A To-Do List for the Pentagon’s New AI Chief

It looks like the beginning of the end for the Defense Department’s Defense Digital Service, Chief Data Officer, and the Joint Artificial Intelligence Center. 

Deputy Secretary Kathleen Hicks’ office announced on Dec. 8 that all three organizations will be collapsed under a new position called the Chief Data and Artificial Intelligence Officer. The CDAO will report directly to DepSec Hicks. 

So what does this mean for the world’s largest military? 

Most immediately, this reshuffle likely means each office’s functions will be redistributed and, potentially, reimagined. The move also reduces the personnel and strategic deconfliction burden on the DepSec. Perhaps most importantly, it also reduces bureaucratic confusion on who “owns” the execution of the department’s artificial intelligence vision. 

Published mission statements, doctrine, and guidance on AI on data have perplexed DOD components and industry partners for years. One need look no further than the mission statements of the to-be reorganized entities. Buried in the DOD Data Strategy, the mission of the Chief Data Officer is described as “govern[ing] the Department’s data management efforts.” The JAIC’s website claims to have a “critical mass of expertise to help the Department harness the game-changing power of AI.” One cannot harness AI without the data to fuel it. How was the JAIC meeting their mission goals without the governing data authority granted to the Chief Data Officer? The Defense Digital Service claims to “work on high-impact technology projects that help the DoD.” One would presume the Service’s technical activities include AI projects and the data to them. These overlapping authorities and activities made it incredibly hard for mission owners to navigate who could help plan, enable, or execute any AI project. It’s no wonder that most DOD components and services decided to grow their own AI capabilities.

The creation of the CDAO also allows a new leader to rationalize and clarify doctrine around AI and data. Even though the Chief Data Officer was first to publish the DOD Data Strategy, the JAIC’s AI Ethics Framework does not acknowledge the Chief Data Officer’s directives to ethically collect and use data for algorithmic solutions. While the Data Strategy makes the Chief Data Officer responsible for all things related to data ethics, the implementation memo for the AI Ethical Framework vests all oversight authority to the JAIC. Despite this authority, nothing was done to communicate how to employ the framework. Instead, the Defense Innovation Unit took the initiative to show the Department a way to pragmatically execute ethical AI.

This consolidation renews hope that the department can reimagine and execute on its AI Strategy. While the concepts were reasonable in 2018, the strategy did not identify programs which could exploit the weaknesses of our pacing competitors. For example, China has been able to publish more research papers, file more patents, and grow its AI industry faster than the United States because it is exploiting our research and development system. As DOD funds more AI research, Beijing will continue to directly benefit from having their scientists jointly work with ours, learn from our mistakes, start companies in China based-on the most promising ideas from our best minds, file patents ahead of our scientists, and, thereby, accelerate their military AI programs. 

The CDAO should oppose China with a cross-department AI counterintelligence program, with robust partnerships across government, to strategically message that U.S. technology will not be exploited. China needs the United States; we do not need China.

The CDAO also has an opportunity to leverage China’s lack of faith in their own industry. As Ryan Sullivan points out, China worries about the power of a free-market economy and tightly controls the growth of its technology companies. While this helps them thread the needle of maintaining their government model and growing their economy, it crushes innovation. A strategic initiative to foster AI companies solving some of the DoD’s most ambitious problems, perhaps scaling the DIU’s use of Other Transaction Authorities, would directly confront a systemic weakness in the PRC. The proposed language in the 2022 National Defense Authorization Act (Section 865 and 867) requiring the service secretaries to pick the top five innovative R&D programs would also help move ideas out of the innovation “valley of death” into funded program lines.

Lastly, the CDAO needs to more meaningfully engage U.S. allies. We never go to war alone. The AI and data challenge is no exception. The JAIC had already started some meaningful exchanges with international partners in 2019. These efforts require expansion beyond the Five Eyes and should include ASEAN and NATO partners. Moreover, this engagement needs strategic direction to exploit the most vulnerable elements of our pacing competitors and amplifies the strengths of our allies.

Make no mistake, we are in a pitched battle for the future of civilization. Data is the ammunition and AI is the artillery. It is only a question of who can deepen the data magazine and who can discharge compute cycles faster. This is the Department’s chance to embrace this future and confront our enemies before it’s too late.

Brian Drake is the Chief Technology Officer of Accrete.AI Government. He is also the President and Chief Executive Officer of the Defense Intelligence Memorial Foundation. He recently left the Defense Intelligence Agency, where he was the agency’s first Director of Artificial Intelligence, drafted the agency’s AI strategy, and oversaw a more-than-$20 million AI investment portfolio.