Are Students Choosing Degrees for Jobs AI Is About to Disrupt?

Fears abound about the impact AI will have on future (and present) employment prospects. Yet at exactly the moment it’s beginning to reshape some of the professions graduates traditionally entered, students appear to be moving towards some of the degrees most exposed to it.

New research revealing UK university application data between 2020 and 2025 suggests that applications to AI-exposed degrees, including areas such as IT, statistics and engineering, actually increased after the launch of ChatGPT in late 2022.

Call it nihilism, fatalism, existentialism, even pig-headedness. Whatever the case, our young are heading into the future full steam ahead. And do you know what? Good on them.

The Changing Shape of Degree Demand 

Research published in January 2026 examined applications across 1,090 degree programmes and linked them to millions of LinkedIn profiles. It was found that degrees with greater exposure to AI experienced greater growth in applications following the release of ChatGPT. A degree one standard deviation above average in AI exposure saw approximately 6% greater growth in applications than a less-exposed degree.

On the face of it, it’s a surprising finding. We are becoming increasingly concerned about AI's ability to automate highly skilled work. However, the students making decisions about their future seem to be moving towards that very work. The effect was not confined to one particular demographic either. The researchers found a similar pattern among men and women and across different ethnic groups.

Young People Keep Their Faith in Universities

Perhaps the most surprising data of all is the fact that overall demand for university places didn’t change substantially during this period. Despite the doom and gloom, many young people are keeping faith in the power of higher education over more skills-based options.

There could be many reasons for this: inertia, the perceived prestige of a university degree, or even the wish to leave home. The trend is certainly worth exploring more.

Technology As the Foundation Stone for Higher Education

So what’s behind these choices? If you’re 18 years old and trying to decide what to study, technology does look like a rather sensible bet. After all, AI is transforming the economy, governments are investing in it, and businesses are scrambling to work out how to use it. The companies building AI systems are among the most valuable in the world. 

The problem is that there is a difference between an industry becoming more important and every job within that industry becoming more secure.

The Degrees That Looked Like Safe Bets

Computer science is perhaps the clearest example. For years, it was almost the textbook future-proof degree. Learn to code and there would be a seemingly endless supply of well-paid jobs waiting at the other end. Then the computers started learning how to code.

The latest UK graduate data though provides a rather awkward footnote to the future-proofing argument. Analysis for the 2027 Guardian University Guide found that the proportion of computer science graduates entering professional coding or programming roles had fallen from around 40% to 28%. The proportion entering any graduate-level occupation had also fallen, from more than 60% two years earlier to 50%.

Those are striking numbers. They don't, however, prove that AI caused the decline. Graduate employment is influenced by the wider economy, hiring patterns and the fortunes of the technology sector. Many of these jobs are boosted by AI, not threatened.

Still, the timing is difficult to ignore. The decline has taken place as generative AI has gone from an interesting novelty to a tool that can write code, debug software, and increasingly handle tasks that once provided an obvious entry point for junior developers.

The underlying Graduate Outcomes data covers more than 350,000 former students, so this isn't a conclusion being drawn from a few anecdotes about graduates struggling to find work. It is a large-scale picture of where recent graduates are ending up, although, again, it can’t tell us precisely why those outcomes are changing.

AI Doesn't Have to Kill a Job to Change It

The assumption that AI will either replace a job or leave it alone is altogether too simplistic. Jobs are simply collections of tasks, and AI can change the economics of those tasks without eliminating the occupation itself.

A programmer may still be needed, but perhaps fewer programmers are needed to produce the same amount of software. A marketing department may still need people, but perhaps fewer people are required to produce the same amount of content. A journalist may still be employed, but the value of being able to produce words quickly may fall when a machine can produce them almost instantly. That distinction matters enormously for graduates because junior employees have traditionally learned by doing the tasks that are now becoming easier to automate.

For a company, asking an AI system to produce a first draft of some code is one thing. Asking it to take responsibility for an entire software system is another. The experienced engineer who understands the architecture, the customer and the consequences of getting something wrong may become more valuable. But what happens to the graduate who was supposed to spend three years learning the less glamorous tasks before becoming that experienced engineer?

That is one of the less discussed problems with AI. It may not simply replace experienced professionals. It may change the ladder by which people become experienced professionals in the first place. 

So What Should an 18-Year-Old Actually Study?

This is where the idea of an “AI-proof” degree becomes unhelpful. There are certainly professions that appear less exposed to automation. Jobs involving physical environments, human relationships, responsibility, and interpersonal judgement are generally harder for AI to perform independently. Medicine, nursing, teaching, and many engineering roles are often presented as relatively resilient.

The more useful question, then, is which degree is least likely to be affected by AI. Rather, it’s which degrees give students something that remains useful when the tasks associated with their profession change.

That could come in the form of deep subject knowledge, the ability to understand complex problems, communicate clearly, exercise judgement, or work with other people. It could also mean simply being comfortable enough with technology to use new tools rather than being displaced by them.

The Problem With Predicting 2040 From 2026

The crux of the matter is that asking an 18-year-old to make that decision while simultaneously acknowledging that we don't know what AI will be capable of in five years is absurd. After all, nobody knows what the future holds. Even the companies developing the technology don't know (or don’t want to admit) exactly where it will take us. And yet young people still have to choose.

It is therefore tempting to look at these students and accuse them of making the wrong bet. Why study computer science if AI is coming for programmers? Why study statistics if machines are becoming better at analysing data? Why study engineering if AI is increasingly capable of designing things? But all of this finger-pointing assumes we know what the right bet is.

The Bet Has Already Been Placed

Perhaps, then, the most interesting thing about these university application figures isn't that young people are choosing degrees that AI might disrupt. It's that they are choosing them despite knowing that AI might disrupt them.

They’ve been told countless times that the future is uncertain. And their response appears to be: fine. Whether that turns out to be prescient or foolish is something we will only know later. The students making these choices certainly don't have access to some secret map of the labour market in 2035. They’re making decisions with the same imperfect information as everyone else. There is something tremendously admirable about that.

That world as we know it may already be disappearing. And if it is, a little pig-headedness might not be the worst quality to take into it.