Artificial intelligence

"Lab IA": Inria at the service of the State for its digital transformation

Date:
Changed on 04/05/2021
The State’s public services have to convert to using digital technology... But to do so, they need tools adapted to their needs. This is the objective of AI Lab, which aims to introduce data science and artificial intelligence into administrative organisations to offer new work possibilities for employees and new functionalities for users.
Illustration IA
©NicoElNino - stock.adobe.com

Research is an essential ally in the digital transition of administrative organisations.

Support French administrative organisations in their digital transition and modernisation through artificial intelligence. This is one of the objectives of Dinum (Interministerial Department for Digital Technology). To achieve this aim, Dinum launched an initial call for expressions of interest (EOI) in 2018 that targeted administrative organisations, whether centralised or not.

The idea was to finance the implementation of digital or artificial intelligence tools to meet a specific need. The problem was that certain needs required significant scientific research and therefore could not be met by solutions that were already available on the market. The solution was to involve Inria to bring in additional research expertise.

This was how the Lab IA was born in early 2019. “The scheme is in fact a group of people pooling their efforts to establish the use of AI at different levels in order to facilitate the lives of citizens and public service employees,” says Ioana Manolescu, Scientific Director of the AI Lab.

Banking on project feasibility 

In concrete terms, Inria takes part in the selection of applications submitted during the EOI, alongside Dinum and the Interministerial Department for the Transformation of Public Action (DITP). In their applications, the administrative organisations give a detailed description of the data they posses, what they would like to do with it and the expected impact. “Our primary criterion is the feasibility of the project because the funding, approximately €100,000 per application, is only allocated for a year, which is very short for research,” says Ioana Manolescu. “Of course, we take into account the interest of the project and the fact it will provide an additional solution to those already on the market”.

Throughout the process, the Scientific Director of the AI Lab bears in mind which Inria teams could meet certain needs. “I joined the Institute in 1997 during my first-year Master’s internship... and I’ve been there ever since. So I know the structure well!” she says. “For several years, I was also a member of the Inria Evaluation Committee, which contributes to the evaluation of researchers and teams. This provides an overall vision of the Institute’s activities and research topics.” Using the annual activity reports of Inria’s 200 teams as reference material, Ioana Manolescu makes sure she stays up to date.

Robin Reynaud, Project Manager of the AI Lab at Dinum who works closely with Ioana Manolescu, recently made a cartographic schema of all the Inria teams that offers an accurate and up-to-date representation of their research topics. It is an invaluable tool for connecting teams to the administrative organisations’ requests. “Through the AI Lab projects, we connect Inria researchers with data scientists from administrative bodies to allow the latter to benefit from innovative techniques at the very forefront of their fields.”

The objective: an open source proof of concept

When we choose the applications, we rank them by level of technological maturity,” explains Ioana Manolescu. “Those that are well advanced are contracted out to companies. Those that require research come to Inria”. Out of the 15 projects chosen in 2019, four of them (three of which are still going: see boxes) resulted in a partnership agreement with Inria: Dinum finances the employment of a post-doc or engineer working in an Inria team to meet the needs of the project owner. “Inria provides 20% of my work time, 50% of that of my colleague Oana Balalau and the time of the researchers who supervise the newcomer during the project”, says Ioana Manolescu.

The long-term aim of this year of collaboration is to obtain a proof of concept. The administrative organisation is then free to move on to industrial production in-house or through private companies... or to submit a new EOI to develop the subject further. “It is important that everything created within the framework of the Lab AI is opensource and therefore reusable by anyone,” the scientific director says. “In this way, we produce technological building blocks that can then be used by other administrative bodies or for other applications”.

High-impact results

At the same time, Inria sets up internship agreements with administrative organisations for less demanding projects, or those that are further upstream in the research process and whose needs have to be further refined before starting an EOI application. Oana Balalau is supervising three internships due to begin in the second quarter of 2021. “The AI Lab is to everyone’s advantage: Dinum and the administrative bodies of course, but also for Inria researchers who can work on real data and obtain high-impact results,” says Ioana Manolescu. The EOI could not take place in 2020 due to the health crisis, but it will be held again this year with, let’s hope, at least as many interesting projects as in 2019!

Discover the projects led by the Lab AI