Early conversational agents (CAs) focused on dyadic human–AI interaction between a person and a CA. Multiparty human–AI interaction, in which a CA mediates interaction among people, has since become more widespread. Research on multiparty CAs is scattered across fields, making existing knowledge difficult to identify, compare, and build upon. This study reviewed ACM publications to identify UX research and synthesized the effects of multiparty CAs across four dimensions: communication, engagement, connection, and relationship maintenance.