In-depth reporting and analytical commentary on artificial intelligence regulation. No legal advice.

Sam Altman says some other European countries should follow French example: op-ed for French newspaper on eve of Paris AI Action Summit

Florian Mueller

Context: Tomorrow (Monday, February 10, 2025), will be the first day (and industry stakeholder day) of the Paris AI Action Summit (event website) hosted by the government of France in partnership with that of India. On the second day, the heads of state and government of many countries will join. A draft statement for the political part of the summit has leaked (February 8, 2025 Transformer article). Critics say it falls short of emphasizing Responsible AI principles and of specific action items.

What’s new: Yesterday (February 8, 2025), French center-left newspaper Le Monde published a (partly paywalled) op-ed by OpenAI CEO Sam Altman. The article describes France as the “nerve center” of AI in Europe and says other European countries should follow its example. It goes on to talk about the innovation gap between Europe on the one hand and the U.S. and China on the other hand, referencing a report by former Italian politician and European Central Bank (ECB) president Mario Draghi, who describes this as a critical challenge for Europe’s future.

Direct impact: The article will certainly appeal to a French audience through its flattery while policy makers in Brussels (the de facto capital of the European Union) and other European countries (EU member states, but also the UK) must ask themselves what they have to learn from France. As this article discusses in multiple sections, there are reasons for which Mr. Altman understandably believes France could be a thought leader and agent of change, but there are also structural issues concerning Europe in general and France in particular that realistically won’t be overcome.

Wider ramifications: Mr. Altman apparently seeks to encourage Europe to focus on innovation (in terms of both making and adopting AI technologies) rather than regulation. The summary above his op-ed refers to “European regulators” who should reflect upon their AI policies. A new legislative term has recently started in the EU, and there is talk of deregulation though no specific actions have been taken. For the time being, various major AI services are unavailable in the EU as providers find the applicable regulations too cumbersome or, in some cases, need more time to bring their services into compliance. At a recent Brussels event, the CEOs of four European technology companies (SAP, ASML, Nokia and Ericsson) raised concerns over European overregulation as one of the causes of the innovation gap, with one of them saying that Europe’s economy “has one foot in the morgue” (January 16, 2025 ip fray article).

(The following comments on OpenAI’s hope that France and the Paris AI Action Summit can bring about change in Europe reflect only the personal opinions of the author.)

Relative to some other European regions, Paris has significant AI-related startup activity. Mistral is the most significant one of all French AI companies.

I live in a small enclave where France is never more than a few hundred meters away. That is why Google Chrome pointed me to Sam Altman’s Le Monde op-ed. In my observation, the realization of Europe’s innovation problem at the highest political levels may indeed be particularly strong in France, but that is a more recent development and there are structural problems due to which France has a lot of homework to do, with some problems being impossible to solve in the foreseeable future.

Florian Mueller
(Founder & Publisher of ai fray)

The governments of France and Germany were less enthusiastic about the EU’s AI Act than some others, but couldn’t prevent its passing into low.

French president Macron, who is a “lame duck” president now without a parliamentary majority, has made AI one of his policy priorities. And no European leader warns as much as Mr. Macron against the threat to Europe’s future that its weakness in innovation represents. Last year he and the above-mentioned Mr. Draghi appeared at the Collège de France, and Mr. Macron said Europe had only two or three years left to manage a turnaround.

For those reasons, it is easy to see why Mr. Altman hopes that a French-led AI summit could make a positive impact, at least in terms of dissuading the EU from shooting itself in the foot through overregulation that only widens the productivity gap.

The EU is aware of its problems, but has not found any promising path forward yet. A recently published Competitiveness Compass (January 28, 2025 ip fray article) contains some significant acknowledgments, but falls far short of what would change the situation.

France wasn’t always skeptical of aggressive tech regulation

It is also a fact that the AI Act and some other pieces of EU regulation were rushed through the bloc’s legislative institutions not least because of a French politician’s aspirations to position himself as a “doer” who could either lead the European Commission (EC) or at least become an Executive Vice President: Thierry Breton. His inability to control his ambition resulted in his ouster. He even took to X (Twitter) to weaken EC president Ursula von der Leyen, who later asked France to replace him, which Mr. Macron did (September 16, 2024 ip fray article).

By now, more and more people realize that overregulation is counterproductive. But until recently, France was actually pushing harder than any other European country for anti-Big-Tech laws. That includes Mr. Macron’s own lead candidate in the 2019 EU Parliament elections. Here’s a tweet by a then-Politico reporter, below which this article will continue:

France and the EU have a history of innovation, but not enough recent innovations

With respect to the French economy in the Digital Era, Mr. Altman’s attempt to flatter France as a cradle of innovation is nothing short of telling. He refers to the Age of Enlightenment (which started in the late 17th century), steamboats, aviation, cinema and nuclear physics. That is yet more “recent2 than what Greece can say, which invented geometry thousands of years ago. But it is objectively reflective of how Europe has fallen behind and how difficult it will be (and how long it will take) to catch up at least in part.

In a recent LinkedIn discussion, an AI developer said that there are actually a number of French developers working on major AI projects. But they are not working for French companies. They have either emigrated (particularly to the United States) or they work for European subsidiaries of U.S. companies.

The single biggest problem is indeed that countries like Germany and France, which lack natural resources, would need to perform much stronger in innovative industries. But there is no realistic hope that the situation will improve in the foreseeable future.

Mr. Macron will go down in history as having identified and warned against various major problems without having solved a single one of them.

The Franco-German sovereign debt crisis looms large

France is in terrible economic shape with no realistic chance of getting back on track in the foreseeable future. Its public deficit amounts to approximately 6% of GDP, and a prime minister appointed by Mr. Macron, Michel Barnier, sought to take measures that would bring down that deficit to “only” 5%, but was voted out of office after a few months because of the country having a solid political majority that prefers ever more debt over any serious reforms.

As a result, the ECB will have to buy up French sovereign debt in contravention of EU law (even though the European Court of Justice (ECJ) in a politically motivated, flagrant miscarriage of justice declared such transactions compatible with an explicit prohibition of lending money to eurozone governments). That will enable the French government to pay its bills for some more time, but hyperinflation will set in sooner or later.

The euro (€) currency went through a major crisis in the early 2010s because of the public debt of Greece (approximately €300 billion at the time) and problems in some other (mostly Southern) countries. Recently, however, the risk premium that lenders demanded from the French government exceeded —even if only temporarily so — that they were seeking from Greece. And with the German economy being in bad shape (with this month’s elections being highly unlikely to get the country back on track), the next sovereign debt crisis in the eurozone could be far more serious than the one triggered by relatively small countries like Greece last time. Germany’s economy is shrinking for the third year in a row, and there is no hope of France experiencing non-negligible growth anytime soon.

Affluent French people are increasingly leaving their country. More and more wealth French citizens relocate to Italy, where they pay lower taxes and find a similar culture in a neighbor country. There have even been media reports of the real estate market in the Milan region having been impacted by strong demand from French citizens (January 11, 2025 BMFTV article).

As a result of France representing a major threat to the € and Germany being unable to provide stability, combined with overall economic weakness, the next eurozone sovereign debt crisis may only be a few years away.

The terrible shape of the European economy, relative to the U.S., was also discussed (with various charts and numbers) in this part of a January 3, 2025 ip fray article. For example, the poorest U.S. state, Mississippi, has a substantially higher per-capita GDP than France or the eurozone average, and is even going to surpass Germany this year.

French math education: too few top-performing students, and too many of the talented ones leave the country

One common misconception is that France is strong in math and offers great math education. While there are apparently some French universities where math is taught at the highest level (quite possibly at a higher one than in Germany), French students are actually much closer to a Third World than East Asian level of math skills.

The Trends in International Mathematics and Science Study (TIMSS) is an international comparative study that measures mathematics and science achievement at the 4th and 8th grades every four years. The participating countries are from multiple continents. The study was carried out for the first time in 1995. France participated with its 8th-graders, but they only ranked 13th in the world. For a country that gave the world famous mathematicians like Descartes, that outcome was a huge disappointment. France didn’t participate for another 20 years, but tried to improve its education system. In 2015, France tried again, fielding its 4th-graders this time as they would have gotten the benefit of the more recent reforms. But France ended up 35th, even worse than 20 years earlier. In the meantime, its students had been overtaken by, those in various other parts of the world, such as various Gulf states.

The most important indicator with a view to digital innovation is the percentage of students that score very high. That is the relevant talent pool for software developers and other engineers. In the 2015 edition of TIMSS, only about 2.5% of all French students reached that performance level. Even in Germany, only twice as many: 5% may be high compared to France, but it is just a tenth of Singapore’s number and not even one-sixth of Japan’s.

There is a clear correlation. Countries with high percentages of top performers in TIMSS tend to have either homogeneous societies or strict criteria-based immigration rules, unlike France. Among any ethnic group, the distribution of performance levels is a Bell curve, but while some countries have top performers where the belly of the curve is, they are outliers in other countries.

The 2023 results don’t readily provide the percentage of top performers, but one can draw inferences from the published performance data. Even in terms of average performance (which, again, is less relevant than the percentage of top performers, yet also meaningful), France is in bad shape. It is ranked behind the following countries and regions: Singapore, Chinese Taipei, South Korea, Hong Kong SAR, Japan, Macao SAR, Lithuania, Türkiye, England, Poland, Ireland, Romania, the Netherlands, Latvia, Norway, the Czech Republic, Sweden, Bulgaria, Finland, Ausralia, Germany, Denmark, Serbia, Flemish Belgium, Hungary, Portugal, the United States, Cyprus, Slovakia, Slovenia, Italy, Armenia, Albania, Canada, Spain, the United Arab Emirates, Georgia, Azerbaijan, New Zealand, the French-speaking part of Belgium, and Kazakhstan. Only a few countries and regions such as Moocco and Kosovo are behind France.

An Excel table with official TIMSS performance data for 4th-graders shows that the score of the best-performing 5% of French students is 604 or higher:

  • 604 is less than the average across 100% of Singaporean students. In Singapore, it takes a score of 565 or higher not to be in the bottom quarter, and a score of 743 or higher to be in the top 5%.
  • In Ireland, a score of 604 means that one outperforms 75% of all students, but underperforms 25%.
  • In the U.S., the 95th percentile is 667 points, despite many Europeans falsely believing that their school system beats the American one. It does not when it comes to math, and it is also weaker in some other areas. There are school subjects that get more attention in Europe, such as foreign languages, but native speakers of English don’t have a pressing need to learn other languages unless they want to live abroad.

France has structural problems that no politician will be able to solve. But it still has a significant number of very smart people, and France should think of how to incentivize them to stay and how to make them more productive.

The Paris AI Action Summit has an important agenda. It’s about strengthening AI in Europe and strengthening Europe in AI. Mr. Altman’s guest piece, though some of it is unfounded flattery for diplomatic reasons, may serve as food for thought for the leaders of other European countries and especially for policy makers in the EU institutions.