Influence of Expertise Complementarity on Ad Hoc Human-Agent Team Effectiveness

As autonomous agents become more capable and widely deployed, teams of human and agent members will be seen more frequently. Ad hoc human-agent teams, formed with team members without prior experience with current teammates and deployed only for a limited number of interactions, will find diverse applications in dynamic environments. We focuses on ad-hoc team scenarios pairing human with agent where both need to assess and adapt to the capabilities and expertise of partner to maximize team performance. We investigate influence of different agent expertise distributions on effectiveness of such ad-hoc teams. We designed, implemented, and experimented with an environment in which human-agent teams repeatedly collaborate to complete heterogeneous task sets where agent and human expertise vary over different task types. Several hypotheses about effect of complementarity of team members on team performance and human satisfaction are evaluated.