On 29 March 2018, the President of the French Republic presented an ambitious policy for artiﬁcial intelligence (AI) and launched a National Strategy for AI, during a day of debates at the Collège de France. This National Strategy for AI (SNIA), with a budget of €1.5 billion in public funds for five years (2018-2022), includes three objectives: to achieve the highest scientific level in AI by training and attracting the best global talent in this field, to widespread artificial intelligence in the economy and society, notably through startups, public-private partnerships and data sharing, and, finally, to give it an ethical frame.
The National AI Research Programme
45% of the National AI Strategy dedicated budget is devoted to the National AI Research Program (PNRIA), the coordination of which has been confided to Inria. This programme has several priorities:
- Promote attractiveness and talent support programs;
- Develop AI training at all levels (initial training, continual education, and top-notch training at Master and Doctorate levels);
- Expand State-of-the-art digital infrastructures;
- Support research in AI through dedicated calls, operated by the French National Research Agency (ANR);
- Strengthen and foster joint collaborative strategies at European and international levels, and
- Boost private-public partnerships and increase the impact of research in AI.
A synergy of talents
In order to contribute to raising France's attractiveness by supporting talent, the scientific community is mobilize throughout the country with the 3IA ANITI, 3IA Côte d’Azur, MIAI Grenoble Alpes and PRAIRIE, the DATAIA, Hi! Paris and SCAI centres, the 43 research chairs selected outside the 3IAs and the 22 PhD programs (~300 PhD grants).
A Program: The Jean Zay supercomputer
The Jean Zay supercomputer is the first French supercomputer where high performance computing and AI facilities are converged. Its computing power reaches more than 36,85 petaflops per second. Jean Zay has already enabled the realisation of more than 1,200 projects, including BLOOM, the largest multilingual "open science" model, with 46 languages including 20 African languages and 176 billion parameters. Jean Zay is also a champion of energy efficiency thanks to its latest generation hot water cooling technology, the energy of which is recovered to heat local buildings in the Paris Saclay area, a first on this scale in Europe.
Coordination of specific actions
The National AI Research Programme is also conducting a series of structuring actions for the French ecosystem.
The CNRS has launched its AI for Science and Science for AI (AISSAI) centre with the main objective of structuring transversal actions involving all CNRS institutes at the interfaces with AI.
The Regalia pilot project, launched by Inria, aims to build a software environment for testing and helping to regulate digital platforms.
The mission of this resource and experimentation centre is to better understand artificial intelligence and its effects on work, employment, skills and social dialogue in order to develop business practices and public action.
The PMIA is a multi-stakeholder initiative, bringing together scientists, industry, international organisations and governments, to foster the emergence of responsible AI. Inria provides support to two expert groups "Innovation and Commercialisation" and "Future of Work".
The launch of the Acceleration Strategy actions
In November 2021, a new phase of the national strategy for AI was launched as part of the France 2030 Plan, with €1.5 billion in public funding.
On the research side, this acceleration strategy allows the funding of the following actions:
- PEPR (Priority Research Programmes and Equipment) piloted by CEA, CNRS and Inria, dedicated to embedded AI and frugal AI, decentralised AI and trusted AI, as well as the mathematical foundations of AI
- Development of the Deepgreen infrastructure for embedded Deep Learning
- Development of the P16 infrastructure for the extension of Scikit-learn