In an edited extract from an interview with LSE, Fabien Curto Millet, the chief economist at Google, explains his excitement about AI, voiced his scepticism that technology is harming labour markets and offers advice to young people entering the AI-economy. Fabien was talking to Maayan Arad.
How will artificial intelligence impact jobs?
Artificial Intelligence (AI) is the most profoundly transformative technology that I’ve had the privilege to witness in my professional career as an economist. AI has the potential for a massive productivity uplift. And it’s coming not a moment too soon given the demographic and fiscal challenges societies are facing.
The evidence is tantalising. There is a growing documentation of the transformative productivity impact that AI has. Eric Brynjolfsson at Stanford looked at call centres and found that AI resolved 14 per cent more issues per hour. And my colleagues at Google found 21 per cent time savings in developer environments such as coding. Another study looked at the impact of AI across the Danish economy and found an average of 2.8 per cent time savings for workers.
It’s not just productivity. It’s also about benefits in terms of the quality of a work environment and the work itself. There’s higher creativity and higher quality of output in work.
Do you think jobs today are getting better?
I spend a lot of time talking to people deploying AI across different sectors and the resounding message is that AI removes drudgery and stimulates creativity. It is also empowering. The doctor using AI to speed up admin and the nurse using AI to do the work of doctors both tell us that.
AI also accelerates scientific research. AlphaFold, released by Google’s DeepMind, was used to undertake an experiment which won the Nobel Prize for Chemistry in 2024. That research could save hundreds of thousands of dollars and potentially years in the lab.
How is AI affecting career paths?
AI has only been around for a few years, and careers take decades. So we don’t yet have the best stories. But I have no doubt that they will come.
David Autor’s paper, New Frontiers, found that 60 per cent of employment in America today is in roles that did not exist 80 years ago. That’s what technology does to economies when you allow enough time to pass. Career paths of people who are around now will include things that we can hardly anticipate. When the internet came around we could not imagine that there would be people employed as influencers and creators. That’s new work.
You can also expect to see new forms of organisations such as “micro multinationals”, which are small and medium-sized enterprises that are global from day one. So you may have fewer than ten people. And yet your management is in France, your engineering in Ukraine and your financing in Silicon Valley. Digitalisation made that possible. AI is going to remove frictions and allow more of those small type of “lean” enterprises to emerge.
What is AI doing to labour markets?
The current labour market is not the most welcoming to enter. But from my perspective, that is largely macro driven as opposed to AI driven.
The best paper on this, topic (even though I disagree radically with its conclusions) is Eric Brynjolfsson’s Canaries in the Coal Mine?, which uses data from millions of American workers. And the headline finding of the paper is that there is a 13 per cent relative decline in employment for early career workers in highly AI-exposed occupations. And it finds that the timing reflects the launch of ChatGPT in November 2022. And from that point onwards, the fate of these young workers in highly exposed industries starts diverging from the rest of the economy.
But to me, that interpretation is not correct. They are looking at employment numbers. Whereas if you look at job postings, they started to shift six months before ChatGPT comes out. And what was happening in April 2022? The Federal Reserve tried to control inflation by hiking interest rates. Which was a massive macro event. And we think that highly AI-exposed occupations are also more sensitive to interest rates. So what we’re picking up in this paper is interest rate sensitivity. So my read of the literature is that AI is not the culprit of the current situation.
What do people get wrong about when they talk about AI and jobs?
Technology impacts tasks rather than jobs. And on the whole, technology is a net creator of jobs. The way that technology works from an economic perspective is it automates individual tasks. And a job is really a collection of tasks. And so a job only gets fully automated away when a significant proportion of the underlying tasks are automated, which is a fairly rare event. James Bessen’s study, published in 2016, looked at 270 occupations and found that only one – elevator operator – had been automated away.
People are also under appreciating how transformative AI is going to be for blue collar occupations. In America we have a shortage of 100,000 electricians. And a lot of that is driven by the AI boom and investment in data centres requiring those traditional trades. All occupations will be modified by AI.
A radical example is Open Network Stacks, which uses AI to allow people with no formal education to interact with bodies of knowledge. A farmer in Uttar Pradesh can ask AI what to plant or what price to sell at. So even in an occupation that couldn’t get more manual, AI will have an impact. There are many reasons to think that this transformation is economy wide.
What advice do you have for young people entering the labour market?
Young people must make sure they are using AI to their advantage. If you want to have the optimistic case for young people look to David Deming, a labour economist at Harvard, who points out that young people are typically the ones who make best use of technology. And this is what we have seen in previous waves where technology disrupts jobs.
It is the young people coming into the organisation are the ones who will best be able to use AI. Whenever I speak to younger people who are in the field, despite thinking that I am pretty clued up on the AI, I always learn from the young. Let’s not be too pessimistic about the prospects for young people.
Become masters of AI, but to use it to your advantage you must also develop judgment so that you are leaning into those human qualities that are irreplaceable.
Joshua Gans wrote that “AI is an advance in the statistics of prediction.” And prediction is an important element in decision making. But decision making also requires judgment, which is a very human element. So what I tell young people is become masters of AI, but to use it to your advantage you must also develop judgment so that you are leaning into those human qualities that are irreplaceable.
I’m as excited as I can be about this transformation ahead of us. It is going to require adaptation by individuals because it’s happening within our lifetimes. And all of us are going to have to tool up. It’s going to happen for all of us one way or another. Google wants to be a responsible, good partner to governments, sectors across the economy and people. We’ll figure this out together.
This article gives the views of the author, not the position of LSE Business Review or the London School of Economics. You are agreeing with our comment policy when you leave a comment.
Image credit: Andrii Yalanskyi provided by Shutterstock.
