We have never known more precisely which skills the AI economy will require, or less about how young people will come to possess them.
That gap has been on my mind for some time, and Ewan Simpson’s recent paper, When Is the Evidence Enough? Youth Labour Markets, AI, and the Cost of Waiting for Certainty, sharpened it. His opening line puts it plainly: “Something is going wrong at the point where young people enter the labour market.”
The World Economic Forum, the OECD, national skills councils and industry bodies have converged on a well-built skills architecture, and a parallel infrastructure of reskilling strategies and talent accelerators has grown up around it. Far less rigour has gone into a simpler question: where and how do people aged 18–25 acquire, apply and transfer those competencies?
That question is now urgent. Entry-level roles, long the place where textbook knowledge became working judgement, are under the most pressure in the occupations most exposed to AI; the Burning Glass Institute’s No Country for Young Grads documents how hard that first step has become. Yet the pathways that carry this load (work-integrated learning, apprenticeships, competency-based models and, above all, mentorship) are still treated as delivery detail rather than economic infrastructure.
Experts Who Never Emerge
The WEF’s own education work shows the problem in miniature. Shaping the Future of Learning (2026) lists sixteen conditions for education readiness, including employer and career partnerships and lifelong, student-driven learning. It is also unusually honest about the ground beneath them. Promising results from controlled studies rarely survive at system scale, and the first students to enter higher education after generative AI arrived are only now graduating, so the evidence is still thin.
Its sharpest warning is that AI may not replace experts so much as stop them from ever emerging.
That warning has an economic logic and warning. Expertise rests on tacit judgement, and tacit judgement is learned by doing the routine, codifiable work that once made up a first job: drafting, data entry, basic research, customer queries. As the LSE’s Grace Lordan points out, these are the tasks AI absorbs first. Each firm that automates them captures the saving, while the cost, a thinner pipeline of future experts, is borne by everyone.
Nor will the problem pass with the economic cycle. In Lordan’s words: “AI is not a recession. It does not pass.”
The blueprint, then, has been drawn on ground no one has surveyed. That calls less for planners than for builders: leaders prepared to construct new entry routes, test them under real conditions, and rebuild what fails.
The timing matters. The generation now entering the labour market is the first to find the bottom rung itself in question. A weak hiring cycle is only part of the reason; the larger part is that the work through which entry was once earned is being redesigned.
And the frameworks that will govern their pathways are being drafted largely without them in the room, even though, as the WEF’s Youth Pulse 2026 makes clear, young people have plenty to say about the world of work they are inheriting.
Waiting for Proof Is a Decision
To be fair, the evidence has not yet proven that AI is the main cause of young people’s struggles to find work. Slow growth, rising hiring costs, health pressures and post-pandemic caution all play a part.
But as Simpson argues, we have crossed the threshold for action, even if not for final attribution. Enough canaries in enough coal mines should change behaviour.
None of this requires waiting for a new skills framework. It requires organisations, educators and policymakers to treat the first rung as what it has always been: the economy’s learning infrastructure. Four moves would turn it from a casualty of AI into a launch pad.
1. Redesign the Rung, Don’t Remove It
When AI takes over routine tasks, the answer is to rebuild entry roles around what remains: supervised judgement, client contact, checking and improving AI output, and problems without a clean answer. Firms that cut junior roles outright are already finding the saving comes with a bill attached, as Fast Company reported this year. Cognizant and Pearson’s AI Workforce Pulse frames adaptability as the new imperative, and a well-designed first job is exactly where adaptability is built.
2. Treat Mentorship as Infrastructure
Mentorship is how tacit judgement passes from one generation to the next, and it is the pathway most at risk when junior roles disappear. It cannot run on goodwill and spare time. It needs to be designed, resourced and written into senior people’s objectives, in the same way we budget for systems and training.
3. Measure Conversion, Not Placements
Counting how many young people enter a programme tells us very little. The measure that matters is conversion: how many go on to do work that requires independent judgement, and how quickly they get there. Track that, and we will soon see which pathways work and which only look busy.
4. Build It With Young People, Not for Them
Many of those entering work are more fluent with AI tools than the managers designing their jobs. Research in Harvard Business Review argues that curious employees, wherever they sit in the hierarchy, can drive digital transformation when given the room. Bring young people into the design of entry routes as co-designers, not consultees.
The Question for Every Leader
If we get this wrong, we will have described the skills of the future economy in exquisite detail and quietly dismantled the mechanism by which a generation was meant to acquire them. If we get it right, the people we nearly locked out may be the ones who lead us through.
Are our organisations building the first rung, or pulling it up behind us?
References
Carnegie, M. (2026). Companies replaced entry-level workers with AI. Now they are paying the price. Fast Company.
Cognizant & Pearson (2026). The AI Workforce Pulse: The Adaptability Imperative.
Hill, L. A., Menon, S., Le Cam, A., Begag, L. & Grazina, K. (2026). Empower Your Most Curious Employees to Drive Your Digital Transformation. Harvard Business Review.
Levanon, G., Sigelman, M., et al. (2025). No Country for Young Grads. Burning Glass Institute.
Lordan, G. (2026a). AI means fewer graduate jobs: how to make sure they still exist in the future. LSE British Politics and Policy.
Lordan, G. (2026b). Is AI behind youth unemployment? LSE British Politics and Policy.
Simpson, E. (2026). When Is the Evidence Enough? Youth Labor Markets, AI, and the Cost of Waiting for Certainty. Narxoz Business School.
World Economic Forum (2026). Shaping the Future of Learning: Education Readiness for the Age of AI.
World Economic Forum (2026). Youth Pulse 2026: Insights from the Next Generation for a Changing World.
