Who gets the new jobs?

A few weeks ago I was in a zoom full of people who are supporting each other to Get Good With AI. Getting together to make sense of something that is moving faster than any of us can keep up with. If you want to join us, drop me a line, I've got a discount code.

It was an interesting and inspiring conversation, even if many of us teeter on the fence regarding the benefits vs the risks which I've written about before.

Someone shared a statistic that is really interesting.

Roughly 60% of the jobs that exist today didn't exist in 1940. This stat comes from MIT economist David Autor and colleagues, who spent years building a database from eight decades of US census records, tracking every new job title that appeared and where it came from. You can read the full 118 pages here. Six in ten jobs that people are doing right now, jobs that feel ordinary and established, simply weren't jobs eighty years ago.

The reason that stat gets deployed so often in conversations about AI is obvious. History says technology creates more jobs than it destroys so you can relax. It'll be fine.

But I had a look at where Autor's stat came from and what the research showed, and I realised that there's a part that gets left out. There are two very different stories of new job creation since 1940.

Before 1980, new work spread across the income and education spectrum. New jobs appeared in production, in clerical roles, in the middle of the labour market. People without degrees got a foothold. The ladder had rungs all the way up. After 1980, that changed. New work opportunities were split between high-paid professional roles for graduates at the top and low-paid health and care services at the bottom. The middle stopped generating a smaller proportion of new jobs. The effects of automation have eroded demand, Autor's team found, and this has intensified over the last four decades.

AI is arriving into an already polarised labour market. Not the 1950 labour market, not even the 1990 labour market. This one which is already hollowed out in the middle. Whether AI will create new jobs is one question. The other question is: for whom?

I was working with a small charity earlier this year, where paper-based data systems are still the reality. The team there is under-resourced, under-supported, and the AI conversation is largely happening without them. Meanwhile, some of the largest professional services firms are reportedly letting go of staff who don't adopt new tools fast enough. Individuals in those firm are being told to adapt or else. Smaller organisations doing essential work on tight budgets are having to fend for themselves.

Enabling people to adapt, in ways that are specific and well-supported and properly resourced, is different from telling them to get on board. It takes longer. It costs more and in a reality where small charities are still using paper instead of Outlook, it requires someone to notice that not everyone is starting from the same place.

As the future of new work arrives, who in your organisation, or your sector, will be in a position to do it? And what would it take to make sure the answer isn't the people who were already fine?

Day of the Week 📆

While I'm writing this, people are voting in key elections in the UK: in the Scottish Parliament and Welsh Senedd and in local elections across England, including London, Manchester, and Birmingham, my home. While I'm writing, people are voting, and by the time you read this, some of the results will be known.

With trust in governments declining around the world, political participation is declining and support for populism (on the right and left of the spectrum) is growing. It's a deeply worrying time.

What am I listening to?👂

I listened to The Squiggly Careers podcast: "How to Change Behaviour at Work." which unpacks a piece of research by James Elfer, Siri Chilazi and Edward Chang on what actually shifts workplace behaviour, and applies it to real team challenges. A practical and genuinely good listen.

What am I reading? 📚

This piece from MoreThanNow, James Elfer's behavioural science consultancy, on recruiting talent from underrepresented backgrounds describes interesting findings. In trials with two large companies, putting diversity and inclusion messaging front and centre in job adverts actually reduced applications, including from candidates from underrepresented backgrounds. Also a photo of a single underrepresented individual performed worse than a photo of a diverse group, apparently because candidates read it as tokenistic. The lesson isn't to abandon diversity efforts in recruitment. It's to test your assumptions before deploying them at scale, because well-intentioned doesn't always mean effective.

What am I watching? 👀

We had a weekend of watching oldish films on the TV, both were free to view here in the UK on Channel Four: Quadrophenia 1979 and The Devil Wears Prada 2006.

Both bearing up well, I think.

😍 Joy-giving things

The bluebells have certainly been spectacular this year.

Here they are in Hampton Wood, Warwickshire, from a couple of weeks ago.

Have a great weekend

Lucy

ChangeOut is created by Lucy Caldicott. You can find more about my work at ChangeOut.org.

If you're a purpose-driven professional wondering whether coaching might help you think through where you are and where you're heading, I'd love to hear from you. I'm opening up coaching spaces later in 2026. Express your interest here.

You can also find me on Instagram, and LinkedIn.

🎬🎬🎬 YouTube 🎬🎬🎬

If you like what you read and you'd like to show your appreciation in cash, you can do that here. I'd be very grateful!