What Happens When Humans Believe Their Teammate is an AI? An Investigation into Humans Teaming with Autonomy

Abstract

As artificial intelligence (AI) continues to grow in proficiency, the potential for AI to be used as team members rather than tools is becoming closer to realization. This advancement is driving new research investigations into the applicability of human-human teamwork knowledge to the context of human-autonomy teaming. In the current study, we apply qualitative methods to explore how the perceived composition of a team (how many humans and how many agents on the team) affects sentiments toward teammates, team processes, cognitive states, and the emergence of a system of team cognition. A total of 46 teams completed a teamwork simulation task and were interviewed afterwards regarding their teamwork experience. All of the teams were comprised of only humans; however, two conditions were led to believe that their teammate(s) were autonomous agents. Interviews were analyzed using grounded theory and the Gioia methodology, which revealed thematic differences between the team compositions. In light of our results, we offer a new model that describes how early-stage action teams achieve effective team processes and emergent cognitive states.