Establishing Common Ground between Humans & Machines Pitch Day for Trusted Artificial Intelligence

Across the military services and into the commercial sector, there is an increased focus on teaming human operators, maintainers, etc. with intelligent agents to improve system performance.  For these hybrid teams, developing, communicating, and maintaining a shared situation assessment is critical to properly calibrating human trust of the intelligent agent’s input and maximizing the performance of the hybrid team.  Researchers refer to this process as establishing and maintaining “common ground” (also called “grounding”).  While many researchers assert that common ground is important for many aspects of teamwork, fewer authors look into the process by which common ground is achieved.  Early models of grounding were not amendable to machine implementation and only functioned in a post-hoc manner.  More recent computational models have considered the roles of various speech acts and media in the development of common ground.  Although progress has been made, a generalizable utility that can be used by a variety of systems for establishing and maintaining common ground does not exist.  To fill this gap, Sonalysts, Inc., the University of Southern California Institute for Creative Technologies, and CAMO, LLC will deliver Situation Assessment Management Extension (SAME).  SAME is a utility that developers can “plug in” to a variety of intelligent agents to assist with the process of developing common ground.  SAME will employ state-of-the-art natural language processing and grounding models to allow humans and agents to assess the degree to which common ground has been achieved and manage any discrepancies in their respective assessments.