The site’s response to the AI Cluster call for expressions of interest was submitted on Thursday November 9. The project brings together 16 local institutions and organizations, as well as a number of companies. Interview with Aurélien Garivier, professor at the École normale supérieure de Lyon, member of Pure and Applied Mathematics Unit (UMPA), associated with Parallel Computation Laboratory (LIP) and project coordinator.
Q: Why AILyS?Aurélien Garivier : This is the 2nd wave for AI (artificial intelligence) projects. We hadn’t responded in 2018, but since then, our forces on the site have developed and networked, so we were ready to commit to a more structuring project for AI at the Lyon Saint-Étienne site level. We are already one of France’s leading players, with 230 researchers working mainly on AI, 15 ERC grants, 8 members of the Institut universitaire de France (French university Institute - IUF), over 200 theses and almost 500 publications. Our ambition is to be one of Europe’s leading AI sites in the future. If we get it, the AI Cluster label will be a definite asset.
Beyond this ambition for the site, the AILyS project is an additional way of addressing the many challenges and opportunities posed by the development of AI, in terms of research, training and innovation, for an AI that is useful to all of society.
Q: What are the particular challenges facing AILyS?Aurélien Garivier : While embracing all aspects of artificial intelligence, the AILyS project has chosen to focus on two particular challenges: developing a responsible and inclusive AI, and working specifically on the issue of frugality.
In first place, it is imperative to anchor the development of AI innovations on solid ethical foundations in order to guarantee their use for and by all, in a way that is reliable and protective of individual and collective rights. The role of the humanities and social sciences is decisive here, not only to take account of social and ecological limits while aiming to reduce the impact of the digital divide, but even to succeed in truly implementing a vision of healthcare based on the 5Ps (personalized, preventive, predictive, participative, proof-based).
Regarding frugality, it is both ethically and practically essential to reduce the environmental impact of AI, but also to design algorithms that are more data-efficient and simpler to use. This is a real ambition, and presupposes a serious trade-off between efficiency and complexity of methods, whereas until now, prototypes of AI have given us very spectacular demonstrations, but which can hardly be implemented to meet the real needs of the economy and society. This is also what we need to address, in connection with the first challenge.
Q: What does AILyS entail as a flagship teaching and research project for the Lyon Saint-Etienne site?
Aurélien Garivier : The plan is to build on the site’s strengths and skills, with the idea of first structuring what already exists and works, and then complementing it with actions that don’t yet exist. For example, in the mathematics and computer science disciplines at the core AI, we already have considerable expertise, with our laboratories based at Claude Bernard Lyon 1 University and ENS de Lyon, in conjunction with organizations such as INRIA and CNRS; health, with the HCL (Civil Hospices of Lyon), Centre Léon Bérard and our universities, as well as the Houraa health data grouping for the Auvergne Rhône-Alpes region; and natural sciences and engineering, with the technical support of MésoLyS, the future meso-computing center and data at the Lyon site, which will bring together the 3 major computing centers on the Lyon site. Our aim is to support and develop this very significant potential with a multi-disciplinary approach in which the social sciences, law and the arts are central.
As for the actions to be created or expanded, they relate to developing the attractiveness of the area for students in initial training, for people in continuing education, and of course for researchers and industrialists, not forgetting the general public and citizens. These actions are based on the richness and complementarity of our consortium. For example, we are creating two double bachelor degrees: math-computer science and IT-law. We are pooling our strengths in AI research, training and innovation, with a common thematic focus on AI and its development. The governance of the project will enable us to include the institutions, while being guided by the shared ambition of being a recognized reference in AI, capable of supporting the players in their transformation.
Some examples of planned actions and tools:
- the creation of a Graduate School to organize the transition from master’s to doctorate and the awarding of "PhD tracks", an attractiveness scheme;
- the creation of some thirty research chairs;
- the creation of a "modelling lab greenhouse" housing teams of 4 to 6 students, supervised by data analysis professionals and research chair holders;
- an ambitious focus on continuing education, with the creation of a multi-year program co-constructed by the academic and industrial sectors;
- the funding of innovative teaching, research and innovation projects through targeted calls for proposals;
- the creation of an "innovation springboard" to stimulate the transfer of AI technologies to the economy;
- the MésoLyS mesocenter, which draws on the cluster of 3 data centers on the Lyon site to provide the necessary computing and storage resources and technical support
- and a wide range of initiatives aimed at the general public, to promote an informed and critical view of AI.
The total budget for the project is ¤94M, comprising ¤29M from public institutions, ¤25M from companies, and ¤40M in support from the ANR (French national research agency).