International Defence Excellence and Security Symposium

Human-Autonomy Teaming for UAS Design, Development, and Deployment

With its ubiquitous application and transformative potential, AI holds the promise of delivering both incremental and profound changes in many aspects of our society including military operations. However, AI (or automation or machine intelligence more broadly) is not a silver bullet for eliminating human errors when information processing and timely decision-making demands exceed the capacity of human operators. Many AI-enabled systems such as UASs operate in environments that are not specifiable a priori. Further, there will always be environments (e.g., for missile and air defense operations) in which data come from multiple sources, including new or partially unreliable ones, and where the integrity of data sources and AI processes and outputs cannot be fully trusted. The problem of trustworthy AI and thus human-autonomy teaming (HAT) is complex and multidimensional and it is largely dependent on both effective human-machine interaction and robust, reliable AI systems. To achieve the full HAT potential for C2 operations, there is the need for mutual trust between human operators and AI-enabled decision aids in the context of wide area surveillance, advanced training systems, and after-action analysis, etc. With such trusted partnership, a persistent and reliable HAT decision support capability will help enable and ensure HAT capabilities and thus achieve information dominance and decision superiority.