A Framework for Developing Synthetic Agents as Pedagogical Teammates: Applying What We Already Know

Abstract

Research efforts are currently underway to use what we know about teams to ensure synthetic team members act like a human teammate. The research proposed here builds on this by leveraging what we already know about team training to develop synthetic teammates who can coach, provide feedback, and implement interventions during virtual exercises. This is potentially much more efficient and effective than the traditional use of avatars or even naïve, live teammates, allowing for personalized training and immediate feedback and coaching organic to the context. To facilitate this, we present a framework to integrate the research base of team training and pedagogical agents so we can begin to identify what leads to effective team training using synthetic teammates given training goals, team characteristics, and training environment.