HOW AI takes jobs

Graduate vacancies down 46%. The jobs that never get advertised.

Join my next live AI webinar

We’re starting the next cohort of the AI Workshop Kit next Tuesday

Learn more

OR Join the next webinar

We’re getting more data that points toward AI impacting jobs.

Not a crude “AI took ma job” but something more subtle (and dangerous).

Graduate vacancies on Adzuna in Britain fell from 15,397 to 8,383 in a year. Down 46%. In one year.

So you get the degree, send the applications, and find that the first rung has gone. The jobs just aren’t there.

The BBC headline is bleak. But blaming that entire fall on AI would be dishonest. You know me better than that! Britain’s (and many other countries’) wider jobs markets are already a mess.

The warning I'm watching is more specific: companies keeping their experienced people and quietly deciding they don't need anyone underneath them.

This is what we need to watch for as it’ll point to AI’s real impact.

Nobody fired Deborah

Stanford's latest research found no widespread, economy-wide displacement from AI. Plenty of people stop reading there. That’s the headline and enough for people to say “cool, this whole AI thing isn’t a threat”.

Careful!

The very next point in the study is from US payroll data showing employment among 22-to-25-year-olds in highly AI-exposed jobs was 19% below where it would have been had it kept pace with less-exposed work. Most of the adjustment came through reduced hiring.

It’s young people who are the canary in the coal mine.

This is actually a canary “resuscitation” chamber but I thought it was interesting! Poor little bugger.

You don't arrive at work to discover Deborah from accounts has been replaced by a robot. Nah. Too crude. Instead Deborah gets AI tools, clears more work, and the business shelves its plan to hire a graduate.

From that company's point of view, I get it. Lower costs. Work still gets done. No months spent training somebody new.

Also remember junior work is where people learn. Researching a market, producing a first draft, checking a case. Boring, documented work that somebody senior can check. Exactly the sort of work language models are getting good at.

The experienced partner knows which client is about to explode and when the textbook answer will fail. They acquired that judgement through years of doing the work, getting things wrong and having somebody correct them.

Remove those years at the bottom and...where do the next seniors come from?

Everybody wants experienced people. Nobody wants to pay for the period in which they become experienced.

Now imagine every company making that decision. Down the line this collapses companies, then industries then potentially the economy.

The thing is we won’t know if this is happening until the world’s economy starts to stabilise and grow. Periods of growth normally lead to more hiring - makes sense…there’s more demand out there so get hiring to produce more and sell more.

This time, for the first time in human history, companies have an option.

They don’t need more labour as there is an alternative: AI. They can give their existing team better AI and leave the junior vacancies closed.

So: if the wider market recovers while those roles stay flat, I'll be a lot more worried. This is the test.

And yes, technology creates new jobs. But AI can do that new work too. I don't think "we'll all retrain" is a good enough answer when we're removing the places people get their first experience.

We are not the horse cart drivers. We are the horses this time.

What to do?

If you are in a job: keep it. I'm certainly not telling you to quit into this market. Jobs are going to become more and more valuable.

BUT: I am telling you to start building something you control alongside it. One freelance client. A small service. Something useful that somebody will pay you for. Aim for your first £500 outside your salary, rather than a million-pound business on day one.

Because AI is double edged. It’s wonderful for employers because they can cut one of their largest costs: people. That’s terrible for the staff of course. And there is not much they can do about it - they lack control over macroeconomic forces and technological progress. Sure they can slow it down but for how long?

So we as individuals need to use these same tools.

The same tools that let a company hire fewer people let you test an idea for very little money. You can have a rough version ready this afternoon and find out whether anybody wants it. There has never been a better time to start SOMETHING and generate extra income.

I've put the full argument, the charts and what I'd do next in the 25-minute video over on Youtube.

And if you are looking to rapidly find a direction I’ve put together a tool for you here: Freedom OS.

To the Task,

Kyle