“This revolution is political but must not be partisan”, for mathematician Cédric Villani

L'ancien député Cédric Villani a été président du rapport parlementaire sur l'IA. AFP – JOEL SAGET

Renowned mathematician and president of the parliamentary report on artificial intelligence, Cédric Villani is one of the experts at the first local citizens' convention on AI in Montpellier.

What do you think of the initiative of the Metropolis which launched this first citizens' convention ?

On AI, there are always questions of sovereignty that arise – in relation to the prerogatives of this or that institution, the way of maintaining control of one's destiny – and including it in an operation carried out by the metropolitan area is very interesting because it allows for citizen reflection. On a political level, it is an initiative that concerns those responsible for digital strategy but also all elected officials in the executive branch. Thanks to the work of this citizens' convention, the Metropolis will be better prepared to face the future around artificial intelligence.

Are the fears surrounding this new technology legitimate or are they psychosis ?

What we know is that this technology will be omnipresent in our daily lives. It has been there for years. It is above all the appearance of the Chat GPT conversational robot and its partial use by ordinary mortals that is intriguing. Now, the question is not whether you are for or against AI, that makes no sense. It's more about knowing how to prioritize its use, what impact it will have on citizens, what system of governance and how politicians will take up the subject.

Are these fears the result of a lack of precise definition of AI ?

Let's not look for a definition of artificial intelligence, there is none. Quite simply because it is a vague, varied set of techniques and technologies that make it possible to carry out tasks that we thought were reserved for humans. But there are many things that fall into this definition and which are not yet called artificial intelligence. This was implemented by intelligent people, for sure, clever people, who found good techniques but it's just not intelligence.
We might have believed that to write a good letter of candidacy for the senatorial elections you had to be intelligent, have knowledge of a territory, notable political experience, but no: ChatGPT writes you an application for the senatorial elections, to the electors, to which there is almost nothing to change for the letter to be impeccable and yet ChatGPT has no intelligence. So let's not look for a definition at all costs.

How are politicians tackling the subject today?

The scientific parliamentary office that I had the honor of chairing at the National Assembly is the perfect example of how to take part politically in the development of this new technology. It was necessary to reconcile scientific, political and technical opinions in contradictory hearings. And ultimately, deliver what looks more like a trajectory, with proposals for controlled sovereignty, than a guide. At the European level, Parliament is proposing, as part of its digital strategy, several texts to regulate artificial intelligence and guarantee better conditions for the development and use of this innovative technology.

Is this policy ambitious enough in France ?

The Court of Auditors recently assessed the impact of the national strategy on artificial intelligence, as it had been decided by the government on the basis of the report that I was conducting. The conclusion is that the implementation of the various policies had not allowed France to really climb into the leading group of countries in artificial intelligence but that this had clearly avoided downgrading. This report concluded that France remained in the top 10 of artificial intelligence nations and that it would have been much worse if we had not taken it in hand a few years ago.

An ethical, sovereign and responsible artificial intelligence is therefore a pure utopia ?

Obviously it is not Montpellier which will dictate its law with a simple citizens' convention. But it must be remembered that we have skills recognized at the highest level in the region. I had the chance to visit the Adastra supercomputer at the national scientific research center and it is part of the infrastructure that makes our country more sovereign with regard to artificial intelligence.

Why is it important that citizens enter into this debate ?

When we ask the precise question of how a territory will address a technical problem, there are several ways to answer.

The first is to ask yourself what the position of elected officials is. Elected officials will respond to what they think in their soul and conscience, the party's position. Sometimes it's the instructions, sometimes it's tradition but they are obviously held by certain filters.

We can also ask the experts. Experts come with their own culture and sometimes their biases. There has been some fascinating work in decision theory that has shown that sometimes experts do worse than average at predicting what will happen in their own disciplines.

You can ask the associations now, but the associations were also founded with a goal, to defend freedom in this or that.

And then we can ask the most representative structure there is: a population elected by chance. Our constitution tells us so; democracy and our republic are supposed to act for the people, by the people, on behalf of the people. And this citizens' convention acts in this direction and presents this ideal as cleanly as possible.

I subscribe to read more