What if AI is not a job problem, but part of the solution?

Hardly a week goes by without another headline about the jobs artificial intelligence will destroy.

Goldman Sachs estimates that 300 million jobs globally are exposed to automation by AI. The IMF says almost 40% of jobs worldwide are exposed to AI. And Sam Altman predicts that entire categories of jobs will disappear.

While this sounds frightening, there is another side.  

The European Commission says labour and skills shortages are already sizeable across countries and sectors. 77% of European companies see them as a barrier to long-term investment. Globally, 72% of employers surveyed by ManpowerGroup struggle to find the skilled talent they need.

This points in exactly the opposite direction: if we increasingly lack the people and skills to do the work, AI could become part of the solution.

So, which of these two forces will matter more?
 

Let us look beyond the headlines

The most substantial global study I know is the World Economic Forum’s Future of Jobs Report. Its latest 2025 edition brings together the views of more than 1,000 employers, representing over 14 million employees worldwide. By 2030, it expects 92 million existing jobs to be displaced. At the same time, 170 million new jobs are expected to be created. Overall, the WEF expects a net increase of 78 million jobs, not fewer.

Technological change has long destroyed some jobs while creating others. And zooming in on AI itself, the WEF expects AI and information processing technologies not only to displace 9 million jobs by 2030, but also to create 11 million new ones.

That is the outlook. What has actually happened since generative AI entered the workplace?

So far, there is little evidence of the predicted collapse in white-collar employment. Over the past three years, white-collar employment has continued to grow. Routine back-office roles are a notable exception, suggesting that the composition of white-collar work is changing, rather than employment declining across the board.

These are US labour-market data reported by The Economist. And some of the details are interesting: the US now employs around 7% more software developers, 10% more radiologists and 21% more paralegals than three years ago, all professions one might have expected to feel the impact of AI.

Three years into the generative AI era, this is at least an indication: the predicted collapse has so far not happened. What happens next remains open.

There is another reason why I struggle to imagine a sudden collapse of the labour market.

The Copenhagen Institute for Futures Studies recently explored exactly such a future in a scenario snippet called «Dario World». It takes us to 2030, where a French insurance company has already laid off half of its workforce as AI rapidly replaces human work.

The interesting part is that the authors conclude that for this scenario to happen, it would require a government responding quickly and coherently, citizens adapting without revolt, and a labour market absorbing the shock without triggering a political crisis.

Remove just one of these assumptions and the scenario starts to collapse.

It illustrates something I find easy to forget when we talk about technology: what is technologically possible does not automatically become reality. An innovation also needs people willing to use it and an economic model that makes it profitable.

And finally, what happens when AI really does replace jobs?

IKEA provides a good real-life example. Its AI chatbot Billie took over 47% of customer-service queries, affecting around 8,500 customer-service roles worldwide.

IKEA then did not ask how to solve the problem of thousands of people being replaced by AI. It looked instead at the opportunity. At the 53% Billie could not handle, and at 8,500 people with deep knowledge of customers and the IKEA range.

Customer-service co-workers were retrained as interior design advisors, providing phone and video consultations and personalised room planning. The new service subsequently generated around €1.3 billion in revenue.

I find the example powerful because AI did exactly what many fear: it replaced a substantial part of human work. But replacing work does not have to mean replacing people.

So, where does this leave us?

None of this means that AI will not replace human work. It already does. And for people whose jobs disappear as a consequence, the impact can be very real.

But looking at the bigger picture, I find the scenario of AI becoming part of the solution to labour shortages much more convincing than the scenario of mass unemployment.

The opportunity lies in reallocating work: letting AI do what AI does better, while people focus on what humans do better. Done well, this can free scarce human capacity rather than simply replace it.

But that requires us to rethink work and, equally important, to reskill and upskill people for the capabilities that will be needed.

Perhaps that is the bigger risk: not that AI takes too many jobs, but that we fail to redesign work and prepare people to make the best of human and AI capabilities.