FOCUS EVENT: Revolutionizing Training with Generative
Generative AI, such as ChatGPT, has burst onto the scene as the latest application of generative natural language processing, taking the public by storm and demonstrating its impressively (and eerily) human-like conversational skills. Millions are now using it and other AI generative models, experimenting with and exploring their possibilities, asking questions, and giving various writing assignments and other tasks. With these AI models becoming more robust and mature, how might they be put to use productively and safely in training, education, and simulation? Beyond how they might support their human counterparts as highly personalized assistants, could they fundamentally change the way we learn, train, and work in today’s digital world? Although these natural language models can create responses that appear accurate and thoughtful, they lack true human-like understanding or insight. And as they become more sophisticated and human-like, they might engender a false sense of trust or an exaggerated mistrust, overreliance, or miscalibrated confidence by their human users. What processes or guardrails will we need to ensure productive human-AI teaming that will provide accuracy, data integrity, and explainability? This panel of leading thinkers from defense and industry will address and explore with the audience the current and potential applications of generative AI in training, simulation, and education in both military and civilian domains.
To create and maintain an advantage with peer and near peer advisories, it is critical that our warfighters are equipped with systems that are designed to consider their physical, behavioral, and cognitive needs. I/ITSEC provides an excellent opportunity for groups such as the government Human Systems COI and industry to reach out to the larger DoD community and discuss human systems issues as they relate to training, modeling, and simulation. Research shows that attention to human systems design is critical for the prevention of human error, which accounts for most accidents and incidents across a wide range of systems. Unfortunately, human performance research is not routinely transitioned to defense acquisition programs. Also, with no specifications required for human systems integration in acquisition programs, Requests for Proposals (RFP) seldom include evaluation criteria for it, and it is ignored by program managers. Human systems design issues must be considered early during system design to reduce subsequent operations and maintenance costs, minimize accidents and incidents that negatively impact safety and costs, and improve the effectiveness of the combined human-system for achieving mission outcomes. This panel of experts will discuss the importance of recent human systems documentation and guidance, specifically the Human Readiness Levels (HRL) as they relate to modeling, simulation, education, and training solutions, all of which include the human. The panel will educate the attendees on what the HRLs are, how program managers can and should build those requirements into RFPs, and challenges associated with implementing the guidance.
What if an authoritarian leader of a nuclear nation could improve their cognitive abilities by 100x using implants? Can multimodal deepfakes turn a population against its government in five days? Will children be cognitively inoculated against misinformation before the end of this decade? These are but a few questions the NATO Industry Advisory Group (NIAG) imagined as part of Study Group 278 on Cognitive Augmentation for Military Applications at the 2040 horizon. The commoditization of generative AI and the vast breadth of emerging cognitive technologies, combined, will yield new opportunities and threats for NATO nations. Understanding this landscape is critical for government stakeholders and industry decision-makers as they seek to leverage such prospects and anticipate/mitigate the related threats. In this special event, servicemembers and industry representatives will get a primer on the SG-278 report, focused on four components for cognitive augmentation: training and readiness, neurotech, emerging tech, and ethical, legal, and societal implications (ELSI).
Demos: "Proficiency & Readiness at Scale"
AI Assistant for Instructional Systems Designers
Accelerating ISD analysis to create and update training at “Speed to the Fleet”
With the US Navy’s requirement to create and update training, ensuring with “Speed to the Fleet,” there is an urgent need to find automated and reliable methods to accelerate for Instructional Systems Design (ISD) analysis. In this demonstration, Aptima is showcasing its application of large language models (LLMs) that support automating the data collection and development of training artifacts, data-enabling components to reduce time and cost.
At-The-Edge Training Matchmaker
Optimizing the match between task requirements, individuals, training needs, and resources
This innovative solution will extend training systems to deployed locations to give Warfighters access to the training they need when they need it. Instead of guesswork and ad hoc task assignment, it will systematically match the mission requirement to individuals with the most relevant competencies and skills, then close the skills gap by delivering targeted training, assessing learning to validate proficiency, and providing job aids to reduce the reliance on memory during mission execution. The system will consist of a decision augmentation tool, human digital twins that represent the competencies of each Warfighter, mission task inventories that map to competency models, training content repositories also mapped to competency models, and a suite of user end devices (tablets, VR headsets, AR glasses, etc.) connected to a deployable network infrastructure that can effectively scale to a wide range of operational environments, austere to modern.
Neurophysiological Measurement for In-The-Moment Adaptive Training
Personalizing training to maximize learning and trainee throughput
Aptima has developed a training enhancement platform that provides a neural indicator of stress and engagement fused with instructor observation to provide multimodal state assessment of trainees. The platform provides instructors with a deep understanding of trainee state to enable them to intervene at the right time, provide targeted after-action reviews to trainees, and ensure maximum trainee throughput. This innovative system leverages Aptima’s extensive experience in physiological measurement, instructional support tools, and performance assessment to enable in-the-moment adaptive training.
Proficiency and Readiness Optimizing Platform and Ecosystem for Learning
Proficiency defined: Tailored training and optimized readiness
The Proficiency and Readiness Optimizing Platform and Ecosystem for Learning (PROPEL) is a precision training solution that ingests and assesses observer- and simulator-based human performance data from training and operational environments. PROPEL uses LLM-based algorithms and recommender tools to provide personalized and tailored training at scale as well as adaptive custom learning paths for every learner. It provides actionable insights and trend analysis to include the measurement of training effectiveness.
Augmenting Training Operations Management with Human Performance Insights
Identifying opportunities for additional learning within simulation data for trainee support and remediation
Aptima and Lone Star Analytics (LSA) are combining LSA’s MaxUp Training Enterprise Model (TEM), which enables “what-if” analysis for planning, operations, impact, and risk mitigation, with Aptima’s Proficiency Based Training tool for tracking, analyzing, warehousing, and visualizing human performance data. The resulting capability identifies opportunities for additional learning within simulation data for trainee support and remediation.
Wires Crossed in A Digital World: How to Prevent Misalignments in Human and AI Decision Making
* Best Paper 2023, Emerging Concepts and Innovative Technologies Subcommittee
Authors: Maria Chaparro Osman, Summer Rebensky, Audrey Reinert, Valarie Yerdon, Chris Jenkins, Jianna Logue, Charles Jusko, Gabe Ganberg
Tuesday, November 28, 3:00 PM, Room W300 – THEATRE
Simulators Provide Adequate Training – Says Who?
Authors: Alexxa Bessey, Brian Schreiber, Mark Schroeder, Steve Macut, Winston (Wink) Bennett
Tuesday, November 28, 2:30 PM, Room W307D
Virtual Reality Provides Real Data: How Data in VR Transforms the Concept of Readiness
Authors: Summer Rebensky, William Stalker, Jonathan Diemunsch, Shawn Turk, Samantha Perry, Quintin Oliver, Winston Bennett
Wednesday, November 29, 11:30 AM, Room W307B
Unobtrusive Measures and Understanding Team Processes
Authors: Alexxa Bessey, Robert McCormack, Kara Orvis, Marissa Shuffler, Dorothy Carter, Amanda Thayer, Eduardo Salas
Wednesday, November 29, 4:30 PM, Room W308C
Toward a Theory of Human-AI Co-Learning and Trustworthiness
Authors: Fred Diedrich, Gary Riccio, Tatiana Toumbeva, and Scott Flanagan
Thursday, November 30, 11:00 AM, Room W307B
Augmented Maintenance: Setting Expectations for Augmented Reality
Authors: B. Adrian Flowers, Michael J. Keeney, Werner Born, and Jeffrey Beaubien
Thursday, November 30, 11:30 AM, Room W307C