Something is changing in how careers begin.
For most of the last century, the first job was not really about the job. It was a learning environment with a salary attached. The junior analyst building spreadsheets nobody read. The trainee sitting in on meetings without speaking. The assistant drafting a document that would be rewritten four times.
All of them were doing low-stakes work whose real purpose was to accumulate something that cannot be taught directly. Judgement. Context. A feel for how things go wrong.
Those are the roles AI is best at.
The rung, not the ladder
The pattern showing up in the data is oddly specific. In the occupations most exposed to AI, employment for workers in their early twenties has fallen sharply — while older workers in the same jobs have held steady.
And it is not happening through redundancies. It is happening because fewer people are being hired in the first place.
Nobody is being pushed off the ladder. The bottom rung is quietly being removed.
This is visible across very different labour markets — the United States, the United Kingdom, India. It is not yet an economy-wide collapse, and reasonable people disagree about how much of it is AI versus an ordinary downturn. But if you are twenty-three and applying for your first job, that distinction is academic.
Two reactions, pulling opposite ways
People are responding in two directions at once.
The first is defensive. More certifications. More degrees. More bootcamps. Applications to law and business schools have surged. University applications are at record highs. In the UK, the average graduate vacancy now attracts around 140 applicants.
This is rational. When the field narrows, you signal harder.
But notice what it is: a queue. It works less well the more people join it. A hundred and forty applications per vacancy does not get solved by making application number 141 slightly more decorated.
The second reaction is quieter. Across recent global surveys, almost every young worker now says a sense of purpose matters to their job satisfaction — and that figure has climbed sharply in just three years.
Part of what is driving it is loss. Around half of workers say AI has already automated work they found meaningful and would rather have kept. The parts being removed are not only the tedious parts.
So people are reaching for meaning partly because it is being taken from them.
Why the quiet reaction is the stronger one
The usual case for following your interests is emotional: you will be happier. That may be true, but it is not what makes it a competitive argument.
The competitive argument is about time.
The things AI does not replicate well — judgement, context, knowing which problem actually matters — are not acquired through instruction. They build up over years of engagement that nobody is forcing you to do. The reading you do because you want to. The problem you keep turning over on a walk. The tenth version of something nobody asked you to improve.
That kind of engagement cannot be manufactured from discipline alone. Discipline gets you through a semester. It does not reliably get you through a decade.
What sustains a decade of voluntary depth is genuine interest.
So interest is not a luxury sitting on top of your career strategy. It is the input to the only strategy with a long shelf life.
The honest caveat
Not everyone can act on this equally.
Following your interests through the lean years is far easier when someone can catch you if it goes wrong. That safety net is unequally distributed — across families, across countries, across circumstances. Anyone who tells you to simply follow your passion without saying so is selling something.
There is a second caution. If you treat your interest as a single true calling waiting to be discovered, the first genuinely difficult year will feel like a verdict — proof you chose wrong. If you treat it as something you are building, the same year feels like a Tuesday.
The difference is not motivational. It is the difference between quitting and continuing.
What we would actually suggest
Not a choice between the two reactions. Both.
Take the certification. Sit the exam. Get the degree if it opens a door that is otherwise shut. Income security is not a distraction from your ikigai — for most people it is what makes pursuing one possible at all.
But while you are standing in that queue, do not let it eat everything. Protect some part of your attention for the thing you would engage with anyway. Keep it alive even if it currently earns you nothing.
That is not sentimentality. It is where the compounding happens.
The credential gets you considered. What you have gone deep on is what makes you hard to replace.
And treat that interest as something you are growing rather than something you have found. This is closer to what ikigai originally meant in Japanese, long before the four circles: not a hidden purpose waiting to be uncovered, but something that grows out of what you decide to care about, day after day.
The rung is missing. That is real, and it is not your fault.
But the ladder was never the only way up. It was just the fastest one, for a while, for people who happened to be standing near it.
Sources and credit
This article borrows heavily from the work below. We have kept the argument free of citations to make it easier to read, but none of the claims are ours — they belong to the researchers and organisations listed here, and we would encourage you to read them directly.
On the disappearance of entry-level roles
- Erik Brynjolfsson, Bharat Chandar and Ruyu Chen, Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence — Stanford Digital Economy Lab. The source of the finding that young workers, and only young workers, are losing ground in AI-exposed occupations.
- ISE Student Recruitment Survey 2025 — Institute of Student Employers, UK. Where the figure of roughly 140 applications per graduate vacancy comes from.
- The Labor Market for Recent College Graduates — Federal Reserve Bank of New York. Ongoing data on graduate unemployment and underemployment.
- Evaluating the Impact of AI on the Labor Market — The Budget Lab at Yale. The most serious counter-argument to the above, and the reason we say this is not yet an economy-wide collapse.
On the search for meaning at work
- Global Gen Z and Millennial Survey 2026 — Deloitte. Surveys more than 22,500 people across 44 countries; the source for the rise in how much purpose matters.
- Work AI Index 2026 — Glean Work AI Institute. Where the finding on automated meaningful work comes from.
- Are We Automating the Joy Out of Work? Designing AI to Augment Work, Not Meaning — CHI 2026. On why the tasks people enjoy are disproportionately the ones being automated.
On interest, persistence and the limits of passion
- Christopher Nye, Rong Su, James Rounds and Fritz Drasgow, Vocational Interests and Performance — Perspectives on Psychological Science. Sixty years of evidence that fit between a person’s interests and their work predicts how well they do it.
- Paul O’Keefe, Carol Dweck and Gregory Walton, Implicit Theories of Interest: Finding Your Passion or Developing It? — Psychological Science. The research behind our second caution: people who think interests are discovered give up faster than people who think they are built.
- Erin Cech, The Trouble with Passion: How Searching for Fulfillment at Work Fosters Inequality — University of California Press. The strongest challenge to everything this article argues, and the source of our caveat about safety nets. Worth reading in full if this topic matters to you.