SOCIAPARIS

Project overview

This research project aims to analyse the uses of AI, and more specifically large language models (changes in professions and how they relate to traditional areas of expertise, and the transformation of working practices), and to understand the challenges (organisational, technical, legal, etc.) posed by AI.

An external survey of 15 cities conducted in 2025 on their initial reactions to the availability of large language models reveals a shared concern about the uncontrolled use of consumer-grade AI and a search for alternatives between general-purpose solutions (such as Copilot), semi-generalist solutions tailored to local authorities (led by Delibia), and sector-specific external solutions (such as MA IA for public procurement or Doctrine for law). Initial applications are for general editorial work and a few rare internal sector-specific tools; however, there is widespread caution regarding chatbots open to the public, due to the risks of hallucination. The implications for staff’s work still appear difficult to assess but are the subject of close scrutiny.

In terms of strategy, developments are being driven by a focus on regional development or the modernisation of the administration. Charters and guides have been developed in all towns, though the issue of environmental impacts still appears difficult to assess. An effort is being led by CIOs to engage elected representatives and other departments, such as HR, on the subject. Cooperation between cities appears necessary, but the approaches vary (CIO network, interconnections, CNFPT, Dinum). This initial approach introduces two surveys conducted within the City of Paris’s departments in 2026.

Internal survey on the use of large language models within the City of Paris. In-depth interviews with city staff who use an AI tool to explain the impact on their work and on relationships with colleagues. Focus group session for users to share their experiences.

Survey on image recognition tools applied to the urban transport sector.