AI Teammates Reduce Human Belonging and Communication Quality in Teams

A study published on arXiv reveals that conversational AI teammates negatively impact human-to-human communication in small-team decision-making. Researchers used Group Communication Analysis (GCA), surveys, and lexical analysis to compare 16 teams of two students plus an AI teammate against 17 all-human teams of three. The AI was the most talkative and self-cohesive member but contributed the least new information and lowest information density. In AI-human teams, humans showed lower responsivity and social impact toward each other and reported lower belonging. The study highlights sociocognitive costs of integrating AI as a teammate.