Could do better. The effectiveness of the national strategy for research in artificial intelligence (SNRIA) is “not proven”, according to a report Monday from the Court of Auditors in a mid-term report. The Court welcomed the “strong political signal on the importance of AI (artificial intelligence) for research”, provided by allocating to the latter 30% of the funding for the first phase (2018-2022) of the National Strategy for artificial intelligence (SNIA).
This first part, with in the end more than 550 million euros committed to research, has “made it possible to avoid a scientific dropout since 2018”, according to the Court. But his report notes that “the effectiveness of the strategy to strengthen France’s position in AI (…) is not proven”. As proof of this, in terms of the number of scientific publications on AI and out of a total of 47 countries compared, France “hardly retains a place in 10th place worldwide and remains in 2nd place at European level”. , behind Germany.
Long time
The Court admits that the “long time of research” can distort this assessment, and calls for closer monitoring of the effects of investments on France’s place in AI. She recommends strengthening the synergies between the centers of excellence in AI, and “adapting the governance and steering of the strategy”, which has “weakened over time and runs the risk of being ineffective. »
The Court also pleads for the sustainability of funding tools allocated to institutes over the first four-year phase alone. This is to attract and retain talent in AI research. The second phase of the national strategy (2022-2025), announced in 2021 and called “acceleration”, aims precisely to compensate for a “currently limited number of high-level public trainers”. To provide for this, the report advocates greater integration into European support programs such as “Horizon Europe” and “for a digital Europe”. »