Science of Human and Computer Teaming
Abstract
The Science of Human and Computer Teaming program will develop and demonstrate data-driven approaches for the formation and training of teams comprised of humans and computers. Conventional approaches to military personnel selection, role assignment, and training are optimized for individual performance, but military operations are typically performed by teams, and future teams are likely to also include autonomous systems that use artificial intelligence (AI) to sense, reason, learn, and interact. Behavioral scientists are studying the performance of groups across diverse sets of tasks and developing performance assessment techniques for group work. Interesting early results suggest that groups exhibit a form of intelligence beyond that of the individual members, and that group intelligence has social correlates. Computer scientists are looking at ways in which humans may team with computers to achieve superior levels of performance. Such human-computer teams have shown great promise in highly structured competitive domains such as chess. Realizing this promise in free-form (battlefield) environments will require intuitive, low-latency, high-bandwidth, human-computer interfaces that enable computers to be better teammates. The program will identify individual characteristics predictive of performance of mixed human-computer teams; develop techniques for measuring these characteristics in military personnel; demonstrate the capability to select, assign roles, and train human-computer teams with performance superior to that of human-only teams formed and trained using current methods; and develop an understanding of how to structure human-computer teams for superior performance on military missions such as cyber defense and intelligence analysis.
Document Details
- Document Type
- Accomplishment
- Publication Date
- Oct 01, 2017
- Source ID
- b08411ae80f3038116f4921e44f70fda
Related Documents
- Root: TACTICAL TECHNOLOGY