Artificial Social Intelligence to support Macrocognition in Teams (ASIMT)
Abstract
The vision of Artificial Social Intelligence to Support Macrocognition in Teams (ASIMT) is to lay the foundation for helping to produce both a symbiotic and a synergistic relationship between human and machine. ASIMT envisions a relationship between warfighter and technology that is symbiotic because members will learn from each other. With present day human-human teams, member collaboration over repeated interaction episodes informs knowledge of their task, their situation, and teammates. ASIMT envisions a relationship between technology and warfighter as synergistic because members will be able to coordinate cognition and behavior. In current human-human teams, effectiveness means coordinating behavior and knowledge such that the collective is able to do more than the individual members could achieve on their own. But ASIMT supports development of future hybrid human-machine teams doing more. ASIMT envisions a future capability to leverage what technology does well and what humans do well to produce a collaborative synergy where cognition and coordination are amplified. To achieve these objectives, we propose to develop innovative, plausible models and associated hypotheses to advance state-of-the-art understanding for human-machine teaming. We integrate a set of disciplines to produce a synthesis that can both advance theory on teamwork but also provide a deeper understanding of how artificial social intelligence (ASI) can be appropriately developed to augment teamwork. First, we leverage theory and methods from Developmental Psychology and Social Neuroscience to examine how social cognitive processes (e.g., theory of mind) can be implemented to create agent architectures capable of monitoring and intervening in teamwork. Second, from the Organizational Sciences, we examine how shared mental models and transactive memory systems can be learned and implemented in ASI to enhance team coordination. From Group Communication theory, we adapt social-roles theory to consider the communicative elements that need to be monitored and/or used to support collaboration in teams with varying forms of expertise and responsibilities on a team. From Human Factors and Computer Science, we integrate theory on intelligent agents, trust in technology, and situation awareness, to explore artificial intelligence and machine learning approaches capable of monitoring teamwork processes in complex environments. Our proposed research will provide the following innovations: 1) The integration of the current theoretical distinction between online versus offline social cognition and how to study/implement these different cognitive processes with agent technology; (2) The integration of theory from the cognitive and computational sciences with fundamentals about team process from the social and organizational sciences using the Macrocognition in Teams Model (MITM) as a foundation; (3) Consideration of fundamental distinctions between teamwork and taskwork and how to account for them when developing Artificial Social Intelligence (ASI) capable of monitoring and intervening in teams; and, (4) Using these aforementioned concepts to help specify multiple theoretically grounded entry points and monitoring paths for ASI. These innovations are explored within the MITM to study collaborative cognition in human-machine teams and how social and technical roles may vary dependent on the phases of problem solving (e.g., solution development versus solution evaluation). In this way, ASIMT supports development of a cybernetic team with an augmented capability that is truly a force multiplier.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Dec 04, 2019
- Source ID
- W911NF2010008
Entities
People
- Stephen Fiore
Organizations
- Army Contracting Command
- Defense Advanced Research Projects Agency
- University of Central Florida