DOD AI Deputy: AI is an ‘Enabler’ of Mission Areas

The Defense Department is expanding efforts to use artificial intelligence as a tool augmenting the human workforce to enhance nearly all mission and warfighter efficacy, the deputy AI official at the Pentagon said last week.

“AI is kind of like electricity. It’s not a specific thing. It’s an enabler of a bunch of different mission areas,” said Pentagon Deputy Chief Digital AI Officer Margaret Palmieri at the 2024 Billington Cybersecurity Summit.

Her office has been building the foundations of “AI electricity” by focusing efforts on hiring the best talent, finding the right compute and getting access to learning models in order to maintain a competitive advantage.

“How do we look through our current portfolio, pull across the capabilities that may be in the research and development phase now, but have to get into the hands of operators, have to be tested and evaluated and trained, and make sure that we’ve got good pipelines to do that?” Palmieri said. “We usually can’t predict the next crisis and we don’t really know what the exact need may be in the future.”

According to Palmieri, the department has over 1,000 AI applications using approximately $1.5 billion across the department.

Palmieri added that DOD is adhering to ethical AI principles and working across the national security ecosystem to coordinate AI compliance.

“What we in CDAO look for is, how do we enable those AI projects that are really integrated into mission areas across DOD, whether it’s on the business side or whether it’s on warfighting side,” said Palmieri. “How do we make sure that they’re done responsibly?”

She cited a governance meeting where her office “work[ed] with all the services and the different departments in DOD to talk about where everybody is on their efforts.”

“We were the first military [in the world] to sign ethical principles for the use of AI in 2020. We have a very robust, responsible AI toolkit and implementation pathway,” Palmieri said. “We take this very seriously and think about how you deploy these capabilities in a responsible way.”

The CDAO is also emphasizing what Palmieri described as AI scaffolding. DOD is investing in, she said, testing, evaluation, providing guidelines and data labeling and sharing, as to reduce barriers to AI adoption throughout the department.

Part of that is reducing friction in working with industry on department tech priorities. Palmieri said DOD has an acquisition “hierarchy of needs,” with AI at the top, analytics and metrics in the middle, and quality data at the bottom.

Working with industry through initiatives like the Small Business Innovations Research path, Tradewinds acquisition platform and Open Data and Applications Government-owned Interoperable Repositories, Palmieri said CDAO is examining how to implement technology like generative AI into DOD operations while maintaining security and resiliency.

“We want to leverage industry’s best, but we enforce technical architectures through contract language sometimes, and this is where we are. We’re going to buy the data platform separate from the applications. We’re going to make sure that data platform is open and that that data is available to all the providers that we need it to be, so that we can really start to unleash innovation at local levels and do that in a more effective way,” Palmieri said.