Program Summary
Humans intuitively combine pre-existing knowledge with observations and contextual clues to construct rich mental models of the world around them and use these models to evaluate goals, perform thought experiments, make predictions, and update their situational understanding. When the environment contains other people, humans use a skill called theory of mind (ToM) to infer their mental states from observed actions and context, and predict future actions from those inferred states. When humans form teams, these models can become extremely complex. High-performing teams naturally align key aspects of their models to create shared mental models of their environment, equipment, team, and strategies. ToM and the ability to create shared mental models are key elements of human social intelligence. Together, these two skills form the basis for human collaboration at all scales, whether the setting is a playing field or a military mission.
Artificial intelligence (AI) technologies have made little progress in understanding the most important component of the environments in which they operate: humans. This lack of understanding stymies efforts to create safe, efficient, and productive human-machine teams. The Artificial Social Intelligence for Successful Teams (ASIST) program seeks to develop foundational AI theory and systems that demonstrate the basic machine social skills needed to infer the goals and situational knowledge of human partners, predict what they will need, and offer context-aware actions in order to perform as adaptable and resilient AI teammates.
ASIST performers will work to create agents that demonstrate a machine ToM and the ability to participate in an effective team by representing and helping to maintain shared models. The program seeks to create and employ a testbed for evaluating these agents using customizable open-world environments and standardized interfaces. The interfaces will include a package of standard sensing channels (e.g., information streams from physical and virtual sensors that will be available to ASIST agents) and communication/action channels (e.g., mechanisms for agents to convey information to human teammates in the testbed and options they can use to engage with the team). ASIST agents must operate in increasingly complex and specialized environments; be adaptable to sudden perturbations in the mission or team, like the loss of communication with a key teammate; and use noisy multi-channel observations to represent the world and do complex inference and prediction.