By Theresa Rézeau 

In 1889, Canadian-born British painter Elizabeth Adela Forbes (1859–1912) painted the end of a school day.

In School Is Out, children gather themselves before leaving the classroom. Books and papers are collected. Hats are retrieved. Bodies turn towards the door. Light enters the room as though the world beyond it is waiting.

Nothing dramatic has happened, and that is precisely what makes the painting so moving. Forbes captures one of the oldest promises we make to the young: learn first, and then go out into the world.

More than a century later, we are still making that promise. Only the world outside the classroom has changed.

Yesterday, I saw something that stayed with me: a graduate being pursued over student debt.

I did not know the circumstances of the debt, and they matter. In Britain, ordinary student-loan repayments are tied to income, and those earning below the relevant threshold are not normally required to make them. Arrears can arise in other circumstances. But it was not the mechanics of that particular account that disturbed me. It was the image itself.

A young person had done what generations had been encouraged to do: study, qualify, prepare, invest in yourself. Somewhere at the other end of that journey stood a financial obligation. What seemed considerably less certain was the opportunity that obligation had been supposed to help create.

For decades, the proposition offered to young people in Britain has been remarkably straightforward: work hard at school, go to university, earn a degree, enter a profession, build a career and repay the cost of your education from the prosperity that education helps you achieve.

It was never a guarantee. No degree could promise a particular salary or career. But beneath the system sat an assumption powerful enough to persuade millions of families that taking on substantial student debt was rational. Education would improve your chances. The qualification would open the door. Work would be waiting somewhere on the other side.

Today, that assumption deserves examination.

By April 2026, hiring across the UK had fallen 14 per cent compared with a year earlier. Of 38 sizeable entry-level occupations examined in a government analysis, 30 were hiring fewer people.

The declines become more revealing when we look at the professions. Entry-level hiring for accountants had fallen 29 per cent; graphic designers, 28 per cent; software engineers, 27 per cent; product managers, 24 per cent; data analysts, 15 per cent; and legal assistants, 14 per cent.

These are not obscure occupations at the margins of the economy. They are precisely the kinds of careers for which generations of young people have been told that education, technical skill and professional qualification would prepare them.

Meanwhile, the average student-loan balance for borrowers who finished their courses in 2023 and became liable to repay in April 2026 was approximately £47,900. The question is whether the old pathway connecting that debt to economic opportunity remains as reliable as we continue pretending it is.

Artificial intelligence complicates that question enormously.

It would be easy, and wrong, to say that AI has simply taken the jobs. Britain’s labour market is weak for several reasons. Employers are hiring cautiously. Economic growth has been difficult. Vacancies have fallen. Entry-level hiring broadly reflects that wider deterioration, and even government researchers caution against treating the simultaneous rise of artificial intelligence and decline of junior employment as proof that one has caused the other.

But something else is beginning to appear in the data.

Research from King’s College London found that companies with workforces highly exposed to artificial intelligence reduced junior employment more sharply than overall employment. Those firms were also considerably less likely to advertise new vacancies.

The pattern matters because technological disruption does not require the wholesale disappearance of professions.

AI does not have to replace the accountant; it only has to reduce the number of graduates the accountant needs beneath her.

It does not have to replace the lawyer; it only has to perform enough research, summarising, document review and drafting that the firm recruits fewer junior assistants.

It does not have to replace the creative director; it only has to make one entry-level design position unnecessary.

There may never be a morning when Britain wakes to discover that artificial intelligence has taken millions of professional jobs overnight. The transformation could be much quieter. A company that once recruited ten juniors recruits six. A manager who once needed three assistants discovers that one employee equipped with artificial intelligence can perform much of the work. Another graduate intake becomes smaller.

Nobody announces a revolution. The door simply becomes narrower.

This is worth returning to in Forbes’s painting.

The children do not rush from the schoolroom. They gather at its edge, caught in that small interval between instruction and whatever comes afterwards. Light enters the room, but Forbes does not show us the future waiting beyond it. We see only the children preparing to meet it.

In 1889, that uncertainty belonged naturally to childhood. No one can know what a child will become. Today, the uncertainty is different.

We can educate a young person for fifteen or twenty years, teach them to specialise, encourage them to borrow against their future earnings, and still be unable to tell them whether the first job through which that education becomes experience will exist when they arrive.

The light is still there, and so is the doorway. What has become less certain is what waits beyond it.

That raises one of the most consequential questions of the AI age: what happens when we automate the bottom rung of the ladder?

Every profession depends upon beginnings. The senior lawyer was once the junior lawyer. The partner was once the trainee. The architect once assisted. The accountant once reconciled figures. The programmer once wrote elementary code. The inexperienced became experienced because somebody allowed them to perform work while they were still learning how to do it.

Efficiency asks an entirely reasonable question: why should a highly paid professional spend hours doing something a machine can accomplish in seconds?

But society must ask another: if the machine performs the work through which beginners traditionally became professionals, where will the next generation of professionals come from?

That is larger than unemployment. It is a problem of succession.

We have built an entire professional world around the assumption that experience must be acquired, while simultaneously discovering technologies capable of eliminating many of the tasks through which experience was traditionally acquired.

The contradiction is easy to miss because it can look like efficiency from inside an individual company. For the employer, recruiting fewer graduates may make perfect economic sense. For society, thousands of employers making the same perfectly rational decision may produce something very different: a generation with nowhere to begin.

Beginning matters. We have constructed much of childhood and education around preparing for it. We tell children to concentrate. We measure their progress. We celebrate examination results. Later, families watch sons and daughters cross graduation stages in gowns while names are called and certificates are handed over.

For a moment, everything looks exactly as the promise said it would. Then the gown is returned and the applications begin: upload the CV, create an account, rewrite the personal statement, attach the degree, demonstrate experience for an entry-level position, submit, wait, receive the rejection, repeat.

It is easy to turn this into a moral judgement about young people. They chose the wrong degree. They expect too much. They should retrain, become entrepreneurs or learn to use AI.

Some of that advice may be sensible. But every generation has been remarkably skilled at turning structural change into personal failure. When housing becomes unaffordable, we discuss spending habits; when wages stagnate, we discuss ambition; when secure employment disappears, we discuss resilience; when graduates cannot find professional work, we ask why they studied what they studied.

Perhaps occasionally we should turn the question around: what kind of economy did we ask them to prepare for, and how closely does it resemble the one waiting for them now?

This is not an argument against artificial intelligence.

AI may accelerate scientific discovery, improve medicine, increase productivity, democratise expertise and allow individuals to accomplish work that once required enormous organisations. Human beings have always built tools to reduce the labour required to accomplish something.

But progress and fairness are not the same thing. A technological revolution can ultimately enrich civilisation while destabilising the people who enter adulthood during the transition.

Eventually, new industries emerge. New professions appear. New forms of work become imaginable. Eventually is one of history’s most comforting words.

But history measures technological revolutions in decades. Human beings experience them on Mondays.

Rent does not wait for the new economy to emerge, and neither does adulthood. The person who borrowed tens of thousands of pounds to prepare for the old economy cannot return the degree and begin again. The years have already been spent. The debt has already been created.

That is why the student-loan question cannot remain separate from the transformation of work.

Education has value far beyond employment. A civilisation should want people to understand history, literature, mathematics, science, philosophy and art whether or not every subject produces an immediate financial return.

But if university is primarily a public and intellectual good, perhaps we should stop speaking about every degree as though it were principally a private financial investment. And if it is an investment, we cannot dismiss the return when asking young people to assume the risk.

We cannot tell an eighteen-year-old that borrowing for university is sensible because graduates earn more and then, when the labour market changes, tell the twenty-two-year-old that employment was never promised.

Technically, it wasn’t. Morally, the question is more complicated, because societies make promises without ever writing them down.

There is another consequence of this uncertainty that receives less attention.

Listen to the advice given to young people worried about their futures and certain professions appear again and again: healthcare, nursing, medicine, allied health, care.

The reasoning is understandable. People will still become ill. Populations will age. Hospitals will still need nurses, doctors, physiotherapists, radiographers and other skilled professionals. These are occupations requiring human judgement, physical presence, trust and care in ways that make complete automation far more difficult.

So healthcare begins to acquire another attraction beyond vocation: safety.

For some young people, the question is no longer simply, What am I good at? What fascinates me? What kind of contribution do I want to make? It becomes: Which profession is least likely to disappear?

That is a profound change in the psychology of education.

A young person may love graphic design but wonder whether generative AI will reduce the opportunities available to them. Another may be fascinated by software engineering but watch entry-level recruitment contract. Someone drawn towards writing, publishing, law or administration may hear repeatedly that machines will soon perform much of the junior work.

And somewhere in that uncertainty comes the advice: go into healthcare. People will always need healthcare.

There is truth in it. But there is also something sad about it.

A civilisation needs people who genuinely want to become nurses and doctors. It needs people with the temperament to care for frightened patients, exhausted families, ageing bodies and human beings at their most vulnerable. Healthcare should not become merely the bunker into which an anxious generation retreats because everything outside it appears vulnerable to automation.

Nor can an entire generation become healthcare workers. We still need architects, artists, engineers, writers, teachers, designers, lawyers, researchers, historians, accountants and entrepreneurs.

The purpose of technological progress cannot be to compress human ambition into the handful of occupations machines find most difficult to perform. Education should expand the imagination of the young, not teach them to organise their lives around what an algorithm cannot yet do.

The question we should want a young person to ask is What might I become? The question we risk leaving them with is What will still be safe by the time I get there?

The answer is not to tell young people to abandon university, nor to fear artificial intelligence. It is to stop pretending that the bargain has not changed.

If technology allows us to accomplish more with fewer people, that may ultimately be one of its great achievements. But a society cannot celebrate the efficiency of removing work while continuing to organise dignity, independence, housing, security and adulthood almost entirely around having it.

Something eventually has to give.

Universities may have to reconsider what they teach and what they charge. Employers may have to accept greater responsibility for training young people rather than demanding experience before offering them the opportunity to acquire it. Governments may have to reconsider how education is financed when the economic assumptions used to justify its cost become less certain.

And there is a larger question still. If artificial intelligence makes human labour dramatically more productive, who should benefit from that productivity? The company that employs fewer people? The shareholder? The consumer? Or should some portion of that extraordinary gain belong to the society whose workers, institutions, knowledge and generations of human creativity made the technology possible?

These questions will become harder to avoid because the children entering classrooms today will graduate into a world we cannot confidently describe. Some will enter professions that do not yet have names. Others may discover that the profession they spent years preparing to enter requires far fewer beginners than it once did.

They should not be frightened of that future, but neither should we lie to them about it.

More than a century ago, Elizabeth Forbes painted children preparing to leave school. They gather their belongings and turn towards the door. The lesson is finished. Outside lies everything they have not yet become.

That is why School Is Out feels unexpectedly contemporary now. Forbes captures the moment before a young person discovers whether the world has kept the promises adults made about it.

We cannot promise every child a particular career or guarantee prosperity. We cannot stop technology from changing the nature of work, nor should we try. But surely we can promise something more fundamental: that we will not ask one generation to carry the full cost of a transition they did not create.

If we encourage them to study, the education must still mean something. If we encourage them to borrow, we must be truthful about what waits at the other end. If we automate the first rung of the ladder, we must create another way for them to climb.

And if the old bargain no longer works, then it is the bargain that must change—not simply the young people who believed in it.

The degree will arrive. The technology has already arrived. The bill will certainly arrive.

The question is what will be waiting on the other side.