Some of the largest companies today are data companies such as Uber, Amazon, Google, and others, each playing a role in how we collect, organize, and analyze data. Nearly every industry is faced with emerging disruptive technologies including machine learning and artificial intelligence (AI) that are making a significant impact on the future. Military training is no different. As training needs and requirements evolve, leveraging data analytics for adaptive training is an appealing solution for boosting training effectiveness and improving readiness. Yet, beyond the hype of emerging data solutions, there needs to be an understanding for the challenges in integrating AI for training.
There is great potential in leveraging data-driven, advanced technologies to train students better, faster, safer, and more cost-effectively. As such, we have seen increases in military investments into training technologies such as simulated training, AR/VR/MR, and Live-Virtual-Constructive (LVC). According to Chris Shank, Director of Strategic Capabilities Office in the Pentagon, who spoke recently at the Autonomous Capabilities for DoD Summit, AI systems account for about 33 percent of DoD’s portfolio of projects. That said, leveraging more data-driven technologies to create a more adaptive training solution comes with its own challenges. Considering the relative novelty of leveraging data for training, we must keep in mind the following challenges that the military training industry faces:
Lack of requirements due the early stages of adoption
The military industry desperately needs training solutions that can induce and enhance critical thinking and situational awareness skills for the higher levels of the cognitive domain (analysis, synthesis, evaluation). Real-time information gathering, processing, and analysis are critical components to the individualized, flexible, dynamic, and adaptive smart training solutions needed today. However, market implications show that while the military industry knows that data-driven advanced training solutions are necessary, it doesn’t quite understand how to effectively define what it needs in terms of the requirements.
To overcome this challenge and provide improved training solutions that will meet the training needs of the military industry, companies need to work directly and proactively with the DoD to gain early insights into their challenges, objectives, and goals. Having this critical information can enable us to provide guidance and help the military industry define the most important requirements. For industry, academia, and small and large business, understanding the requirements is critical because it provides structure in understanding the most important areas for future internal research and development investments and, in turn, shapes the future of training.
Lack of data access
Data has been around for a long time, but the military industry has yet to utilize its full potential. Big data is not necessarily better than small data; what is important is the right data. The power is in combining datasets to draw meaningful conclusions based on correlations and patterns from various data sources.
The biggest hurdle when it comes to government’s data is access for industry partners. The lack of access to the right data can hinder the development of advanced training solutions that utilize machine-learning techniques. Because machine-learning algorithms require data to learn, these algorithms become “smarter” with increased access to the right data. DoD continues to collect and store great amounts of data, but the access to this data remains a challenge.
A formation of a collaborative agreement, specifically the Cooperative Research and Development Agreement (CRADA), between a government agency and a company or academia is one way to overcome this challenge. Through CRADAs, companies or academia can access personnel, services, facilities, equipment, and data for a joint research and development effort, but no funding from the government agency. This means that the vendors can potentially use the government data to train the algorithms. While the formation of a CRADA can be a lengthy process that takes months, it is well-worth it at times when a critical element like data access is needed.
Competition for resources and competency
Today, the scarcity of data science talent poses a challenge for the defense industry. The future of AI will likely not reside in a single algorithm, but rather a system of algorithms, where the right algorithm can be called at the right time. This system requires a team of data scientists, and already there is a shortage in the current talent pool. So, what happens when the availability of resources is limited?
Attracting the talent is difficult, specifically, when the pay structures within the DoD can be less attractive to the key talent compared to that of private organizations. However, the scarcity of data scientists is an issue across the board. Encouraging greater collaboration between DoD, large and small businesses (traditional and non-traditional), and academia to drive innovation and create advanced training solutions could address this challenge.
Addressing a cultural shift
Traditionally, military suppliers have built and delivered tangible products (hardware) to the military industry customers, for which the customers typically provided the requirements. Data-driven deliverables can differ in a major way, specifically because they are founded on an intangible source code – software. The realization that the deliverables are moving from hardware (product) to software (service) will drive a major cultural shift for both DoD and industry.
This shift will likely cause the need for the government’s procurement process to change. Procuring software or services requires a process that is more efficient to keep up with the rapid technological advancements. For example, an algorithm created today may not be sufficient six months later. In its current state, government’s procurement process is vastly bureaucratic and not designed to support the acquisition of data driven solutions efficiently. Additionally, this cultural shift is likely to affect the industry as well, driving it to re-evaluate and adjust its own business models, contracts, and procurement processes to remain relevant and competitive.
The Road Ahead.
While there are many challenges that lay ahead, military training is evolving and industry partners and defense agencies will need to collaborate to realize the full promise of AI for training to improve training effectiveness and readiness.
