One of the biggest questions young people are frequently asked is: What do you want to be?
By extension, we are asking a much bigger question:
What do you need to study to become that person?
Implicit in this question is a difficult predicament. Young people must make the best possible guess about what kind of work will still be available several years from now, decide which qualification will get them there, and then commit a significant amount of time—and either their own, their parents’ or a scholarship foundation’s money—to that decision.
Every time I return to the World Economic Forum’s Future of Jobs reports, the outlook has changed. The occupations expected to grow change. The skills employers say they require change. Technology changes what workers do within jobs that continue to exist.
Career planning was never easy. Artificial intelligence has made it considerably more difficult.
In a recent Financial Times article, Sarah O’Connor and John Burn-Murdoch examine whether AI is already affecting what students choose to study.
The initial finding is surprising.
Researchers studying UK university applications between 2020 and 2025 found that, following the release of ChatGPT, applications increased for degrees associated with occupations that have greater exposure to AI. These included information technology, statistics and engineering.
In other words, students appeared to be moving towards some of the fields in which entry-level opportunities were beginning to weaken.
The advice changed while students were following it
When I was a primary and secondary school student, we were told to pursue STEM because that was where the future would be.
Eventually, that advice became more specific: study computer science, choose a practical degree and enter a rapidly growing industry.
Many readers will remember the rise of FAANG—Facebook, Apple, Amazon, Netflix and Google. During the 2010s and into the early 2020s, entry-level software engineers at leading US technology companies could receive compensation packages exceeding $100,000.
That sent a powerful message to students and parents: computer science was not simply an interesting degree. It appeared to be a safe investment.
Many young people listened.
But consider the timeline. A child in secondary school hears that technology is the future. That student selects the relevant examination subjects, completes sixth form or its equivalent, applies to university and then spends another three or four years completing a degree.
By the time that student becomes fully qualified, the ship they were told to board may have already left the port.
Now some of our young people are entering a labour market in which employers can use AI to perform parts of the junior work through which graduates would ordinarily gain experience.
This creates a problem that extends beyond the number of available jobs.
Senior professionals did not begin their careers with senior-level judgment. They developed it by researching, drafting, checking, correcting mistakes and completing the less glamorous work that organisations are now increasingly interested in automating.
If AI removes much of that entry-level work, we must ask how the next generation will acquire the experience necessary to become senior professionals.
Students did not make irrational choices
We should be careful not to look at the present situation and conclude that these students simply chose badly.
There are several possible explanations for the increase in applications to AI-exposed subjects.
Students may have viewed AI as an opportunity and wanted to work with the technology. They may also have been responding to the rising cost of living by choosing degrees traditionally associated with higher salaries. And for much of the last decade, computer science genuinely appeared to be one of the safest educational choices available.
We cannot judge decisions made several years ago using information that only became clear afterwards.
The evidence itself also requires qualification.
The study examined applications through 2025 and is still a working paper. The FT reports that preliminary 2026 figures suggest the earlier pattern may be starting to reverse. Computer-science enrolment in the United States has fallen from its 2024 peak, while preliminary UK application figures also show a substantial decline.
However, it is too early to know whether this represents a lasting change.
AI exposure does not necessarily mean that an occupation will disappear. In many professions, AI may assist skilled people rather than replace them. New occupations may also emerge around technologies and industries that are still developing.
The honest answer is that nobody knows exactly what the labour market will look like in four years.
Yet 17- and 18-year-olds are still required to place their bets.
“Choose the right degree” is no longer enough
As a school counsellor, I think there is a danger in placing too much of this responsibility on the student.
We can tell young people to research the labour market. But which labour market should they research—the one that exists when they apply, the one that exists when they graduate, or the one they will face ten years into their careers?
Employers themselves are still determining how AI will affect their organisations. Universities do not know precisely which occupations will expand or contract. Governments are still trying to understand what these changes mean for education, employment and social policy.
It is unreasonable to expect a teenager to confidently predict what these institutions cannot.
This does not mean that subject choice no longer matters. It means career guidance must become more honest about uncertainty.
Instead of asking only, Which degree will get me a good job?, students should also consider:
- What capabilities will I develop through this course?
- Can those capabilities transfer between professions and industries?
- Will I gain practical experience as well as academic knowledge?
- How is technology changing the entry-level route into this profession?
- If that pathway changes, what else could I do with what I have learned?
- Am I genuinely interested in this field, or am I pursuing it because adults told me it was safe?
These questions do not produce a simple list of secure careers. But perhaps giving students a false list of secure careers is part of what created the problem.
Adaptability cannot become another empty slogan
The easy response is to say that young people simply need to become more adaptable.
That is partly true. But “be adaptable” can become a convenient way for institutions to transfer all the risk onto individuals.
Students are told to keep retraining while employers reduce junior opportunities. Universities continue marketing programmes using career outcomes built during a different labour market. Governments frequently respond only after the disruption has already occurred.
Then, when the promised opportunity does not materialise, the young person is asked why they chose the wrong degree.
Adaptability must be supported structurally.
Universities should be transparent about changing graduate pathways. Employers must consider how junior workers will acquire professional judgment if the tasks that once developed it are automated. Schools need access to current labour-market information, not assumptions inherited from ten years ago. Governments must make continued education and retraining realistically accessible throughout a person’s working life.
The FT article points to Sweden’s support for mid-career learning as one possible model. That matters because a decision made at 17 or 18 should not become a permanent sentence.
Education still matters—but the promise must change
I come from a culture in which education was strongly connected to social mobility.
The message was that if you went to school, worked hard and became qualified, education would provide a route out of poverty and into a more secure life.
I still believe deeply in education.
But believing in education does not require us to keep repeating a promise that the labour market may no longer honour.
The more honest promise is that education should help a person understand, communicate, reason, adapt and contribute. It should provide specialist knowledge, but it should also give people enough intellectual range to move when the world around them changes.
Students will need technical competence. They will also need judgment, communication, creativity, interpersonal ability and the capacity to continue learning. Interdisciplinary education may become increasingly valuable because real problems rarely remain neatly within the boundaries of a single subject.
Young people should think carefully about what they study.
But adults also need to recognise that we may not know—and cannot reliably predict—which choice will remain perfectly safe.
We are asking students to place a bet on a future that none of us can reliably predict. Our responsibility is not to make that prediction for them with false confidence.
It is to build an education and employment system in which one decision made as a teenager does not determine the limits of the rest of their working life.