Bridging Minds and Machines: Mapping the Terrain of Communication in Human-AI Teams — A Systematic Literature Review

Abstract

As AI becomes increasingly adept at communicating with humans in natural language and complementing human skills in various collaborative and teamwork settings, how the CSCW community understands the evolving role of communication in human-AI teams (HATs) urgently needs to keep pace with these technological advancements. To provide a structured overview of the scientific knowledge regarding communication in HATs, we conducted a systematic review of 74 articles, synthesizing how communication has been conceptualized, operationalized, and linked to team processes and outcomes. Our review reveals that existing research overwhelmingly relies on text-based communication and is concentrated in controlled experimental and simulated settings, with communication outcomes often measured using platform-specific or frequency-based metrics, limiting generalizability across tasks and contexts. Across this literature, communication is most commonly modeled through a functional input-output lens, leaving the communicative processes that transform inputs into taskwork and teamwork outcomes under-explored. We synthesize these findings into an input–process–output framework that reconceptualizes communication as a socially embedded team process, which can inform methodological choices and the purposeful and responsible design of communicative AI teammates in CSCW contexts.