Henrique Ferraz presented a study on AI and social behaviour at NetSci 2026

During his presentation, Henrique Ferraz introduced a study focused on the use of large language models (LLMs) to simulate social interactions based on game theory, exploring to what extent these systems can reproduce human behaviour in collective settings.

The study involved the digital replication of a sociological experiment based on the networked Prisoner’s Dilemma, comparing results obtained from human participants with those generated by autonomous AI-based virtual agents. The findings showed that one of the analysed models was able to reproduce, in general terms, human collective behaviour, but not individual behaviour.

This work opens up new possibilities for the use of artificial intelligence as an experimental tool in the social sciences, enabling the analysis of complex social dynamics at scale, beyond the limitations of experiments involving real human participants.

NetSci is one of the leading international references in the field of network science, bringing together researchers from different disciplines to address challenges related to complexity, collective behaviour, and interactions in social, technological and biological systems.

ARAID’s participation in this conference strengthens the international visibility of research carried out in Aragón and its contribution to innovative approaches at the intersection of artificial intelligence and social sciences.