Everyone Wants AI-Ready Workers.
Nobody Wants to Pay to Train Them.

The paradox

There's a PwC analysis going around: employers now want entry-level workers with senior-level skills in the age of AI. PwC read over a billion job ads across 27 countries. AI-exposed entry-level roles are now seven times more likely to demand senior-level judgement; those "seniorised" openings have grown 35% since 2019 while ordinary entry-level roles shrank. The new work leans on what AI can't do — judgement, empathy, the call you make when the playbook runs out.

Fine. We can all nod at that. It's true, and it's not even surprising.

The real question

But it skips the only question that matters. You cannot hire senior judgement at an entry-level desk — it doesn't arrive at birth. Someone has to build it. So the real question isn't what skills the AI era needs. Everyone already knows what it needs. The question is: who pays the cost to train these people?

Look at the four candidates, because the answer matters more than the trend.

Universities: too slow

Too slow, by design. I watched this happen with live-selling. By the time institutions woke up and started talking about a proper academic program for live hosts, the wave had already peaked — the skill had been learned on the job, in the dark, by people who couldn't wait for a syllabus. Curriculum committees move in years. AI moves in months. By the time a degree exists, it's describing a world that's already gone.

Businesses: won't pay

Will they train their own staff? Be honest about the market we're in.

Most owners do the quiet math and decide not to.

Government: budget without expertise

Isn't public upskilling budget exactly for this? This is where everyone instinctively points, and where the conversation goes straight to numbers: how big the allocation, which agency, how many seats. But that skips the prior question, the one that actually decides whether the money works: where is the expertise?

You can allocate funds in a budget line. You cannot allocate expertise that does not yet exist. In the AI era, everyone is back at the same starting point — the officer drafting the grant, the trainer cashing it, the worker in the seat. There is no shelf of seasoned AI experts waiting to be deployed with public money; the technology is too new for anyone to have twenty years on it. So a budget without expertise doesn't buy training. It funds the guru economy — repackaged hype, now with a government invoice attached.

The individual

So the bill rolls downhill. Universities can't catch it in time. Businesses won't carry it. Government can write the cheque but can't manufacture the expertise to spend it well. Which leaves exactly one party standing at the bottom of the slope.

The same person who's expected to show up already senior is the one now responsible for making themselves that way — on their own time, with their own money, while still doing the day job that AI is busy reshaping under them.

The bill nobody will pay

That's the part nobody says out loud. The cost of the AI transition doesn't disappear because no institution wants to pay it. It just gets pushed onto the person with the least power to refuse — and the least margin to absorb it.

I run an AI company. I'm not above this; I live inside it. And I keep landing on the same uncomfortable conclusion: everyone agrees on what the AI era demands, and nobody is volunteering to pay for it. The market has decided what it wants. It just hasn't decided to fund it.

Which leaves the individual holding a bill they didn't choose — and a harder question still unanswered: where do you even get the right education for this? Not the prompt tricks. The real thing.

Published by IMA AI — June 2026. PwC figures: 2026 Global AI Jobs Barometer.