Meet the ‘Lady Gaga of Mathematics’ helming France’s AI task force

0
69
- Advertisement -

A spider brooch-wearing mathematician

On a crisp Saturday morning in Orsay, a southwestern suburb of Paris with some 16,500 inhabitants, the rue de Paris was bustling. But while many residents were doing their usual weekend shopping at the fishmonger or the butcher shop, further up the street, in a small former chateau that is now the town’s cultural center, about 80 people had set aside their late-morning hours to hear the “voeux” of their legislative representative to the National Assembly, Cédric Villani.

The voeux, or “new year’s wishes,” are a standard exercise of French politicians from the president on down, in which they review activities of the past year and lay out projects for the year to come. Villani, a mathematician and Fields Medal winner (often shorthanded as the equivalent of the Nobel Prize in mathematics), was new to the practice; only six months earlier, he was still an academic. He was dressed as always — winter or summer — in a black three-piece suit, a shirt with cufflinks, a spider brooch on his lapel, and a large, floppy tie called a lavallière (today’s version in purple). He cut an unmistakable figure, sporting a three-day beard, his dark hair styled in a pageboy. He mingled, smiling with attendees, and posed for selfies before taking the stage.

The fact that a mathematician could be considered, as he is, a “rock star” — or, better yet, “the Lady Gaga of mathematics” — says perhaps more about the French than Villani. Nonetheless, Villani, 44, has become a darling of President Emmanuel Macron’s young technocratic government, accompanying the president to Ouagadougou, Burkina Faso, in November and Beijing in mid-January. The government has piled the work on his desk, which is evidence, Villani says, of the need for people with scientific expertise in politics. But of all his projects — from math education to the future of New Caledonia to tax evasion — perhaps his most all-consuming mission is his task force on artificial intelligence and the highly anticipated report it’s set to release tomorrow. If successful, the report will help set the AI agenda in France and Europe for years to come.

In view of a world where “artificial intelligence will be everywhere, like electricity,” as Villani has said, becoming a leader in the field is critical for France. Many feel that Europe is already at an enormous disadvantage compared to the US and China and will need to do some Usain Bolt-style sprinting to catch up. For one thing, France and Europe don’t have the data-gathering platforms necessary to fuel machine learning: they lack the power of what the acronym-loving French call GAFA (Google, Apple, Facebook, and Amazon). French bureaucracy has also historically been a drag on entrepreneurship and invention. Compared to the US, cooperation between academia and industry is much less frequent. And though France is known for the quality of its engineers and scientists, much of the top-level talent goes abroad, where there is more money and freedom to pursue research without constraints. Addressing these issues by sketching the nation’s AI road map has fallen on the well-tailored shoulders of Villani.


Several months before delivering his new year’s wishes, after a local TEDx presentation in November on “How AI Will Revolutionize Health,” I sat down with Villani, who was cordial if a bit distant. After winning the Fields Medal, Villani, a self-described “formerly shy” person, took a media training workshop. His large eyes, luminous skin, thin body, and slightly walleyed expression accentuated the impression of speaking to an “extraterrestrial,” as Paris Match once put it.

The report Villani is set to release isn’t a first for France. At the very end of François Hollande’s presidency last spring, his administration released a rushed AI report, offering broad brushstrokes. But, Villani says, his report should offer both “a panorama … and make a diagnosis of the subject…. It must pose questions explicitly and offer practical solutions and implementations.”

Villani’s six-member task force (@MissionVillani) is made up of a machine learning researcher, an engineer with the defense ministry, and four members of a French digital technology advisory council, with expertise in everything from philosophy to law. The group was charged with a broad-ranging mission, covering industrial, data policy, employment and training, environmental, ethical, and research issues. Contrary to the reports issued by the Obama administration (which one international observer notes “have not led to a single bit of U.S. policy”), Villani’s team expects that some concrete measures will be put into play within months. Focusing on four key sectors — health, transportation, environment, and defense — the team also emphasizes that it has devoted substantial attention to the ethics of data policy within the context of the General Data Protection Regulation (GDPR), which is slated to go into effect in May and will broaden privacy protections for individuals in Europe. Considering the current Facebook meltdown over the Cambridge Analytica scandal, the French team has probably chosen wisely. As for the question of financing, while France’s previous AI report suggested a 1.5 billion-euro investment in AI, it is unclear precisely how much funding will be allotted.

Given the breadth and complexity of the subject of artificial intelligence, a skeptic may doubt one man’s ability to understand the field without years of study, but there is probably no other member of Parliament better suited to lead this project on the subject. By all accounts a quick study with an enormous work capacity and a diplomatic, optimistic temperament, Villani was, of course, also a high-level researcher in a related field, which helped him grasp AI concepts quickly.

Asked if he had an interest in artificial intelligence before he was assigned to the task force, Villani enunciated clearly and deliberately in his high-pitched, slightly theatrical French: “The subject had grown so significantly that you would have to have been blind and deaf not to be interested.” In any case, “the big concerns are not really about the most technical issues.” He has indicated in the past that he hopes to avoid AI’s potentially “devastating effects on economic issues and the democratic fabric,” partly by making sure that AI is “everybody’s business.” Hence his large-scale offensive in the French press to educate the public and his push to seek broad-ranging input for his report.

In November, Villani estimated that he would speak to 250 people for the AI report and finish it by the end of January. But, solicited by hundreds of people who wanted their say, he kept speaking to more and more parties. (“He always says yes,” said one of his team members.) Among these were scientists, lawyers, doctors, philosophers, labor union representatives, business leaders, startup entrepreneurs, and even one roundtable of 15 girls, who discussed the involvement of girls in the sciences. In the end, the task force interviewed about 350 people in groups of 10 to 15, gathered according to topic, in an off-white conference room at the Digital Ministry. There were also 1,600 contributors on a public online platform.

Now, the report is slated to be officially delivered to the president at a March 29th ceremony at Paris’ Collège de France, which will be attended by an estimated 500 guests, with the participation of tech-celebrity guests like Facebook’s chief AI scientist and deep-learning guru Yann LeCun.

AI experts in France and abroad are anxious to see the report. “Governments around the world are struggling with whether they have to do something preemptively about AI,” said Joshua Gans, professor of strategic management at the University of Toronto and co-author of the forthcoming Prediction Machines: The Simple Economics of Artificial Intelligence. “There have been a lot of concerns about potential issues, from safety to jobs to what are its privacy implications.”

If he had to sum up international government reaction to AI, Gans said, “it is a great dropping of the ball … not so much on the research side but on ensuring that privacy laws are up to date and moving toward international agreements … on autonomous weapons.” The fact that France has put a “very high-profile person” in charge of the task force attests to how seriously it is taking these issues, he said.

Jean-Gabriel Ganascia, a professor of computer science at the Sorbonne and author of Le mythe de la singularité (“The Myth of the Singularity”), was one of the people interviewed by Villani and his team. “We’re waiting for the report and then its translation into concrete actions,” Ganascia said. “The important thing is not to write reports and create strategies. We have to act.”

And there is not a moment to waste; some maintain that it is already too late for France and Europe. “A country that doesn’t have an AI industry will be underdeveloped tomorrow. It will be a slow process of technical, political and military colonization,” said provocateur Laurent Alexandre, author of La guerre des intelligences (“The Intelligence War”), in an interview with the French business daily Les Echos. Noting the dominance of American and Chinese tech giants, he said, “Europe has completely lost the AI battle.”


Roxanne Varza, the 33-year-old American director of Paris’ new Station F, which bills itself as the world’s biggest startup campus, let out a melodious laugh when I mentioned Alexandre’s grim views. “That is such a passé view,” she said, sitting in a glass-walled room overlooking her monster tech playground. “Maybe you could have said that five years ago, but I don’t think you can say that anymore. With the Brexit climate, Donald Trump, high Silicon Valley prices, the political situation worldwide, and with Macron now in power in France, I think we’ve seen a huge shift.”

In fact, Varza said, the top countries that apply for programs at Station F are the US and the UK, respectively. “France has a huge opportunity in [the AI] space because French engineers and data scientists are so well known.” And feared American tech giants Facebook and Microsoft are right on the Station F campus, apparently eager to tap into some of that je ne sais quoi in Macron’s “Start-up Nation.” (Add to this the announcement made earlier this year about Facebook’s 10 million-euro investment in France for AI research and a Google AI research center in Paris.)

For his part, Villani doesn’t want an AI war; he wants competition, yes, but also collaboration, such as one he set up with Microsoft when he was director of the Institut Henri Poincaré, a prestigious math institute. At an “Ask Me Anything” event at Station F, he said, “We need more infrastructure, we need a European cloud, we need more European intensive computing centers, we need a European hardware industry, we need more European research centers — and that will take time and money. But it’s worth it because that will [help ensure] European sovereignty. And it should be constructed not in an ambiance of war but in the spirit of competition.”


Marc Schoenauer, 60, Villani’s guide in the AI netherworld, arrived at a dive bar south of Paris on a motorcycle, his long, wavy gray hair and beard askew, wearing a baggy sweater and jeans. If Villani’s attire and studied manner evoke a 19th century dandy with Goth overtones, Schoenauer gives off a former hippie vibe. A researcher at the French Institute for Research in Computer Science and Automation (Inria), Schoenauer has spent 30 years studying artificial intelligence and is the AI expert on Villani’s task force. He accepted the invitation, he said, “naively,” ignorant of the long hours and the months the report would go into overtime.

During the task force’s three-hour hearings, Schoenauer said, interviewees would, one by one, make a statement of their recommendations, which would be followed by a discussion, after which Villani would circle back to issues that interested him. Each task force member was responsible for a chapter of the approximately 200-page document, which would then be reread by their colleagues. In the final phase, a draft was distributed to French ministries for feedback on the feasibility of the recommendations and the necessary funding.

On working with Villani, Schoenauer commented, “He’s very impressive, first of all, because of the amount of work he can do.” He also noted that Villani had to head or contribute to other task forces and is a regular member of Parliament. Furthermore, Schoenauer said, “He remembers everything that was said in the hearings and fills up small notebooks. Later, he knows exactly which notebook he put the note in.”

“I think his motivation is pure in politics. It’s not at all to put himself forward,” said his task force collaborator Yann Bonnet, general secretary of the French digital advisory commission Conseil National du Numérique. “He’s a very good politician. … He always listened carefully, found compromises, was able to understand the balance of power. … It made it so everyone was happy to have contributed to the task force.”

“I’ve never seen him angry,” said another task force member, Anne-Charlotte Cornut. “I never heard him raise his voice in the six months we worked together. And we worked 24 hours a day.”

Schoenauer allowed that perhaps Villani enjoys his public profile and persuading others to his point of view. However, contrary to the “shocking” attitude of some cabinet members and politicians who “don’t give a shit,” he said of Villani, “I think he’s motivated to do something useful for France and mankind in general.”

Interviewed this month in a private room at Le Bourbon, a brasserie favored by the French political elite, Villani said he was motivated to enter politics because of what he saw as the recent “chaos” in French politics. And Macron’s pro-European, “neither right-wing nor left-wing” stance suited Villani’s own long-held political views.

“There was a need to serve the nation in a moment of great confusion, and [there was also] the idea that this was a chance that shouldn’t be missed.” He continued, sometimes spasmodically tapping the table for emphasis, “During the second round of voting [during the presidential election], it was possible that it was going to be a choice between the extreme left or the extreme right. It was chaos!”

Villani admits that, as a new representative, he had to learn “everything” about political life. “You learn the way laws are made, how political influence works, relations between [governmental] groups, cultural questions, work on constitutional reform,” he enumerated.

When it was suggested that some mathematicians might not understand the abandonment of research by one of its highest practitioners, he was slightly defensive. “Every time scientists have the feeling that there’s a subject that mixes science and politics — baf! — they come to see me, or they write me.” He said he had spoken to a half-dozen AI researchers just that morning who urgently wanted to explain their point of view. “I see in all these examples how important it is to have scientists in politics. It’s important for politics. It’s important for science.”

As for the long-awaited AI report, Villani is satisfied. “I think we have the diagnosis, I think we have the recommendations, and I think we’ve listened to enough people to be fairly sure of our recommendations,” he said, noting that not only had his team formally interviewed hundreds of experts, but his constant presence at conferences large and small, as well as his blanket coverage in the mainstream media have elicited feedback — good and not so good — from government and industry observers.

“When you do an interview in a mass circulation publication, if there’s something wrong in what you say, you can be sure that people will tell you so!” Villani remarked.

He paused to take a sip of orange juice. His pale skin seemed a bit gray, having lost the translucent glow observed during a previous meeting. While his colleague Marc Schoenauer was looking forward to returning to his research after the delivery of the AI report (“I learned a lot … but that’s it now”), Villani’s talents will be required for its implementation.

“It’s not the end, but it’s a step we’ve taken,” Villani said of the report’s completion. “We’re going to the next stage of the mission.” A waiter informed him he had a group of people waiting in the wings to meet with him. The deputé seemed game, if a little tired. He was still on the nation’s clock.

Written by Sono Motoyama
This news first appeared on https://www.theverge.com/2018/3/28/17170104/cedric-villani-french-mathematician-ai-report-interview under the title “Meet the ‘Lady Gaga of Mathematics’ helming France’s AI task force”. Bolchha Nepal is not responsible or affiliated towards the opinion expressed in this news article.