AI Isn't Going to Take all the jobs

Let me start with the good news, because you deserve it before we get into anything complicated.

Economists who have looked seriously at AI and employment, not the headlines, the actual research, broadly agree that AI will create nearly double the number of jobs it displaces. New industries, new roles, new needs that don't exist yet. The same thing happened with the industrial revolution, with computing, with the internet. The jobs that vanished were replaced by ones nobody had imagined.

That doesn't mean the transition will be painless. It won't. There will be a gap between the jobs that disappear and the jobs that emerge, and some people will fall into that gap. That's worth saying honestly, because pretending otherwise isn't helpful.

But the panic, the specific panic that AI is going to hollow out the workforce and leave nothing behind, that's not what the evidence suggests. What it suggests is something more nuanced and, I think, more interesting.

AI is going to take certain kinds of jobs. And when you look carefully at which ones, a pattern emerges that tells us something important about how we've been organising work for the last forty years.

Bear with me.

At some point in the latter half of the twentieth century, something shifted in how organisations were run.

The people doing the doing, the makers, the carers, the builders, the teachers, started to be outnumbered by the people managing the doing. Middle management expanded. Layers of administration appeared. Roles multiplied whose primary function was to coordinate, report on, and measure the work of other people.

This wasn't unique to any one sector, but it was most visible, and most measurable, in the NHS.

In the late 1980s, a political decision was made to run the NHS like a marketplace. To make that work, you needed administrators. People to manage contracts, process invoices, track performance, write compliance reports. None of this delivered a single minute of patient care. But suddenly, all of it was essential.

Before this change, administration accounted for roughly 5% of the NHS budget. Within a decade it had doubled. By the early 2000s it was approaching 14%. On a budget that now sits at £205 billion, that's roughly £25 billion a year spent on the apparatus of running a market rather than running a health service.

And here's the part that should make you think: throughout this entire period, technology was arriving that should have been reducing that headcount. Databases. Email. Digital records. Online systems. The paperwork just went digital, with the same number of people managing it. Sometimes more.

The NHS is just the clearest example. The same pattern played out across local government, financial services, education, large corporates. The doing shrank. The managing of the doing expanded. And the technology that was supposed to make administration leaner just gave the administrators better tools.

Those are the jobs AI is coming for. And in many cases, those are jobs that technology was always going to come for eventually. AI is just the point where it becomes undeniable.

The irreplaceable ones

Now here's where it gets interesting.

The jobs that remain, the ones that genuinely cannot be automated, aren't the management roles. They're the doing roles. The ones we've been systematically undervaluing for decades.

The nurse. The junior doctor at 3am. The cleaner who's been on the same ward for fifteen years and notices when something is wrong with a patient because she knows them. The surgeon whose hands know something her conscious mind hasn't caught up with yet.

Physical presence. Embodied judgement. Relational trust. The things that are built between two humans in a room over time, that have no digital equivalent.

And beyond the caring professions, the trades. The plumber, the electrician, the joiner. The work that requires a body, a pair of hands, the ability to read a situation in three dimensions and respond to what's actually there rather than what the manual says should be there.

And beyond that, the artists. The makers. The people who create with soul and imperfection and beauty. The ceramicist whose thumb leaves a mark that no machine would allow. The musician who plays slightly behind the beat in a way an algorithm would correct but a human feels. The writer whose sentences go somewhere unexpected because they followed a thought into the quiet and found something there.

These are the things that are precious precisely because they couldn't have been made any other way. And in a world where everything that can be automated is automated, they don't become less valuable. They become more.

What if the correction is coming?

The money was never not there. In the NHS, in local government, in organisations everywhere, it was being spent on the architecture of systems that technology was always going to replace. If AI absorbs that work, and does it better and cheaper, that isn't displacement. That's correction.

What if the jobs that emerge on the other side are the ones that should have been valued all along? Properly paid nurses. Respected tradespeople. Artists and makers whose work is understood as irreplaceable rather than decorative.

The staffing crisis in the NHS isn't a recruitment problem. It's what happens when you systematically undervalue the irreplaceable while overfunding the replaceable for four decades. The same is true in other sectors and other places.

AI isn't the threat to that world. It might be, if we're brave enough to let it, the thing that finally corrects it.

So is your job safe?

If your work requires you to be present, relational, creative, caring, or skilled with your hands, yes. More than safe.

If your job exists primarily to manage a system that was designed before the technology existed to manage it automatically, that's worth thinking about honestly. Not with panic, but with clarity.

The transition will be bumpy. Some people will need support through it. That's a conversation worth having seriously, and it isn't one I'm dismissing.

But the endpoint, if we get there, looks less like a world where humans are redundant, and more like one where the things that are genuinely, irreplaceably human are finally valued the way they always should have been.

That feels worth working towards.