Like with anything new, the fact that AI/ML is largely an external development has not helped its cause. The military intelligence community was an early adopter of AI/ML. Faced with a continuous and insatiable demand for timely and accurate intelligence to close kill chains and to enable strike operations, AI/ML’s improvement of targeting provoked rapid change. This is why the Joint Artificial Intelligence Center’s (JAIC) origins are found in Project Maven within the Under Secretary of Defense for Intelligence and Security (USD-I&S). Locating the effort under USD-I&S produced the perception by operations professionals that Project Maven and AI/ML efforts are just an “intel thing,” a misperception contributing to broader lethargy in the services. It is not a coincidence that the current and previous JAIC Directors are intelligence officers. When an infantryman or pilot assumes the leadership of the JAIC you can be confident that real progress is afoot.
As in previous inflection points in U.S. military history, Congress seems increasingly willing to direct the military to move forward when it cannot move itself. The recently passed NDAA contains the most direct language to date on the military use of AI, including guidance to ensure its ethical use. The JAIC Director will now report directly to the Deputy Secretary of Defense, ostensibly raising JAIC visibility and influence within the Department. The Department will also report to Congress on the number of active-duty military personnel assigned to the JAIC, putting the services on notice that Congress is monitoring penetration of AI into areas where its application should be more tangible. Most importantly, Congress now requires the Department to provide five use cases of AI/ML application beyond autonomous weapons systems, in areas such as enterprise acquisition, personnel, audit, or financial management functions.
Despite these positive steps and JAIC Director, Marine Lieutenant General Michael S. Groen’s efforts to get the Department and services on board for broader applications, these actions do not yet address the prevailing data illiteracy. To get at this problem, Congress will have to change two military habits: First, Congress must close the education gap in this technical subject by directing changes to curriculum requirements for military officer accessions to produce a data-native officer corps. Believe it or not, in this twenty-first century, highly technical, security environment, there are services and occupational fields where there are no stated educational requirements beyond the acquisition of a bachelor’s degree. The Department must establish and adopt minimum standards for courses that require an appreciation for and analysis of data such as economics and statistics. This sounds very reasonable, but my own experience pursuing this change as the Marine Corps’ Manned-Unmanned Teaming team leader indicates there is significant resistance. Second, Congress will have to direct the targeted elimination of services and organizations where AI/ML applications in civilian organizations at corresponding scale are the current standard. Areas like human resources management and administration, and supply and logistics come quickly to mind. This, too, will be difficult since many have a natural vested interest in retaining a non-automated workforce. These personnel also have the least exposure to the potential of AI/ML transformation seen in other functional areas.
Despite these obstacles, Congress and the Department must rapidly move forward together. If they do, American ingenuity and the entrepreneurial spirit will close the gap and meet the stated goal of leveraging the opportunities AI presents in order to preserve the peace and provide security for future generations.
J. Darren Duke spent a 30-year career as a Marine Corps special operations, intelligence, and foreign area officer with operational and diplomatic tours in the Middle East, Africa, and the Indo-Pacific regions, and combat development experience in data science, automation integration, and policy development for reconnaissance and target acquisition applications.
The views expressed here are the author’s and do not represent the U.S. Government, the Department of Defense, the Department of the Navy, or the United States Marine Corps.