State AI Lab (FR0037)

Set up an open artificial intelligence (AI) lab for the State
Lead institution(s):
Minister of State for the Digital Sector, attached to the Prime Minister
Other stakeholders:
Artificial intelligence institute, all ministries
New commitment
OGP principles with which the commitment is associated:
Innovation and technology for openness
Challenges
Promoting the movement of data between the public and private spheres requires the necessary skills for making use of this data. A network of distributed skills is therefore the ultimate aim, beginning with a cluster within a more centralised AI lab. Ideally, there would be a mix of public sector and external skills.
Ambitions
Lay the groundwork for the open AI lab for the State, define the intended organisation of the team and set the trial processes in motion with the ministries.
The intended organisation of the team is as follows:
A core team made up of data science specialists and public reformers;
For each project, dedicated recruitment modelled on the General Interest Entrepreneurs (EIG) competition for attracting specialists, who will be given the opportunity of embarking on pathways of excellence.
The road map in detail

Publication of the ministries' AI and digital road maps Semester 1 or Semester 2 2018
Calls for AI proposals for the attention of government departments Semester 2 2018
Setup of the AI lab Semester 1 2019

8. Set up an open artificial intelligence (AI) lab for the state

Language of the commitment as it appears in the action plan:

Promoting the movement of data between the public and private spheres requires the necessary skills for making use of this data. A network of distributed skills is therefore the ultimate aim, beginning with a cluster within a more centralised AI lab. Ideally, there would be a mix of public sector and external skills.

Lay the groundwork for the open AI lab for the State, define the intended organisation of the team and set the trial processes in motion with the ministries.

The intended organisation of the team is as follows:

A core team made up of data science specialists and public reformers;

For each project, dedicated recruitment modelled on the General Interest Entrepreneurs (EIG) competition for attracting specialists, who will be given the opportunity of embarking on pathways of excellence. [37]

Milestones

8.1 Publication of the ministries' AI and digital road maps

8.2 Calls for AI proposals for the attention of government departments

8.3 Setup of the AI lab

Start Date: 2018

End Date: 2019

Context and Objectives

Under the guidance of Cédric Villani, mathematician and member of Parliament, the government tried to give more visibility to artificial intelligence (AI) and to integrate it in the work of the public sector. The national action plan mentions the need to acquire sufficient skills to profit from the potential of big data and data circulation. Villani’s 2018 report adds that AI has the potential to better anticipate the transformation of the labor market (using data on skills, unemployment, and needs), to improve health services (developing prediagnostic tools) and public education, to optimize the transportation of people and goods (developing tools for an adapted regulation of traffic), to mitigate climate change (helping consumers understand and limit their use of energy), and to strengthen national defense, especially in the context of new threats. [38] In 2018, internal knowledge about AI was limited and unequally distributed among government agencies. [39]

A team of data scientists within Etalab work across the government. In addition, small Etalab teams work within select agencies (e.g., the national employment agency, Ministry of Interior, Ministry of Defense, and Ministry of Finance). [40] The commitment aimed to improve the integration of AI in the government through the establishment of an AI Lab and a call for proposals related to AI for government agencies. The IRM researcher considers the commitment relevant to access to information. One of the milestones requires the publication of ministries’ AI and digital road maps.

The IRM researcher considers the commitment, as written, specific enough to be verifiable. However, the milestones do not provide enough detail about what the road maps for ministries entail, and exchanges with officials in charge did not provide additional information. Milestones 8.2 and 8.3 are vaguely explained in the action plan, and additional information is available on dedicated government websites. The call for proposals gave agencies in charge of delivering public services a chance to experiment with this new technology. [41] The AI Lab will be an interministerial entity, within the Digital Information and Communication system. It will serve government agencies in their efforts to use AI, through the support of in-house data scientists and a network of researchers. [42]

Although AI is undeniably a growing area of interest, the IRM researcher finds that the commitment has a minor potential impact. It mainly sought to stimulate internal interest in AI and did not, per se, require any actual concrete AI project. Moreover, it did not aim to train officials to use AI but required the support of trained data scientists.

Next steps

Despite the general relevance of AI in today’s world, the IRM researcher suggests this commitment not be carried forward to the next action plan. It only required the publications of governmental road maps but did not include any public-facing element that would improve citizen participation or public accountability.

8. Set up an open artificial intelligence (AI) lab for the State

Completion: Substantial

A decision of the interministerial committee on public transformation set up the Lab IA to help administrations use and develop AI (via projects, trainings) and anticipate the effects of AI on public administration. [xxxvii] It published a call for proposals for administrations that wished to develop AI projects, with six projects selected in 2019 (such as from the Nuclear Safety Agency on the use of data from inspections, the French Agency on Biodiversity on the improvement of controls thanks to data) and 15 in 2020 (such as the General Directorate for Health on optimizing alerts on undesirable health events, the Council of State on the automatic identification of cases that refer to the same decision) . The Ministry of Armed Forces published an AI roadmap in 2019, but no evidence was found that any other ministries published such a strategy.