This Wednesday, at the AI for Social Sciences and Humanities Workshop, hosted by BSC AI Factory, Paula Szewach and Alejandro de la Fuente presented “LLMs and the Study of Social Behavior: Using AI to Understand Social and Political Phenomena”, featuring work from WHAT-IF, including the platform developed by the WHAT-IF team.

“AI for Social Sciences” was the focus of the workshop’s first day, making sharing the features of STAGELab, Simulated Theater for Agentic-Generative Experiments, a natural fit for the programme.
A little bit of context:
- STAGELab is a generative multi-agent coordination framework, developed by WHAT-IF members, that lets researchers describe experimental conditions in plain language and produces a live chatroom that realises those conditions around a real human participant.
- In our project it is used to conduct field experiments exploring how disinformation, uncivil messages, and selective information exposure affect democratic citizenship.
The workshop also addressed the limitations and challenges of using AI in research, aiming to equip social science researchers with the tools and critical perspectives needed to harness the potential of LLMs in their work while maintaining high standards of research quality.
A big thank you to Paula and Alejandro for presenting and sharing updates on WHAT-IF’s work!

