The AI Catch-22: Experience Wanted, but First Jobs Are Vanishing

For decades, the path from higher education to professional success followed a clear, predictable formula. You spent years acquiring a degree, secured a junior or entry-level position, performed the routine, foundational work, and gradually earned the experience needed to climb the corporate ladder.

Today, that ladder is missing its bottom rung.

As generative AI reshapes the global economy, it is not just threatening traditional blue-collar and manufacturing jobs, it is also quietly targeting the cognitive, entry-level positions historically reserved for recent university graduates. Data entry, basic coding, initial market research, routine drafting, and first-tier customer support, the exact tasks where young professionals once acquired foundational experience are now performed faster, cheaper, and around the clock by AI systems.

This is no longer a prediction; it is measurable. A landmark Stanford study (Brynjolfsson, Chandar and Chen, 2025), based on payroll records covering millions of American workers, found that early-career workers aged 22–25 in the most AI-exposed occupations have suffered a 13 per cent relative decline in employment since late 2022, a figure revised upwards to 16 per cent in the study's latest update while employment for their more experienced colleagues remained stable. Europe is on the same trajectory: in the United Kingdom, the job platform Adzuna reports that entry-level vacancies graduate roles, internships, apprenticeships and junior positions have fallen by roughly a third since the launch of ChatGPT.

This shift has created a dangerous market contradiction: the AI Catch-22.

Employers continue to demand two to three years of practical experience for new hires, yet the traditional entry points where that experience was earned are vanishing. Corporations face a short-term financial incentive to replace junior roles and internships with AI agents. But this short-sighted efficiency triggers a long-term crisis: if young graduates are never given the chance to perform basic work, how will they ever develop the critical thinking, strategic intuition, and leadership skills required for senior roles? Companies are, in effect, eating their own seed corn.

Sceptics will rightly note that every technological revolution has destroyed some jobs while creating new ones. But this wave differs in two crucial respects. It is moving at a speed no previous transition has matched, leaving education and labour-market institutions no time to adapt. And for the first time, it targets cognitive rather than manual work precisely the tasks that have always served as the training ground for professional careers. Tellingly, the Stanford data show that youth employment is falling where AI automates young people's work, but growing where it augments it. The policy conclusion writes itself: Europe must deliberately build an economy of augmentation, not automation.

The stakes for Europe's youth are severe. Higher education systems across the continent are moving linearly in an era of exponential technological change. Thousands of students graduate each year with degrees tailored for a job market that no longer exists, holding skill sets rendered obsolete the day they receive their diplomas. Without policy intervention, Europe risks creating a new "lost generation"  highly educated, heavily indebted, yet structurally unemployable due to a systemic skills mismatch.

To prevent this, policymakers, academic institutions, and industry leaders must act decisively and they do not need to start from scratch. The instruments already exist; they must now be repurposed for the AI era.

Make Erasmus+ the bridge over the experience gap. The European Commission has proposed raising the Erasmus+ budget to €40.8 billion for 2028–2034, an increase of roughly 50 per cent on the current programme. A meaningful share of that money should be earmarked for a dedicated traineeship pillar: paid, structured, AI-augmented placements across the single market, in which graduates learn to manage, audit, and direct AI systems from day one rather than compete against them.

‍Erasmus+ transformed how Europeans study; its next mission must be to transform how Europeans start to work.

Reform higher education curricula. Universities must pivot away from assessing rote knowledge and basic content creation, which AI executes effortlessly, and refocus on advanced problem-solving, critical evaluation, human-AI collaboration, and emotional intelligence. The building blocks are already on the table, the Union of Skills, the Commission's AI Skills Academy, and the AI Act's new AI-literacy obligations, but they must reach every degree programme, not only computer science departments.

Reward companies that train, not replace. European labour policies should tie tax incentives and subsidies to structured graduate programmes focused on mentorship and strategic oversight, drawing on the proven dual-education model of Germany, Austria, and Switzerland. A reinforced Youth Guarantee and the proposed EU rules on quality traineeships should ensure that "entry-level" never becomes a synonym for "expendable".

Generative AI does not have to mean the end of youth employment in Europe. But if European institutions fail to update their labour frameworks and educational models, the next generation will be left staring at a career ladder they can no longer reach. It is time to rethink how we train, hire, and empower young talent before the bottom rung disappears entirely.