Philippe Aghion was awarded the 2025 economic sciences prize for his work on creative destruction, a process where new innovations build upon, and take over from, previous innovations – creating economic growth. Here, he reflects on the role of creative destruction in the AI era, and what societies need to do to make the most of AI.
How can AI drive economic growth?
Philippe Aghion: AI makes it easier to produce goods and services because you automate a number of tasks. You’re replacing scarce human capital with physical capital, which is less scarce. That is one reason why AI boosts growth. The other reason is that new ideas often result from recombining old ideas. With AI, you can recombine much more.
A very interesting experiment is AlphaFold, which is a way to create new proteins not through true experiments, but through AI. This is a revolution in medical science and it creates many new possibilities. There is the potential to overcome a number of illnesses much faster.
“AI makes new ideas easier to find and new prototypes easier to conceive. It’s a fantastic revolution.”
Philippe Aghion
In 50 years, how will we look back at this time in history?
Philippe Aghion: It’s very hard to predict. It all depends on what we do. I always think of a technological revolution like a horse. The horse can kill you. But it can also lead you wherever you want, if you know how to steer it. So it all depends on the policies and institutions that we put in place. The problem is that with AI, the technology goes much faster than the institutions. Those societies that will know how to set up good institutions and policies will manage.
Do you think the mechanism of creative destruction is still relevant, that newer technologies could come along and disrupt the massive tech companies that we are seeing today?
Philippe Aghion: I think, if anything, there is a speeding up and intensification of creative destruction with the AI revolution. The issue is how we can have institutions that allow us to make the best of it. On the one hand, AI has great growth potential because it automates tasks in the production of goods and services, and it makes new ideas easier to find.
But there are also limiting factors such as lack of human capital. Inappropriate competition policy can also be a problem, if you have predominance of large companies who discourage new entrants who do not share data, who put entry barriers to new entrants. In Europe, we lack computing power. So if we want to harness the growth potential of AI, we also need to have smartly designed industrial policy to invest in computing power.
There is also the employment aspect. There are jobs for which most of the tasks will be replaced by AI, therefore AI will destroy jobs like any previous – maybe even more – than any previous industrial revolution. On the other hand, firms that adopt AI become more productive and competitive with a bigger demand for their products. That leads to job creation.
Also, AI makes new ideas easier to find. New ideas are new jobs, so we need to have labour market policies to smooth out the transition from old to new jobs.
“Manual jobs will resist much better, I think, to the AI revolution. We should enhance them. We should also reward firms that enhance the soft skills – such as taking initiatives and interacting with others – and the polyvalence of their workers.”
Philippe Aghion
If you have a mixture of good competition policy, smartly designed industrial policy for computing power, a good education system and on top of it labour market policies, you can face the AI revolution and make the best of it.
Is there a best and worst outcome for how AI can be used?
Philippe Aghion: The worst outcome is that you don’t adapt labour, but most people become redundant. You have mass unemployment. Another bad outcome is that you let the kind of perverse AI develop, the kind that will pretend to be you, for example. Another nightmare would be that AI decides to trigger wars. You need us, the humans, to be in control and to have the AI that will help us be in control.
The best possible outcome is that AI helps us solve cancer, that many big illnesses will be overcome, that we will live longer and in better health, that it will give us lots of growth, and therefore better lives and improved standards of living.
We can imagine the nightmare scenarios, but I’m kind of an optimist, and I believe that there are policies to avoid those. Maybe it will take key actors, countries and firms to meet regularly, like we did with nuclear energy. We will need a multilateral organisation where major countries and firms meet to make sure we avoid the worst. That will be very important.
Philippe Aghion is participating at this year’s Nobel Week Dialogue on 9 December in Stockholm. The event will be live streamed here on nobelprize.org.
Published November 2026