In early 2003 it was identified that the aging T-38 Talon fighter trainer aircraft, then over 30 years old, was coming due for replacement. A replacement plan was eventually formulated and went through multiple rewrites and congressional budget delays. Finally on the 20th of March 2015 the United States Air Force (USAF) released the official requirements for the replacement aircraft, called the T-X program. In September of 2018, Boeing-Saab was selected as the winner of the T-X bid, worth an estimated $9.2B. One year later, Boeing designated the aircraft as the T-7A Redhawk, pictured below.

Photo Curtesy of Boeing

Unlike previous training aircraft programs, this request involved not just the development of the aircraft but an integrated training environment involving simulators that can communicate with other simulators and aircraft in air, realtime via data links. The aircraft and the simulators are required to allow instructors on the ground or in the air to create a host of simulated training scenarios for the students. This landmark forward-looking plan is intended to train pilots to interact with a new generation of aircraft, some already fielded and some still to come, in a way that the Air Force has never trained students before.

Coincidently, the United States Air Force is facing a significant challenge in pilot training. Over 10 years ago, it was identified that there was a shortage of pilots in the Air Force. Several half measures were attempted, none of which ultimately solved the problem. In 2022, it was outlined that the Air Force was over 12% below its minimum pilot manning requirements, almost 1700 pilots short (Parrish & Hoehn 2022). Since then, there has been a sense of urgency to find a solution, and while the shortage is shrinking slightly, there is still a major shortfall. The two primary solutions identified were: 1) to increase incentives for pilots to stay in active-duty service rather than leave for other careers like commercial airlines, and 2) to produce more pilots at a faster rate.

As a USAF instructor pilot with nearly 8 years of flying, and currently working on the T-7 project as an operational test pilot, I've aimed the focus of this work to the educational challenges associated with the second solution; producing the most high-quality USAF pilots as possible. Currently, the USAF has the fastest pilot training program in the world. Trimming this timeline down even more presents a significant risk to the level of training required for student pilots to saftley progress onto highly challenging aircraft. On top of that challenge, the instructor pilots used to train new students are drawn from those required manning numbers, thus exacerbating the manning issue further. So, how does the USAF produce pilots at a faster rate using fewer instructor pilots and ensuring the quality of student education and skill level?

The development of the T-7 Redhawk and the Advanced Pilot Training environment of systems provides the unique opportunity for the USAF to solve this problem.

Developments in Artificial Intelligence (AI) can be utilized to create adaptive learning management systems and free up instructor task loads. This would also reduce the required number of instructors as instructors could oversee more students. Additionally, the integration of Extended Reality (XR) into the program can significantly increase student access to low cost, high return simulators. XR is an "umbrella term that encompasses any sort of technology that alters reality by adding digital elements to the physical or real-world enviroment by any extent" (Tremosa 2023).

Tremosa 2023: The Interaction Design Foundation

While the focus of this work is on military aviation training, the challenge of providing high-quality learning opportunities to the greatest number of students, while freeing up teachers for the tasks AI can’t and shouldn’t replace applies to the educational community at large.

Since 1939, USAF pilot training has almost solely followed a pattern of didactic pedagogy, with instructor driven classroom settings (example below), leading students through the syllabus lessons to prepare students for some form of test. With the emergence of new technology, and facing a pressing need, the USAF Air Education Training Command (AETC) has the opportunity to facilitate a shift in the educational approach for pilot training. This transition towards a student-centered pedagogy focuses on providing the greatest amount of educational opportunity, peer-interaction, and timely assessments and feedback as possible for the student. Additionally, new technology in military aircraft has driven a change in the role of operators moving away from a role characterized mainly by sensors and motor skills to a role that now delves more into cognitive skills and decision-making processes (Mosier and Ficher 2010).

U.S. Air Force photo/Dan Hawkins

Currently USAF pilot training consists of 3 phases. Phase 1 is approximately 3 months of instructor led PowerPoint classes where students are spoonfed a topic for a day or two followed by a test. This process is augmented by computer-based learning which consists of more PowerPoint slides, sometimes paired with audio. This approach focuses on providing as much information as possible to the student as quickly as possible. This gauges a student’s ability to temporarily memorize facts and numbers, not actually testing the student’s comprehension or retention of the material. The majority of pilot training across the aviation industry follows a similar format. According to Moore, Lehrer, and Telfer's (2001) study, the conventional pilot ground education that professionals perform in classes is more on a superficial or mechanical stage as opposed to a deeper or intrinsic level. Most instructors adopt the approach of "teaching to the test" to prepare students for exams. Lakowske, Breese, and Callejo (1999) argue that computer-based training (CBT) falls short due to the lack of comprehensive user interaction needed for in-depth exploration of complicated tasks which demand "what-if" analysis. They propose a "closed-loop" training system, which incorporates feedback from flight operations quality assurance programs and visualization systems, as a superior alternative to the prevalent "open-loop" system characterized by infrequent feedback and insufficient objective data for evaluation in pilot education (as cited in Dincer 2023).

Following phase 1 academics, phases 2 and 3 are mixed aircraft and simulator flight training. Currently, most simulation and flight events are a one-on-one, instructor-to-student setting. The instructor will give a short brief of the scenario and the desired learning objectives then the teacher and student will go fly that scenario. Immediately following the event, the student will receive a grade with a small amount of feedback and either progress past that training event or fail, sometimes being added to a probabtion status. It’s a very didactic and linear pattern, providing very limited trial-and-error opportunity for non-retribution practice. Additionally, analysis and feedback is limited due to a particularly difficult challenge for instructors to manage split attention between insuring safety of flight, while multitasking verbal and control corrections, and taking shorthand notes on a small lap-mounted notepad. This picture of students flying in formation with mearly 7 feet of seperation, moving at hundreds of miles per hour, is just one example of many challenging scenarios where the distraction of notetaking could be deadly. 

USAF photo by: Senior Airman James Crow

 Student feedback often relies on the combination of these instructor shorthand notes, and a recollection by memory of the past 1-2hr event time. Additionally, this method relies on a student’s ability to absorb feedback after a mentally and often physically taxing experience. Finally, students and instructors debriefs are often rushed to prepare for and fly the next flight event only a few hours later. This drastically reduces the student’s ability to ingest feedback and practice techniques before being evaluated again.

The objective of pilot training currently is to complete a defined list of syllabi events directed as requirements in the shortest possible time. The objectives of student comprehension and capability takes a backseat to the priority of timeline and event completion. If students are slow to grasp a topic, then it often reflects in their gradebook as the student being a slow-learner or a weak swimmer. This can change a student’s trajectory in a matter of days with a few failed flight events. This timeline provides the student with very few opportunities to practice the problem area on their own, receive useful feedback, and make improvements.

This operations tempo has also created several challenges for the training units, requiring a high number of daily flights from each instructor, and strenuous demand on the aircraft and maintenance teams to generate the necessary sorties every day. This is a particular challenge for the aging T-38 fleet as mentioned earlier. These challenges present a need for a fundamental change in how pilot training is approached.

This is where technology now provides opportunities that have never previously existed to solve the challenges facing USAF pilot training and produce the high-quality pilots in the most cutting-edge way possible. This requires the educational paradigm shift away from a didactic, numbers driven focus, towards a focus on what the most effective educational tools and learning opportunities are that can be provided to students. Additionally, we can take a closer look at what measures we are analyzing to gauge a students capability and readiness to succeed as a pilot in the USAF.

The most significant area technology now provides a margin for improvement is in advances in simulators. Emerging technology to include the integration of Extended Reality (XR) and Artificial Intelligence (AI) paired with Machine Learning (ML) hold significant potential throughout pilot training. The T-7 training systems could be developed to include these technologies in impactful ways.

Currently in pilot training, students have very limited access to simulators for practice. Each training base has a few Utility Training Devices (UTD’s) that have some capability for a student to interact with cockpit switches and include a single forward-view screen for practicing some basic flight scenarios. Other than these few devices, all other simulators are reserved for evaluated flight syllabus events as mentioned previously. Under the T-7 contract, the USAF has ordered a large amount of desktop trainers that somewhat replicate current UTD’s, giving students access to actual stick and throttle controls, touch screen controls, and a forward-looking view monitor. Moreover, these highly accessible lightweight and relatively inexpensive simulation devices could be paired with XR possibilities already being tested by the USAF at Randolph AFB, TX  as part of Pilot Training Next (PTN).

Recent studies show that VR training in aviation has emerged as an effective learning modality. The consulting firm Price Waterhouse Cooper released a study in June 2021 that showed learners using VR mastering content four times faster than in a classroom setting (AINONLINE, 2022).

VR training currently being conducted within the context of aviation for training purposes is promising. Ongoing and recent experimental training conducted by US military personnel is showing some positive and relevant insight into the world of virtual reality training for aviation training. The US Army through an “Aviator Training Next” (ATN) experiment (ARMY AVIATION, 2020) sought to determine if low-cost, low-fidelity commercial-off-the-shelf virtual reality (VR) technologies could be used effectively to train pilots, reported that during the initial phase of flight training for several classes with 296 students using VR devices (See data in the figure below), enough evidence was found which indicates that Aviator Training Next is a valid training methodology, and continued research is to be expected to better understand the adaptability of the technology into other areas of training (as cited in Flores et al. 2023).

Initial Entry Rotary Wing Training Performance Comparison, (ARMY AVIATION, 2020)

With the addition of a VR headset and the new T-7 desktop trainers a student could have access to a full visually immersive simulation suite mixed with physical stick and throttle and touch screen interface. This could replicate almost all aspects of flight that the much larger and significantly more expensive flight simulators can replicate. Having access to near realistic flight practice at any time provides students the ability to practice and prepare for their evaluated events and practice feedback from previous events.

The opportunites really expand when these desktop simulators and high-end simulators are now paired with AI and machine learning technologies to form a "smart-sim," a term personally developed. With technology that now exists, a student could have their own unique learner profile when they log into simulators. They could fly a profile driven by their own learning objectives or they could engage in an individualized training plan. AI can replicate all necessary entities that currently require a simulator instructor such as Air Traffic Control, other air traffic, weather, emergency procedure scenario injectections, and even a formation partner. AI can track an incredible amount of data points each sortie to include, eye movement and attention, aircraft handling, procedural accuracy, and all flight path and positional data points. This is data that would be impossible for an instructor to gather alone. Additionally, this data analysis can be paired with flight principles and best practices to provide the student with instant and objective feedback. For example, if a student struggles with altitude and airspeed control, the AI-incorporated smart-sim could identify immediately if the problem is an improper trim and power setting causing the student to fight the aircraft, or if the cause was a slow instrument cross-check. Many of these instructional fixes currently rely on best guesses from instructor’s perspectives and experience. AI inhanced simulators could also provide insights into the students performance to facilitate instructor-led debriefs and feedback. 

Training profiles can adapt to focus on skills more challenging to that student. If the student struggles with Instrument Flight Rule (IFR) approaches and adapting to bad weather conditions, the smart-sim could simulate poor weather on the students return-to-base, forcing the student to adapt and apply real world decision making and IFR procedures, building practice through repetition. This provides a non-retribution training option for students to practice challenging skills, and shifts the weight of focus from evaluation to rewarding the student’s determination to practice and improve. These smart-sims could also incorporate gamification of this training platform against peers or even instructors to incentivize logging as much time in training events as possible, and fostering teamwork and leadership skills in multi-participant virtual scenarios.

In 2023, Dincer provided an analsis of the current research on the influences of technology incorporation in aviation training, learner involvement and skill retention using immersive technologies, artificial intelligence, and game-based education. This analysis stated there was a clearly favorable link between technology incorporation and enhanced learning results, elevated learner enthusiasm, and superior knowledge retention (Dincer 2023). 

Furthermore, research done by Purdue University’s School of Aviation and Transportation Technology provided an analysis of research on artificial cognitive systems and extended reality. Their comprehensive analysis results “show that the use of XR technologies has the potential for enhancing learning and performance in safe flight instruction environments, a possible reduction in student pilot turnover… and an overall low cost for both flight training organizations and trainees due to the high levels of portability."

The capability of smart-sims would undoubtedly improve students’ readiness and airmanship through repetitions. Instructors could focus time on student performance analytics and see trends of students to better understand what specific training a student needs help with or access if the student is proficient and ready to advance to the next block of training. Actual flights could more appropriately be checkpoints of capability, building student confidence by allowing them to demonstrate to the instructor skills they have practiced and mastered, rather than stressful evaluations of skills they are just beginning to grasp with little-to-no prior exposure. Student access to easily available smart-sims could replace a significant amount of flight hours. This in turn further reduces demand on instructor pilots and aircraft, allowing for the reduction in flight hours and cost that AETC is seeking.

There are of course unique challenges associated with the implementation of these emerging technologies. First and foremost, would be the cost associated with the design and development of the AI and XR integrations into the T-7 program. Furthure research, and a cost benefit analysis would need to be examined carefully and weighed in on by USAF education leadership, operational flight instructors and pilot training instructors, acquisition specialists, and research and development teams. According to Laughlin (2008), some positive preliminary results suggest that training costs for new pilots could be reduced by as much as 70%.

There are those in the USAF pilot training community that feel that putting this much focus into simulators draws away from the flight portion of training, and that there are several aspects of flying that cannot be replicated in simulators. For example, the forces of gravity on the body while maneuvering, the physical requirements of flying a high-performance aircraft, and the real threat to one’s life cannot be replicated by a simulation. While this is true, the USAF is trying to reduce the number of flight hours required in pilot training. The reduced experience from a reduction in flight hours needs to be replaced with time and repetitions in simulators provide the best secondary way to gain this experience.

Additionally, there are legal and cybersecurity concerns with the database the AI system draws from and the data collection. “While users of AI (in aviation and other domains) may have access to domain-agnostic AI tools, the data upon which the underlying models are trained are often not fully known or understood and may have substantial drawbacks for domain-specific use (e.g., validity, bias, traceability, plagiarism/consent). There is also a risk that adversarial examples—inputs intentionally designed to disrupt an AI model – and other types of mis- or dis-information can cause inaccuracies in the processing and products of those models, raising consideration for how the cybersecurity of these systems will be managed and/or regulated (Nguyen et al. 2023). Finally, this technology is just now beginning to be adopted in the aviation industry and this would mark a new foundational approach for military aviation, driving a benefit for more empirical data on the learning curve, retention rates, and operational performance improvements attributed to AI/XR training methodologies.

(Grand View Research, 2023)

Due to the inherent challenges in aviation training and associated costs of flight hours, the aviation industry has and will continue to move towards emerging XR and AI technologies to reduce costs, and improve retention and skill of pilots. As depicted above, the market value for AI in Aviation grew from 3/4 of a billion to over 4 billion in 10 years, over a 530% growth rate (Grand View Research, 2023). Enablers such as machine learning, AI, predictive analytics, quantum computing, and other technologies have a powerful impact on the design, development, and implementation of XR technologies across a broad spectrum of contexts. The related implications for the aviation industry in terms of pilot training, require our full attention moving forward (Flores, Ziakkas, & Dillman (2023). I believe the USAF, with some additional research and commitment to advance our educational paradigm, is positioned to make a major transition into the future capabilities available through AI and XR integration, and the development of the T-7 aircraft and associated training systems provide the bridge for this transformation. My goal is to be a part of the team long term that insures these educational and technological advances are achieved, ensuring the highest quality of training for generations of USAF pilots to come.

1. AIN Online. (2022, June 1). Virtual Reality- The Future of Flight Training. https://www.ainonline.com/aviation-news/business-aviation/2022-06-01/virtual-reality-future-flight-training

2. AirForceTV. (2018, June 28). Air Force Tech Report: Pilot Training Next [Video]. YouTube. https://www.youtube.com/watch?v=PFl4t6PA7Z0

3. Armendariz, N., & Walcutt, J. J. (2023, May 22). Re-Engineering Aviation Training: Applying human-focused learning engineering processes to modernize training pathways, interventions, and use of simulation. Paper presented at MODSIM World 2023, Norfolk, VA. https://www.modsimworld.org/papers/2023/MODSIM_2023_paper_4234.pdf

4. Dapica, R., Hernández, A., & Peinado, F. (2022). Who trains the trainers? Gamification of flight instructor learning in evidence-based training scenarios. Entertainment Computing, 43, 100510. https://doi.org/10.1016/j.entcom.2022.100510

5. Dinçer, N. (2023). Elevating aviation education: A comprehensive examination of technology's role in modern flight training. Journal of Aviation, 7(2), 317-323.

6. Flores, A., Ziakkas, D., & Dillman, B. (2023). Artificial Cognitive Systems and Aviation training. In Tareq Ahram et al. (Eds.), Intelligent Human Systems Integration (IHSI 2023): Integrating People and Intelligent Systems (pp. 69). AHFE International. http://doi.org/10.54941/ahfe1002838

7. Grand View Research. (n.d.). Artificial Intelligence in Aviation Market Size & Share | Industry Analysis Report, 2023. Global Market Insights. https://www.gminsights.com/industry-analysis/artificial-intelligence-in-aviation-market

8. Laughlin, B. (2018). XR Drives Aerospace Excellence at Boeing. Manufacturing Engineering, 82. https://www.proquest.com/trade-journals/xr-drives-aerospace-excellence-at-boeing/docview/2164486002/se-2?accountid=13360

9. Maxwell, A. F. B., & AL AFB, A. M. (2021). Form from Function: Applying Flow Driven Experiential Learning to The Integration of Immersive Technology in Formal Military Aviation Training Programs.

10. Moore, P. J., Lehrer, H. R., & Telfer, R. A. (2001). Quality training and learning in aviation: Problems of alignment. Journal of Air Transportation WorldWide, 6(1), 5-11.

11. Mosier, K. L., & Fischer, U. M. (2010). Judgment and decision making by individuals and teams: issues, models, and applications. Reviews of Human Factors and Ergonomics, 6(1), 198–256. https://doi.org/10.1518/155723410X12849346788822

12. Nguyen, B., Sonnenfeld, N., Finkelstein, L., Alonso, A., Gomez, C., Duruaku, F., & Jentsch, F. (2023). Using AI Tools to Develop Training Materials for Aviation: Ethical, Technical, and Practical Concerns. Proceedings of the Human Factors and Ergonomics Society Annual Meeting, 67(1), 1343-1349. https://doi.org/10.1177/21695067231192904

13. Parrish, P., & Hoehn, J. R. (2022, November 29). U.S. Air Force Pilot Training Transformation. Congressional Research Service. https://crsreports.congress.gov/product/pdf/IF/IF12257/2

14. Tremosa, L. (2023, July 25). Beyond AR vs. VR: What is the Difference between AR vs. MR vs. VR vs. XR?. Interaction Design Foundation - IxDF. https://www.interaction-design.org/literature/article/beyond-ar-vs-vr-what-is-the-difference-between-ar-vs-mr-vs-vr-vs-xr