Why DoD’s New Approach to Data and Artificial Intelligence Should Enhance National Defense
The ability of the United States to compete in the 21st century depends on U.S. leadership in data and artificial intelligence (AI). In response, the Department of Defense (DoD) is taking a new and much-needed approach to U.S. defense efforts in data and AI. David Spirk, the departing Chief Data Officer of the Pentagon, made clear yesterday that the office of the Chief Digital and AI Officer (CDAO), in addition to its other functions, will be the successor organization for and replace DoD’s much-touted Joint Artificial Intelligence Center (JAIC). While the JAIC symbolized DoD’s efforts to get smart on AI beginning in 2018, the integration of data and AI represents a maturation of the U.S. AI approach—one that elevates the importance of AI in national defense. The JAIC itself was not as important as what the JAIC stood for—DoD’s commitment to U.S. defense AI leadership. In paving the way forward and getting AI on the agenda, the JAIC succeeded. From this point on, a more cohesive approach to AI and data through the CDAO is more likely to accelerate AI adoption throughout the U.S. military because it links DoD’s AI efforts with data, the fuel AI requires. For U.S. defense AI adoption, in particular, aligning these organizations could be game-changing. Addressing DoD’s siloed data, standardizing and improving its quality and access, is a precondition to having the data necessary to train algorithms for many defense uses, and any future technologies that rely on collecting, processing, and using information. Implementation will be critical and heavily dependent on two things. First, to catalyze AI adoption, the CDAO will need to develop close relationships with the military services and combatant commands. Second, the CDAO will need to coordinate with DoD’s research and development organizations, such as the Defense Innovation Unit, leading on AI experimentation and research. There is hard work ahead, but the new organizational design is promising.
The office of the CDAO brings together previously independent components of DoD: the JAIC, the office of the Chief Data Officer, the Defense Digital Service (DDS), and the Office of Advancing Analytics (ADVANA). The office of the Chief Data Officer is in charge of data management and coordination, DDS finds digital solutions for internal data and security issues, and ADVANA aggregates data and conducts data analytics. The combination of these offices raised questions about whether an independent JAIC was necessary for U.S. defense AI leadership. Departing CDO Spirk says that the CDAO will be “taking the best parts of all the organizations it is overseeing and redistributing them for faster and better decision-making.” We agree. At present, not only is DoD’s data siloed but its AI efforts and initiatives are as well. According to the company Govini, in FY21, fifteen separate departments and organizations funded and worked on AI and AI-adjacent technologies, often without formal coordination or throughlines. This has led to redundancies, gaps, inconsistencies in application and access to data and resources, and an overall hodge-podge of AI efforts. DoD has acknowledged this and is making organizational changes necessary to accelerate AI adoption even more by restructuring its AI approach from the ground up. Now, CDAO will have teams working on policy and governance, technology development, and rolling out data and AI for the Pentagon and the military services, to avoid bureaucratic duplication and confusion that could undermine the CDAO’s overall authority. In particular, bringing the data and AI teams together will improve the data DoD needs for AI development.
Robots and Artificial Intelligence
Some might fear that the reorganization of the JAIC’s functions within the CDAO means the United States is not as committed to the role of AI in the future of U.S. national defense. Based on current information, this concern is misplaced. First, the JAIC was created so the U.S. military could effectively take advantage of the way AI will shape the future of war. It succeeded in many ways. Recognition of the importance of AI for the future of U.S. defense, and national security in general, is much more widespread. The JAIC made headway on AI adoption and data literacy, with initiatives like “AI 101,” and on the data integration issue, as part of the Artificial Intelligence and Data Initiative (AIDA). The military services are investing more in AI and related technologies such as autonomous systems. This, ironically, makes an independent JAIC less necessary.
Second, the JAIC has also faced challenges that the CDAO approach can address. The JAIC had multiple missions, including advising DoD on AI adoption, funding AI research, and building AI tools itself. The JAIC also lacked the authority to advance military service adoption of AI on its own, or to itself transform the connection between AI and overall DoD policy and strategy. The JAIC ended up arguably not policy-focused enough to lead on policy, and not technically equipped enough to lead on algorithm development. While the JAIC encouraged AI investment within DoD, its existence also highlighted how the uncoordinated DoD AI portfolio required even more organization. The CDAO approach will address some of these issues by fusing DoD data and AI efforts, as will a growing focus on AI in other DoD components, from Research & Engineering on the technology development side to OSD-Policy (Office of the Secretary of Defense Policy) on the strategy and governance side.
Third, it is difficult to get things immediately right when it comes to converting emerging technologies into adopted innovations, especially for conservative institutions like militaries. We think about experimentation as a critical part of how the technology invention process works, but the same is true when it comes to transforming organizations. Given the way data access and integration are essential to innovation, consolidating data and AI, rather than having a specific JAIC only focused on AI, will make technological adoption across DoD more likely.
While it will hopefully spur AI forward within the department, subsuming the JAIC into the CDAO does come with some risks. Currently, there is a widespread understanding that AI is essential for U.S. success in strategic competition and defense leadership. However, we may be taking the prioritization of AI for granted, and future DoD leaders might have a different perspective even if the capabilities of AI tools continue to mature and advance. If that were to happen, the absence of an independent JAIC could lead to a withering of focus on AI, and a downplaying of its importance and relevance, just at the key moment advances in algorithms become more relevant for many military functions.
Reward requires risk, especially when it comes to innovation with emerging technologies like artificial intelligence. Technology development is not a linear process and often involves failure along the way. Innovation becomes even more challenging when it requires organizational change to facilitate adoption. Saying goodbye to the JAIC will be bittersweet—the JAIC played a critical role in advancing the U.S. military’s emphasis on AI and set the table for what is next. Moving forward, however, bringing the JAIC into the CDAO will create a more integrated approach to AI and data that is likely to help the United States achieve defense AI leadership.
Robots and Artificial Intelligence