something without the aid of AI and the time to do so with AI’s help. If the AI correctly produces the outputs of a given process that, for example, an operator formerly produced, it has captured the knowledge of that op- erator in the AI program. Consider the task of producing an executive summary based on a meeting transcript. Someone would need to be trained to understand the role and format of an executive sum- mary and likely practice determining the right level of detail for the audi- ence. It may take less time for that same person to learn how to write a prompt to ask a large language model (LLM) to write an executive summary based on the transcript. However, the LLM may not have the context to know what is important to the audi- ence and to perform this seemingly mundane process.
It follows that, in this case, the re- turn in using AI may not be worth the investment. In other cases, perhaps with a more complex task like anom- aly detection, AI may provide a more significant advantage by combining the knowledge of many operators into a single, fast output. KVA provides an objective, supportable measure of value to reach that decision. Conclusion Optimistically, AI is a powerful tool that may improve the precision, ac- curacy, capability, and power of our systems. But we should not mistake promise for results. The task of quan- tifying returns for our investments— whether in capability, readiness, pro- ductivity, pleasure, or pain—remains a challenge. A deliberate, extensible, and repeatable approach, such as KVA, will ensure that our infrastruc- ture—not only our systems of mea-
surement and accounting, but also testing, training, and practice—keeps pace with the technology we procure for our soldiers. The next steps will include further pilots for KVA or similar complexity- based techniques to provide ready examples for future analyses. Unde- niably, there remains the challenge of formalizing a process to enable pre- cise complexity measurement and ensuring its consistency across the DoW. But we have a blueprint. HAWKINS is a computer scientist with U.S. Army Manager for Command and Control Data and Artificial Intelligence. He holds an M.S. in Library and Information Science from Drexel University and is currently a student in the In- formation Sciences Ph.D. program at the Na- val Postgraduate School (NPS). His published work includes research papers and editorials focused on the intersection between people and technology. HOUSEL, a tenured professor of Information Sciences at NPS, specializes in valuing intel- lectual capital, knowledge management, and IT. With a Ph.D. from the University of Utah, his research, published in journals such as MIS Quarterly , also includes behavioral account- ing. He was previously an associate profes- sor at the University of Southern California and Chief of Consumer Market Research for Telecom Italia, and his value research was featured in a Fortune magazine article. The author can be contacted at jeffrey.t.hawkins10@army.mil . The views expressed in this article are those of the authors and do not necessarily reflect the official policy of the Department of the Army, Department of War, or the U.S. Government. Reproduction or re- posting of articles from Defense Acquisition maga- zine should credit the authors and the magazine.
Related Resources • Innovate to Win Playbook (Online Tool) • EXE 4050V Leading Change to Drive Innovative Culture (Virtual Instructor-Led Training) • Scaling Innovation (Video) • Driving Innovation in Acquisition (Web Event Series)
16 | DEFENSE ACQUISITION | November-December 2025
Made with FlippingBook - Online Brochure Maker