“The only good news on the economic landscape is AI.”

Fabian Kurt Millet is Google's chief economist. The French-Spanish expert oversees a division responsible for a wide range of analysis, from global macroeconomic fluctuations to changes in the regulatory environment, to help the tech giant make the best decisions. Unsurprisingly, he's now obsessed with artificial intelligence and its impact on the economy “from morning until afternoon.”

LE FIGARO — Are global macroeconomic tensions complicating the deployment of artificial intelligence?

passer la publicity

Fabien Kurt Millet — We are living through a complex period in economic history. A few weeks ago, I was in Washington attending a meeting of the World Bank and IMF. The main takeaway is that while economies are resilient in the short term, that resilience can mask long-term vulnerabilities. Everyone in the economic world is feeling nervous. But the only good news on the economic front is AI. This is the most profound technological revolution I have ever seen in my professional career. This is what we call “universal technologies,” a very select group of innovations that fundamentally change the economy. Think about steam engines, electrification, and personal computing. AI will revolutionize science, research, and development.

As China and the US invest heavily, what can Europe do to avoid being left behind?
Competition between the United States and China is clearly advancing. Still, Europe has some aces. We boast a pool of AI talent with excellent training and research programs. According to the World Intellectual Property Organization, France ranks 13th in the world in innovation, behind Japan and Germany. Europe also has a strong industrial base.

Unfortunately, however, there are headwinds that Europe has created for itself, such as its regulatory zeal. The Draghi report made this clear. Europe has built a difficult regulatory jungle to navigate, with around 100 instruments adopted in recent years and implemented by 260 different regulatory authorities. Two out of three European companies cite regulation as a barrier to investment.

What do you think about concerns about AI bubbles?
We are seeing similar symptoms to the dot-com bubble of the 1990s. Inflated market valuations and cyclical funding structures within the sector. But the big difference is that in the dot-com era, cables were run underground even though there was no real demand for them. Now, every time a new GPU is added to the cloud, its computing power is immediately put to use, increasing the demand for AI. This is a major difference that is not emphasized enough. Will this level of demand continue? That's another question. What is clear is that AI is a technology that is here to stay, just as the internet was here after the bubble burst in 2000.

Research points to the strong economic potential of AI. However, according to MIT, 95% of companies do not generate profits from using AI. How do you explain that contradiction?
The MIT study is highly controversial. The best research in microeconometrics demonstrates the significant productivity gains that AI brings across a variety of fields. Many studies exist. What is unclear is when these micro-level gains will turn into macro-level gains. Predictions are even more uncertain. In France, we worked with the Concorde Foundation to estimate a 9% increase in GDP over 10 years. Most recently, a new report from the Implement Consulting Group, developed in collaboration with economist Antonin Bergaud (HEC) and Google, found that accelerating AI innovation could bring France €90 billion in annual economic benefits.

passer la publicity

Is there a risk that executives see AI as a magic wand?
absolutely. AI won't magically transform your company. We still need human intelligence. Deep change takes time. Let's take the industrial revolution as an example. The real productivity leap occurred when electrification forced factories to rethink all production processes. Simply swapping one technology for another only gives you a one-time gain. Systemic benefits are needed, which requires a redesign of the entire workflow. It doesn't happen overnight.

Are small businesses at risk of missing the AI ​​wave?
That's a big concern. According to the McKinsey Global Institute, all sectors are benefiting from digitalization, but adoption rates vary widely. Most sectors were only around 15% of the level achieved by the most digitalized industries. And if we're not careful, the same pattern will happen with AI. We need to ensure that these technologies reach small and medium-sized enterprises and small businesses. That could include training programs built through tax credits or public-private partnerships.

What does AI mean for the labor market?
We should not deceive ourselves, but neither should we fall into sterile pessimism. The labor market is resilient. It absorbed industrialization, women's social advancement, and the baby boomer generation. Computing has cost the United States 3.5 million jobs but created 19 million new ones. Still, these changes have a profound impact on individuals.

In the 1920s, the telecommunications company AT&T was one of the largest employers of women in the United States, but automated telephone exchanges had a major impact on the operators' jobs. What happened next? At the macro level, women's employment remains high. But when we zoom in on these former business owners, we find that many have seen their incomes decline or have left the workforce. That's why we need policies that support this transition so that everyone can share in the prosperity that new technologies bring.

Will AI-related layoffs fuel public distrust in technology?
These announcements should be treated with caution. It's easier to blame AI for layoffs than admit a strategic mistake. The truth is that AI adoption in enterprises is still in its infancy. Technology also creates enthusiasm. For example, in the medical field, caregivers can feel relieved by offloading administrative tasks to AI, allowing them to spend more time with patients.

passer la publicity

AI can take over tasks typically handled by junior employees. What does this mean for young workers?
In the United States, the number of job openings began to decline in April 2022, six months before ChatGPT. This decline was caused by rising interest rates. Employment continues to be sluggish, and young people are the first to feel the effects. But it's not the AI's fault. That doesn't mean it won't be a problem in the future. We must continue to be extremely vigilant.

What should the government do?
First, we need to deploy AI at scale. Because this is how companies can grow and protect jobs in a globally competitive environment. And managing labor market transitions requires a set of training policies built with employers to ensure that training matches the real needs of companies.