Jean-Michel Loubès, Toulouse specialist for hallucinations of artificial intelligence, took part in the international summit of the AI in Paris. He underlines the amazing jumps of the AI, but is aware of his risks: transformation of business, loss of knowledge and the need for regulation for an ethical AI. France can play a key role with his experts.
What are the main axes of their research on artificial intelligence?
I am working on mathematical methods for regulating AI algorithms. My goal is twice: On the one hand, understand why these algorithms can create fatal distortions and mistakes that are called hallucinations. On the other hand, the development of instruments to check the compliance with new European regulations and the improvement of their robustness. It is therefore both an exam and research work to design secure and better controlled algorithms.
Some compare AI with the Internet revolution. Do you share this opinion?
Yes, but I would even say that it is a deeper revolution. AI disrupts our daily use with unprecedented speed. If we compare ourselves with the energy volution – the discovery of fire or nuclear energy – we see a big difference: immediate democratization. In contrast to nuclear power that is reserved for a scientific or military elite, AI is accessible to everyone. Today, a teenager can develop an AI -based application in a few days and spread it worldwide. This accessibility changes the situation radically.
Is AI a threat to work or a lever for innovation?
Both. Like every technological revolution, this leads to winners and losers. AI will not replace people, but it will change the professions. Those who can use it as an assistant are more efficient. On the other hand, there is a less common risk of loss of skills: If we delegate too much to AI, we risk forgetting a certain know-how, as is the case with calculator and mathematics, the GPS and the ability to find their , the case was gone on cards. It is important to think about the transfer to avoid this loss of knowledge. If the AI becomes omnipresent in education, there is a risk of attaching certain skills.
The AI environmental impression is increasingly criticized. Can we make it more economical?
It is a big challenge. The AI consumes a lot of energy, especially due to the data centers that accommodate and guide the models. Efforts are made to optimize this infrastructure, with less gourmet algorithms and more ecological data centers. AI itself can help manage the resources better, for example by optimizing energy flows or logistics chains. But pay attention to the “rebound effect”: When a technology becomes more effective, your use explodes and sometimes cancels environmental profits.
Can France and Europe weigh before the American and Chinese giants?
Absolutely. Europe was a pioneer with the first world law on the regulation of AI. Contrary to the popular opinion, this regulation does not slow down any innovation, it stimulates it. Today AI is based on an empirical approach: We are on the market very powerful models without absolutely understanding their functions. Europe wants to structure this approach to ensure reliable and ethical AI. The recent creation of the National Institute for the Assessment and Security of AI in France is an important step to strengthen our position.
What is it for France to become an essential leader in the AI?
Thanks to the quality of his research, France is already an important player. Many French experts occupy leading positions in the world, such as Yann Lecun, one of the pioneers of the deep neural networks of Meta. We now have to use this specialist knowledge with strategic investments, not only in research, but also in education. AI should not be reserved for some specialists; It must be integrated into all disciplines, from human sciences to medicine. After all, we have to strengthen our international partnerships to compete with the USA and China. With European collaborations and targeted funds, we have the means to stay in the race and offer more ethical and safe AI.
You recently participated in the summit for actions on artificial intelligence in Paris. What do you think were the main points that were discussed during this event?
The summit emphasized several crucial aspects of AI. First, the importance of the development of a sustainable and environmentally friendly AI was largely discussed. The need for international cooperation to determine common ethical standards was then underlined. After all, the role of AI in the transformation of industries and their effects on employment were central subjects.
Can France and Europe play an important role in the regulation and development of a responsible AI in the face of American and Chinese giants?
With its ethical values and its tradition of protecting individual rights, Europe has a unique position to promote a responsible AI. European initiatives aim to establish regulatory managers to ensure that the development of AI respects ethical principles and fundamental rights.