Thesis Title
Understanding the Determinants of Online Social Conformity in Mixed Human-AI Environments

Research Description

Social conformity occurs when individuals adjust their behaviour to align with an opposing majority’s opinion, often to fit in, be liked or be right. Traditional conformity research shows that factors such as majority-minority group dynamics, task objectivity, self-confidence, and individual characteristics shape alignment in face-to-face settings, and that these effects persist in computer-mediated communication (CMC) groups despite reduced social cues. Advances in automation and artificial intelligence have also enabled agent-only environments, where conformity-like behaviours have been observed. However, previous literature has largely focused on homogeneous groups and has only briefly acknowledged the presence and effects of conformity in mixed settings. In contrast, contemporary CMC groups increasingly involve mixed human-AI interactions, where humans engage alongside LLM-based AI agents. In these environments, agents can act as both informational authorities and normative social actors, supporting informational influence through perceived expertise and reasoning quality, and normative influence through human-like language and social embedding. My work aims to investigate how social conformity manifests in mixed human–AI groups and how it shapes information consumption behaviours. It will extend traditional conformity theory to mixed environments by examining established contextual and personal determinants in mixed CMC groups, and by identifying new conformity determinants introduced by AI agents, with implications for the responsible design of CMC platforms.

Supervisors
Dr Damiano Spina, RMIT University
Dr Danula Hettiachchi, RMIT University