The First Rung
Something is breaking at the bottom of the career ladder, and most organizations won't feel it for years.
The first rung is disappearing. At companies adopting generative AI, entry-level hiring has fallen sharply while senior headcount keeps growing. The junior roles that remain are being rewritten to demand skills that used to take a decade to acquire. Researchers call it experience creep. Employers want experience for the very jobs where experience used to be earned.
It's tempting to file this under labor market news. That's a mistake. The entry-level job was never just a line item of inexpensive labor. It was how organizations grew judgment, the one thing they need most and can't buy on demand.
Judgment doesn't arrive by download. It's built by doing real work with real consequences, at a scale where mistakes are survivable, near people who have made those mistakes already. Strip that layer out and the efficiency gain is real, bookable, and immediate. The cost is real too. It's just deferred, unbudgeted, and sitting on nobody's ledger.
We are running down a reserve we don't know how to refill.
I know what that reserve looks like, because I was built by it.
I left school just before my sixteenth birthday and, in the summer of 1987, joined a Youth Training Scheme at the Brooklands Museum Trust in Surrey, England. For a boy with lots of energy and a fascination for history, machinery, and generally how things worked, surrounded by people who had lived remarkable working lives, it was close to perfect.
I don't remember exactly what my role description said. Like most role descriptions, it stopped mattering the day I started and people got to know me. What became clear quickly was that the Trust was full of smart, deeply experienced people from many walks of life, and almost none of them had any practical experience with desktop computers. I did, more or less. I'd taught myself on my parents' business machines, mostly out of curiosity, never imagining it would amount to anything.
Before long a sixteen-year-old was teaching. I was also learning, and this was the part that mattered more. One of the museum's volunteers, Frank, had years of experience on far larger systems and an abundance of patience with me. None of what he passed on was in a manual. He taught me to think about a system rather than a symptom, to sit with a problem instead of rushing past it, and to be useful to people who knew more than I did about everything except the one thing I knew.
I wouldn't have called any of this an on-ramp at the time. I was just a kid who'd been given real work, a gap I could grow into, and someone generous enough to teach me. It's only looking back that I can see the shape of it, and see that none of it required a credential, only that the door was open in the first place.
And the door wasn't there by accident. The Youth Training Scheme existed because someone had thought about the people the conventional path didn't fit and built a way in for them. That seems worth remembering now, because anything that was designed once can be designed again.
I've spent the decades since inside the world that door led to, eventually leading technology functions for global companies. Along the way I hired and worked alongside many people whose degrees mattered enormously, to them and to the work, and I'd never argue otherwise. But the degree was one entrance among several. What worries me now isn't the value of the credential. It's the monoculture. We're closing the learn-by-doing entrance at the very moment we've made the credentialed one the only entrance most employers will accept.
The vanishing first rung would be serious on its own. What makes it dangerous is what came before it.
We spent two decades narrowing the entrances. Jobs that once welcomed capable school-leavers began asking for degrees. Then degrees plus internships. Then the right degree from the right kind of institution, plus experience that a first job used to provide. Each individual hiring decision was defensible. The aggregate effect was a monoculture, one sanctioned path into professional work, expensive to walk and closed to anyone whose life didn't bend to it early enough.
For a while the monoculture held, more or less, because the learn-by-doing entrance still existed alongside it. Plenty of people still found their way in through the side door, the apprenticeship, the junior role that trained you, the unglamorous first job where someone like Frank was waiting. The system had redundancy, even if nobody was managing it as such.
Now the side doors are closing at the same time. AI is absorbing exactly the tasks through which newcomers used to prove themselves and build judgment, and employers are responding by asking the entry level to arrive pre-seasoned. We've made the credential the price of admission while removing the places where the uncredentialed, and the newly credentialed, could become good.
None of this is an argument against degrees. Formal education changes lives, and for many kinds of work it is the right and necessary preparation. The argument is against a system with a single point of failure. A labor market with one entrance isn't a meritocracy. It's a queue, and the people who fall out of it aren't the ones short on talent. They're the ones short on circumstances.
If AI now does the junior work, why carry the cost of junior people? It's a fair question, usually asked by a CFO doing their job, and it deserves a better answer than most of us have ready.
The answer is that none of us will be here forever, and we hold a fiduciary responsibility to an organization that has to outlast us. That responsibility doesn't stop at optimizing operating expense. It includes making sure there is someone capable standing behind the people we have now. The savings from an empty first rung aren't discipline. They're a loan against the successor pipeline, with no plan to repay it.
Framed that way, the question changes shape. It stops being "can we afford to develop people AI could replace" and becomes "can we show continuity of capability the same way we'd show continuity of cash." Most organizations can't, because no one has been given the job.
The answer isn't to romanticize the old entry-level job. A fair amount of it was tedium, and pretending otherwise helps nobody. The answer is to rebuild what the tedium accidentally provided, this time on purpose.
That starts with being honest about what the first rung actually produced. Not output. Output was always the cover story. It produced people who had watched real decisions get made, made small ones themselves, been wrong at survivable scale, and absorbed judgment from people a few rungs up. If AI now does the output, the development still has to happen somewhere. It won't happen by default. It has to be designed.
Some of that design is already visible in organizations taking the problem seriously. Junior roles built around reviewing and verifying AI output rather than producing from scratch, so newcomers learn to interrogate work before they've learned to fear it. Deliberate pairing of early-career people with senior ones, working through problems together rather than delegating downward and hoping. Real work placed into education, through co-ops and apprenticeships, so people build workplace judgment before the market demands they already have it.
And some of the design is older than any of this. The scheme that gave me my start was a policy choice, made by people who understood that talent is distributed more widely than opportunity. That understanding didn't require artificial intelligence to become true. It just requires organizations, and occasionally governments, willing to build doors.
The path of least resistance is clear enough. Companies bank the efficiency, the first rung goes unfilled, and somewhere a decade out a leadership team discovers there is no one behind its senior bench and no way to buy what it stopped growing. The alternative is not complicated, just deliberate. Let AI take the tedium, and design early careers around what they were always for, the making of judgment. The technology permits either. The choice is being made right now, mostly by default, one unfilled junior role at a time.
The first rung was never overhead. It was the foundation everything above it stood on.
