What’s Ahead for AI in National Security? Insights from Booz Allen’s CEO and Panel

Oct 25, 2018 2:05 PM ET
Multimedia with summary

As artificial intelligence (AI) offers capabilities we could not have imagined, what are the promises and pitfalls for national security leaders? Which challenges can be addressed with AI, and which will require human management?

On October 3 at The Atlantic Festival, Booz Allen Senior Vice President and Artificial Intelligence Lead Josh Sullivan moderated a discussion with Susan Kalweit, Director of Analysis the Directorate, National Geospatial-Intelligence Agency, and Derek Strausbaugh, Federal Chief Digital Lead for the Department of Defense, Microsoft. Later in the morning, Booz Allen CEO Horacio Rozanski spoke about AI with Eric Schmitt, Senior Correspondent at the New York Times.

The summaries of the questions and conversations are below.

How is AI changing the game in national security right now?

According to Strausbaugh, “You don’t have to read through 4,500 documents to find the needle in the haystack. We’re narrowing the aperture for analysts, operators, and decision-makers to have the information in front of them that they really want.”

Kalweit cited efficiency and effectiveness. She talked about automation and change detection tools freeing up analysts’ time for more complex issues, plus the power of AI to deliver structured observations that can map to other data streams—e.g. for a photo of a crowd, identifying the location, finding information about the crowd on social media, then connecting the dots to describe the activity as a possible protest.

What are the biggest challenges and concerns?

Kalweit expressed concern about scalability—having the technology ready in time to support customers “from the warfighter to the White House.”

Strausbaugh said he worries that the AI hype will hinder execution and adoption. To truly embrace this technology, “There’s a lot of modernization that needs to happen and a lot of legacy hardware and applications we need to bridge. Some of the cultural and procedural things we have in place right now need to be reformed to allow the technology to rise.”

Both panelists cited the urgent need for education and building the AI workforce with emerging talent and experienced professionals.

What do you hope to see in AI in the next five years?

Strausbaugh is working on making AI “portable” for field missions with limited or no connectivity and talked about applying AI technologies and algorithms to humanitarian and disaster response—for instance, identifying impassable roads or people waiting on rooftops for rescue.

Kalwait shared how NGA is aiming to turn 75 percent of the agency’s image analysis to her “Triple A” (Automation, Augmentation, and Artificial Intelligence” program. “We’ve had successful pilots that are actively supporting operations, and now we’re moving it to enterprise scale.”

How do we make it all work from a technology and human standpoint?

Rozanski emphasized the importance of trust to the AI adoption curve. “The key question is not about what the technology can do, but what we, as a society, trust it to do.”

He suggested a national strategy where government and the private sector work together to ensure that AI will to work as intended and that we will be able to determine what went wrong when it does not. “With AI, we have an opportunity as a society to minimize unintended consequences.”

Watch a recording of this and other sessions, here.