Former top manager Anne Bouverot advises French President Emmanuel Macron on artificial intelligence (AI). She is a co-initiator an open letter that advocates more basic AI research in Europe.
THE TIME: Madame Bouverot, let’s start with a fundamental question. They are committed to that Europe is catching up with China and the USA in the race for artificial intelligence. Do we absolutely have to?
Anne Bouverot: Yes. Because it’s not about the race itself. Artificial intelligence is a transformative technology from both an economic and social perspective. If we only rely on Chinese or US technologies, we will be a continent of consumers. We need our own economic sectors, our own companies. Otherwise we will be dealing with chatbots propagating their vision of the world.
TIME: What do you mean by that?
Bouverot: If you ask a Chinese AI model a political question, or even a US model, the answers often do not correspond to the European way of thinking. AI chatbots are concentrated expressions of cultures.
TIME: Are you worried that we will become as dependent on the US for AI as we currently are for operating systems, search engines or social media?
Bouverot: Absolutely. If we don’t do anything, it will artificial intelligence lead to a large concentration of power. When it came to social media, we first thought: Oh, that’s nice, that’s cute, it won’t really matter. But they have a huge impact on how we interact with each other. And even our elections.
TIME: And what does this mean for AI now? What do we have to do differently?
Bouverot: Europe needs to rethink its position towards the USA. We need more digital independence. This impression has been reinforced since the Munich Security Conference…
TIME: … at which US Vice President JD Vance accused Europe of a loss of democracy …
Bouverot: … felt stronger every week. Unfortunately. Or maybe fortunately. Because it is a wake-up call for Europe to finally do something.
TIME: Can Europe even catch up in the AI race?
Bouverot: I believe that very firmly. In January, the US announced $500 billion for its AI offensive called Stargate. The money will flow into infrastructure in the USA. That was a clear message. Just a few days later, DeepSeek was unveiled, the Chinese AI model developed for just $6 million, a fraction of the money invested in models how GPT and others flowed. This shows that the race is not over.
TIME: What is needed to catch up?
Bouverot: A lot of talent lives in Europe. In Germany, France and other countries we train engineers, mathematicians and computer scientists. We also have the computing capacity needed to train base models for artificial intelligence. Not the largest capacities in the world, but considerable. The Jupiter supercomputer in Germany, for example. We are planning gigafactories, i.e. gigantic AI data centers. And we are sitting on data that is not available to everyone in the world. Data from hospitals, companies, industries. This data is very valuable. All of these factors, combined with agility and urgency, can secure us a place at the forefront of the AI race.
TIME: More resources for AI pioneering research from Europe – This is what leading scientists like Yoshua Bengio and Yann LeCun demand in an open letter that you helped initiate. What do you want to achieve?
Bouverot: If we want to catch up, we can’t do exactly the same thing as the USA and China. The two countries have pioneered the development of language models; they have an almost unassailable lead in this area. We need to explore other areas.
TIME: So neither the USA nor China copy?
Bouverot: Exactly. We have to go beyond that. So far it’s like this: the architecture of artificial intelligence consists of transformer models…
TIME: … i.e. models that are based on artificial neural networks and, to a certain extent, learn themselves. In its current form, this was only made possible by the Transformer architecture developed in 2017.
Bouverot: They are fed texts and improve themselves. But it’s not just language. One could also think about how to use AI to generate voices, such as using them in robotics. And it is also worth thinking about new architectures.
TIME: Europe already has a good reputation in research. What is missing are large companies. Don’t we need to think more about how to translate research into business models?
Bouverot: We are seeing initial success at the start-up level. We have AI companies like Mistral in France or Black Forest Labs in Germany. But if we really want to build our own large companies, we need more research capacity. We have to think even further into the future. The aim of the open letter is to get more resources for research. We can’t do everything at once.
