The Defense Systems Information Analysis Center (DSIAC) was asked for a literature review of U.S. Department of Defense (DoD) publications that describe the employment or attempted employment of artificial intelligence (AI) or machine learning relevant to business mission areas (BMAs). DSIAC searched open-source documents and the Defense Technical Information Center’s repository for relevant documents and complied them into this report. There is a high priority across the DoD to employ AI across a variety of areas. Multiple branches have been experimenting with a variety of AI to aid in their BMAs. These areas include acquisition, logistics, human resources, installations energy and environment, training and readiness, and financial management. Although many of these programs are in their early stages, the DoD has an urgent need to hire data scientists and experts in AI and related fields to best utilize AI resources.
As the rise of artificial intelligence (AI) continues, the Department of Defense (DoD) is continuously looking for ways to employ AI in their business mission areas (BMAs). These areas include acquisition, logistics, human resources, installations energy and environment, training and readiness, and financial readiness.
The DoD established the Joint Artificial Intelligence Center (JAIC) in 2018 to guide the DoD’s AI Strategy, released in response to the White House Executive Order on Maintaining American Leadership in AI. Multiple systems are being tried and tested for use across the DoD and its BMAs.
2.1 United States Air Force (USAF)
The USAF is utilizing IBM’s Watson to assist with acquisition . They asked contractors to develop an AI-based system to help government employees and potential contractors navigate the acquisition process. The transparency and ease of use should make it easier for small businesses to bid on government contracts.
The USAF is also hoping to address the challenge of making contracting more accessible to internal and external customers by providing an intuitive and critical thinking system that understands the complexities of requirements and flexibility within the DoD contracting statutes, regulations, practices, and policies. The new system would be a benefit to program managers and contracting personnel, specifically for automated processes that support warfighting requirements, systems engineering, and risk management. The technology would also provide contracting officers the ability to swiftly answer difficult questions, allowing more time to effectively focus on creating agreements and other components of acquisitions. A small business innovative research solicitation for the system is divided into three phases: (1) developing a design that utilizes cognitive computing to provide users with answers to natural language questions about the contracting system, (2) evaluating the system, and (3) developing a ready-for-deployment application, along with several expansion support requirements. Applied Research Acoustics, LLC in Washington, DC, and KaScott Engineering in Lawrence, KS, were awarded the contract to create a natural language query system that leverages commercial sector advances in cognitive computing algorithms to provide users insights into defense contracting .
2.2 U.S. Army
The Army partnered with Sandia National Laboratories to create a system called the Capability Portfolio Analysis Tool (CPAT), which used sophisticated optimization software with useful visualization products to compare alternative investment strategies. They used CPAT to decide whether to invest in the ground combat vehicle to replace the Bradley Fighting Vehicle or invest in modernization of existing systems. Comparing the options, CPAT used 40 business rules, 70,000 constraints, 8,000 integer variables, and 2,000 binary variables. CPAT ultimately provided the Army with a solution that generated more combat power than the Army’s previous plan, even after factoring in a 25% budget cut. This resulted in the Army extending its use to other major programs .
3.1 Aircraft Maintenance
The USAF is starting to use AI for predictive aircraft maintenance by testing an AI-enabled approach that tailors maintenance schedules to the needs of individual aircraft . This approach is currently used by the F-35’s Autonomic Logistics Information System and extracts real-time data into a predictive algorithm to determine when technicians need to inspect the aircraft or replace parts.
3.2 Logistics Support Activity (LOGSA)
The Army’s LOGSA has contracted IBM’s Watson to develop tailored maintenance schedules for the Stryker fleet based on information pulled from the 17 sensors installed on each vehicle. LOGSA started a project in 2017 to use Watson to analyze shipping flows for repair parts distribution, attempting to determine the most time- and cost-efficient means to deliver supplies .
4.0 Human Resources
Beginning in the mid-1990s, the DoD, National Aeronautics and Space Administration, and other agencies used a system called Resumix to scan resumes for knowledge, skills, and abilities needed for a vacant job . Technology at the time did not exist to allow the system to be successful. The DoD did not find a wide market, and the system was discontinued.
4.2 Employee Monitoring
The Defense Security Service (DSS) Office believes a new AI-enabled pilot program will be helpful in evaluating which employees to trust . The program seeks to sift and apply massive amounts of data on people who hold or are seeking security clearances. The goal is to detect those who have betrayed their trust and predict which ones might. The program involves collecting an individual’s digital footprint, or cyberactivity, and then matching it with other data the DoD has on the person. A full snapshot of the person is hoped to be created during this process. Machine-learning algorithms are being used to derive insights.
The pilot will also provide a person’s Standard Form 86 information and other data not related to cyberactivity. This will join data from what is called continuous evaluation—a current effort to monitor life events related to clearance holders. This includes getting married or divorced, accumulating a lot of debt, tax returns, arrests, and sudden foreign travel. The goal is to allow the system to sense impending insider crime by detecting “micro changes” in a person’s behavior. The overarching goal is to have the system reveal changes in mode, outlook, or behavior of an employee before punishment becomes necessary.
4.3 Army Talent Alignment Program (ATAP)
The Army has been tweaking its personnel management systems as part of its People Strategy, released in October 2019, which outlines how they plan to acquire, train, and keep their talent . The Army recently assigned about 15,000 officers using an automated program as part of the ATAP, which seeks to better match unit needs with officers’ skills. This was the first time the board selection was conducted by combining ATAP with an automated job application portal, the Assignment Interactive Model 2.0. With AIM 2.0, officers can view all open positions and list their preferences and skills, while unit commanders can use it to have more interactive input with applicants to figure out who is best for the assignment. Army officials noted a 35% improvement in market participation, and more than half of officers and units received their top choice, where the unit and officer matched. The Army hopes that future iterations of the system will help define and curb nepotism and commanders hiring predominantly within their comfort zone.
5.0 Installations Energy and Environment
The Army plans to use smart technology to help defend against the increasing complexity of security threats to military installations . Ten technologies are currently being explored to innovate installations, including automated assessments of systems with limited manpower and monitoring utilities for anomalies in energy consumption. They are also looking to track fault detection, install smart thermostats, create autonomous vehicles, and explore frictionless entry. The Army Corps of Engineers is considering its Virtual Test Bed Installation, which involves AI that will analyze data of military installation facilities.
6.0 Training and Readiness
The Defense Acquisition University (DAU) is looking to move to an adaptive learning environment that responds in real time to student needs based on AI analytics measuring performance and activity. This will hopefully accelerate learning and reduce overall training time. The DAU will evaluate proposed solutions, with the intent to award one or more Other Transaction Authorities for prototype projects through the Training and Readiness Accelerator. Past AI experiments have included a 13-week adaptive, AI-powered learning pilot to test whether students can learn more efficiently, retain learning longer, and find value from an AI-enabled learning experience.
In December 2019, the Air Force launched an adaptive learning pilot, issuing tablets to 110 trainees with the 321st Training Squadron at Joint Base San Antonio-Lackland, TX, instead of hard-copy textbooks currently used for basic training. The solution includes Cerego’s online adaptive learning platform and Microsoft Surface Pro tablets .
7.0 Financial Readiness
The head of the Pentagon’s new AI capability believes AI could play a key role in helping the DoD shore up its accounting and make a clean audit more than just a dream . The DoD’s JAIC says the department has plenty of manual processes, like financial accounting, that could benefit from AI to drive greater back-office efficiencies by augmenting manual, human work. The JAIC has had very early discussions with the Pentagon’s Chief Management Officer and Chief Data Officer to predict what the department’s widespread adoption of robotic process automation might look like. The issues of deciding the scope of AI financial auditing are being explored. But the biggest challenge is obtaining the personnel who would not only understand it but embrace it.
 Hutchison, T. E. “Artificial Intelligence in Defense Acquisition.” Gravely Naval Research Group, Naval War College, Newport, RI, 29 May 2018.
 Araujo, D. “Cognitive Computers Primed to Change Air Force Acquisition Landscape.” https://www.military.com/daily-news/2015/08/05/cognitive-computers-primed-to-change-air-force-acquisition.html, accessed 16 April 2020.
 Hoadley, D. S., and K. M. Sayler. “Artificial Intelligence and National Security.” https://fas. org/sgp/crs/natsec/R45178.pdf, 21 November 2019.
 Neal, J. “From Human Resources to Helpful Robot—How AI May Transform HR.” https:// federalnewsnetwork.com/commentary/2019/08/from-human-resources-to-helpful-robot-how-ai-may-transform-hr/, 21 August 2019.
 Tucker, P. “The US Military Is Creating the Future of Employee Monitoring.” https://www. defenseone.com/technology/2019/03/us-military-creating-future-employee-monitoring/155824/, 26 March 2019.
 Williams, L. C. “The Army Turns to Automation for Officer Assignments.” https:// defensesystems.com/articles/2020/02/12/army-talent-management-automation.aspx, 11 February 2020.
 Lacdan, J. “Army Turns to Smart Technology to Defend Against Threats to Installations.” https://www.designworldonline.com/army-turns-to-smart-technology-to-defend-against-threats-to-installations/, 31 May 2019.
 Miller, S. “DoD Takes AI to School.” https://defensesystems.com/articles/2020/01/30/dod-ai-adaptive-learning.aspx, 30 January 2020.
 Mitchell, B. “Could AI Be the Key to DoD Completing a Clean Audit?” https://www. fedscoop.com/artificial-intelligence-key-dod-completing-clean-audit/, 13 May 2019.