided by the first job, while someone is still present to observe and guide the student./h
What employers can change
The employer’s task begins with two practical questions: Before automating a junior task, ask what it was teaching and where that learning will happen now. The answers point to things that can be rebuilt deliberately – real work done with AI but observed by a human, room for survivable mistakes, and coaching treated as a scheduled responsibility rather than something left to chance. None of this is charity. An organization that automates entry-level work without replacing its developmental function may gain productivity now. But it’ll find, several years later, that too few employees are ready for senior responsibilities. The employers that thrive in the AI decade may simply be the ones who noticed, in time, that experience is something they must continue to develop, not merely demand at the point of hiring.
Restoring the first rung
AI will keep replacing the beginner’s work. Much of it deserves to go, and no one should wish the drudgery back for its own sake. But some of that work had a second purpose, and the second purpose cannot easily be automated. The task ahead — for faculty, employers and the institutions that connect them — is to deliberately recreate what entry-level work once provided: a place where beginners can practise, be observed, make recoverable mistakes and earn trust, until they’re ready for greater responsibility.
At the beginning of this commentary, we introduced the partner in Ottawa who combined the old approach with the new. She gave the new hire a live file the AI had already processed, asked him to find what it had missed, and pulled up a chair to watch him work. It’s a modest example, but the principle is important: Automation should not eliminate the opportunity to learn under observation. The larger question is whether we can do that across professions and organizations so the next generation has a genuine place to begin.
Sources
This commentary draws on: Finch and Murgatroyd’s work on AI and entry-level employment for the Learning City Productivity Project (2026); reporting on declining entry-level hiring in AI-exposed occupations; and the discussion of the weakening first rung in Proof on Demand, published on teachonline.ca.


