​In military training environments, critical incidents such as injury, friendly fire, and non-lethal fratricide may occur. Quantitative performance data from training and exercises is often limited, requiring more in-depth case studies to identify and correct the underlying causes of critical incidents. The present study collected Army squad performance, firing, and communication data during a dry-fire battle drill as part of a larger research effort to measure, predict, and enhance Soldier and squad close combat performance. Soldier-worn sensors revealed that some quantitatively rated top-performing squads also committed friendly fire and a fratricide. Therefore, case studies were conducted to determine what contributed to these incidents. This presentation aims to provide insight into squad performance beyond quantitative ratings and to underscore the benefits of more in-depth analyses in the face of critical incidents during training. Squad communication data was particularly valuable in diagnosing incident root causes. For the fratricide incident specifically, the qualitative data revealed a communication breakdown between individual squad members stemming from a non-functioning radio. The specific events leading up to the fratricide incident, and the squad’s response, will be discussed along with squad communication patterns among high and low-performing squads in the context of various critical incidents. We will examine how the conditions surrounding critical incidents and the underlying causes of those incidents can be recreated and manipulated in a simulated training environment, allowing instructors to control the incident onset and provide timely feedback and instruction.  

​Artificial intelligence (AI) can facilitate personalized experiences shown to impact training outcomes. Trainees and instructors can benefit from AI-enabled adaptive learning, task support, assessment, and learning analytics. Impacting the learning and training benefits of AI are the instructional strategies implemented. The co-learning strategy is the process of learning how to learn with another entity. AI co-learning techniques can encourage social, active, and engaging learning behaviors consistent with constructivist learning theory. While the research on co-learning among humans is extensive, human-AI co-learning needs to be better understood. In a team context, co-learning is intended to support team members by facilitating knowledge sharing and awareness in accomplishing a shared goal. Co-learning can also be considered when humans and AI partner to accomplish related tasks with different end goals. This paper will discuss the design of a human-agent co-learning tool for the United States Air Force (USAF) through the lens of constructivism. It will delineate the contributing factors for effective human-AI co-learning interaction design. A USAF maintenance training use case provides a context for applying the factors. The use case will highlight the initiative of leveraging AI to help close an experience gap in maintenance personnel through more efficient, personalized, and engaging support.

Extended reality (XR) technologies have been utilized as effective training tools across many contexts, including military aviation, although commercial aviation has been slower to adopt these technologies. While there is hype behind every new technology, XR technologies have evolved past the emerging classification stage and are at a state of maturity where their impact on training is supported by empirical evidence. Diffusion of innovation theory (Rogers, 1962) presents key factors that, when met, increase the likelihood of adoption. These factors consider the relative advantage, trialability, observability, compatibility, and complexity of the XR technology. Further, there are strategic approaches that should be implemented to address each of these innovation diffusion factors.
 
This presentation will discuss each diffusion factor, provide exemplar use cases, and outline evidence-backed considerations to improve the probability of XR adoption for training. Considerations will discuss various effects that may occur with the introduction XR technology, such as the novelty effect where improved performance initially improves due to new technology and not because of learning. Key questions will be presented that should be addressed under each diffusion factor that will help guide the information needed to support the argument for XR adoption. Importantly, the quality of research evidence to support XR implementation and adoption is critical to reducing the risk of ineffective training. Therefore, a discussion of research-related considerations will also be presented to ensure an appropriate interpretation of existing XR research literature. The goal is to provide the audience with an objective lens to help them determine whether XR technologies should be adopted for their training needs.