Macron and AI: towards the end of the French paradox?

Macron and AI towards the end of the French paradox

In the bustle of the VivaTech show in Paris, the most important show dedicated to technology in France, a face is of particular interest to image hunters. That of Arthur Mensch. Passed by Polytechnique, the Ecole Normale Supérieure (ENS), the young man, just 30 years old, comes out of two years at Deepmind, the Google lab specializing in artificial intelligence (AI). With two acolytes with comparable CVs, he has just launched MistralAI, a competitor to the American OpenAI in the creation of a powerful language model (LLM). A tool necessary for the development of ChatGPT type applications. Tuesday, June 13, the day before the opening of the show, the announcement of a massive fundraising – 105 million euros – has already brought the valuation of MistralAI to more than 200 million euros.

Nothing surprising if the face of Arthur Mensch, discreet in the spans, finally appears in full light on the big stage of VivaTech alongside Emmanuel Macron. The President of the Republic makes it the mirror of his own measures on AI. The young man studied in France? Some 500 million euros will make it possible to develop “poles of excellence”, in order, in particular, to double the number of training courses and continue to supply the country with talents like him. Mistral will require massive computing power for its future language models? Macron raises 50 million euros to quadruple the capacities of the supercomputer Jean Zay, father of the largest French-speaking LLM to date (Bloom). Mistral has carried out a massive fundraising campaign, but its financing could prove to be more perilous in the future? The executive draws Tibi2, a device endowed with 7 billion euros to supply investment funds, as well as a complete overhaul of tax systems for investment in innovative companies, according to the report delivered by MP Paul Midy.

The HuggingFace case

Mistral offers, beyond a great success, the ideal opportunity for Emmanuel Macron to resolve a very French paradox. Yann Le Cun, Luc Julia, François Chollet, Antoine Bordes, Thomas Wolf… A large number of French experts in AI, recognized worldwide, have so far almost exclusively filled the ranks of Gafam or companies under foreign flags, such as HuggingFace, a nugget valued at two billion dollars at the New York headquarters. Alas, without creating French champions able to compete with the major American and Chinese companies that dominate the field. MistralAI now crystallizes this passage strewn with pitfalls between basic research and its transformation into a tech giant on an international scale. A real tour de force. Well, if all goes as planned.

Mistral, it must be remembered, has not released any product. And probably won’t show much to the public until 2024. A world for AI, which is advancing at full speed. Nothing indicates, moreover, that Mistral will be a real competitor to OpenAI, whose means are even more considerable, thanks to the patronage of Microsoft, which sprinkles it with billions of dollars. “We have no chance of competing with the United States in this area. We can do so, on the other hand, on other strategic axes, in particular that of AI safety”, recently judged Milo Rignell, resident expert on new technologies at the Institut Montaigne interviewed by L’Express. An orientation also called for by Emmanuel Macron, Wednesday June 14, who used the words “ethics”, “open-source”, denouncing in passing the racist or sexist “biases” carried by certain AIs. The President of the Republic is betting more than ever on this breath of fresh air. May he one day, even, blow away his rivals.

lep-general-02