Social AI agents in service: The new frontier

This dissertation studies how the growing use of AI agents in services changes what people expect from them socially. As AI moves from scripted, behind-the-scenes tasks into roles like subordinate, coworker, teacher, and leader, success depends less on technical ability and more on behaving in socially acceptable ways.

Across six chapters, the dissertation shows that even simple service AI must communicate emotion appropriately, that human-like looks can help at first but fail during service problems, and that coworker AI must follow key social norms like reciprocity to avoid being ignored or mistreated. It also finds that AI tutors need to create a safe social climate, for example by ensuring confidentiality, and that AI leaders can motivate more through charismatic signals.

Overall, the work argues that social legitimacy comes from enacted behavior, that people grant socialness selectively, and that “more social” only helps when it fits the role.