Artificial social intelligence (ASI) agents have great potential to aid the success of individuals, human–human teams, and human–artificial intelligence teams. To develop helpful ASI agents, we created an urban search and rescue task environment in Minecraft to evaluate ASI agents’ ability to infer participants’ knowledge training conditions and predict participants’ next victim type to be rescued. We evaluated ASI agents’ capabilities in three ways: (a) comparison to ground truth—the actual knowledge training condition and participant actions; (b) comparison among different ASI agents; and (c) comparison to a human observer criterion, whose accuracy served as a reference point. The human observers and the ASI agents used video data and timestamped event messages from the testbed, respectively, to make inferences about the same participants and topic (knowledge training condition) and the same instances of participant actions (rescue of victims). Overall, ASI agents performed better than human observers in inferring knowledge training conditions and predicting actions. Refining the human criterion can guide the design and evaluation of ASI agents for complex task environments and team composition.

Aptima and the Cognitive Engineering Research Institute are developing a mixed-initiative decision-support system for planning multi-Unmanned Aerial System (UAS) missions. The prototype capability is called MIMIC: Mixed Initiative Machine for Instructed Computing. At the core of the system is a model that employs machine learning algorithms to learn from operators during mission planning, and to use what is learned to aid subsequent mission planning tasks. This paper reports on the design of the prototype algorithms, the early interface design, and a series of three experiments performed to support the design of the system. First, machine learning algorithms were implemented that use Markov Decision Processes (MDPs) and Bayesian inference to learn a model of the human operator’s strategies. Second, a mission planning graphical user interface was designed to enable an operator to conduct basic multi-UAS mission planning, and to allow MIMIC to easily capture operator actions. Finally, three experiments were conducted (1) to identify typical operator planning priorities, in order to define the model’s features and to gather planning data to train the algorithms, (2) to evaluate the model by comparing its outputs to operators’ self-assessments of priorities and goals, and (3) to test the model’s ability to predict what an operator’s next actions will be, and compare the predictions to the actual operator actions. The MIMIC project represents a step toward increasing levels of UAS autonomy allowing for multi-UAS control by single operators.

Cyber insider threat is intentional theft from, or sabotage of, a cyber system by someone within the organization. This article explores the use of advanced cognitive and instructional principles to accelerate learning in organizational supervisors to mitigate the cyber threat. It examines the potential advantage of using serious games to engage supervisors. It also posits two systematic instructional approaches for this training challenge -optimal path modelling and a competency-based approach. The paper concludes by discussing challenges of evaluating training for seldom occurring real world phenomena, like detecting a cyber-insider threat.

The advantages of simulation in team training are well documented and accepted by trainers and students. The opportunity for teams to apply knowledge and practice skills to meet realistic problems in a realistic environment is intuitively more motivating, and can be a more valuable training experience for a team than merely reading about, being lectured on, or observing someone else perform a task. Training managers also are attracted to the economic and logistical advantages of linking multiple teams in a simulated environment when compared to the cost and effort required for comparable live exercises. To date, the primary focus of research in distributed simulation-based training has been the continuous quest for greater realism in visual displays; enhanced fidelity of the models that drive the simulation; and the need to improve the speed, reliability, and capacity of communications among different simulators and trainers. Less effort has been devoted to developing technologies that can increase the instructional value of a simulation exercise or that could help an instructional designer make the best use of a simulation-based training opportunity. Such technologies would help an instructor to design more effective scenarios; more accurately and efficiently observe, record, and analyze the trainees' performance; and support the necessary learning events before, during, and after the scenario to meet specific training objectives.

Training teamwork does not necessarily require teams, at least not human teams. Providing trainees with teammates that are synthetic entities can dramatically reduce the logistical complexity of teamwork training and the attendant costs. It can make teamwork training accessible on demand. This may increase the frequency with which individuals engage in teamwork training and, by increasing time on task (or time in team), boost individual proficiency in teamwork skills and improve the performance of teams of trained individuals. In this chapter, we describe the development of a teamwork skills training platform for Airborne Warning and Control System (AWACS) Air Weapons Officers (AWOs; formerly called Weapons 261