1. médialab Sciences Po
  2. News
  3. Retour sur la participation de l'équipe EL2MP à l'atelier Data & Society

Retour sur la participation de l'équipe EL2MP à l'atelier Data & Society

On May 8 and 9, the Écologies des pratiques LLMs (EL2MP) project team took part in a two-day workshop organised remotely by Data & Society.

Post

On May 8 and 9 2025, the Écologies des pratiques LLMs (EL2MP) project team took part in an international workshop organised online by the Data & Society Institute. Bringing together around fifty participants, this event alternated between research presentation sessions and collective discussions on contemporary AI issues.

The EL2MP team was the only French and European representative to take part in a dedicated session.

Composed of Donato Ricci, Gabriel Alcaras, Zoé de Vries and Tommaso Prinetti, the team presented the initial results of its survey during a one-hour session. Based on a text shared in advance, the presentation and project were informed by feedback from a dozen researchers from a wide range of backgrounds, including sociology, economics and media studies.

Main feedback on the presentation:

  •  highlighting the work generated by LLMs, beyond their impact on existing tasks (prompting, assessment, etc.). The hypothesis is that AI introduces new forms of work and practices.
  • the preliminary observation that these new tasks, particularly prompting, are generally not appreciated by those who carry them out, as they are often perceived as unskilled or tedious.
  • the intimate perception and little-shared use of AI, rarely discussed between colleagues, reinforcing the dynamics of isolation and non-transparency within work groups.
  • the importance of contextualising usage, depending on the degree of autonomy users have in relation to the tools, particularly between open environments and more restricted systems.
  • a discussion on deskilling, which could apply to higher intellectual professions not traditionally affected by AI. Indeed, the use of AI is not systematically perceived as a qualifying skill.
  • finally, feedback on the team's methodological protocol (group work, exercises) highlighted its potential in an educational context, encouraging the sharing of experiences and reflexivity.

The feedback provided invaluable input for improving the text with a view to publication, and highlighted key aspects of the uses of AI: the work it generates, the intimate and sometimes problematic nature of these uses, and the potential for downgrading rather than upgrading.