Current operational human-agent teaming paradigms place the full burden of danger on non-human agents. Shifting this burden entirely to the robot is currently possible due to the nature of the limited situations in which teleoperated robots are currently employed in military contexts. However, as the roles of non-human agents grow, robots are expected to function as teammates rather than tools. Here, we present a theoretical framework and metric for quantifying commanders’ attitudes toward risk and effort for humans and robots under their command. Twenty-one participants tasked a Soldier, a robot, and a heterogeneous team to rescue civilians in dangerous environments. Participants were risk averse when tasking each agent individually, but exhibited a risk averse attitude for the Soldier and an effort averse attitude for the robot when tasking the team. These findings show that risk attitudes can change as a function of team composition. The framework developed herein has utility for studying tradeoff attitudes across a wide breadth of contexts.
