NGA Embraces AI to Help Manage Increase in Data, NGA Director Whitworth Says

The National Geospatial Intelligence Agency (NGA) is on a growth spurt, specifically around developing workforce and incorporating artificial intelligence for more data gathering and assistance in managing that data. That was the message from Vice Admiral Frank Whitworth, the director of National Geospatial Intelligence Agency (NGA) during his keynote speech Monday at SATELLITE.

Whitworth talked about resources available for the NGA and what needs to be added. “Structured observations don’t come cheaply, especially when you start generating the type of inferences and number of detections that we do,” he said. “We move more ones and zeros than any other agency in the United States. A lot of Americans don't realize that, and accordingly, with the size of the constellation growing and the number of terabytes coming from space over the next eight years, that data collection will go up precipitously.”

Artificial intelligence has been a key focus of the agency going forward. “In the past, we've used the A in NGA to go with a theme like acceleration or action. This year we're talking about a true acceleration of AI,” he said.

Whitworth outlined five goals for developing AI, including task organization of the leadership. “We're talking about an entity that is not necessarily human, but can be as productive, in some cases, as a human,” he said. “So while we won't necessarily have a chief of AI development like you have a chief of human development, we're getting close to figuring out how to structure that.”

There is automated collection occurring thanks to some of the AI models, he said, from automation that was put in place just last year. “We're keeping custodies of some things that we didn't use to keep custody of, and that's a very powerful thing.”

Whitworth said that they also get pixels from international partners. “That is something that we at the NGA take very seriously,” he said. “I don't talk about those numbers, but trust me, it's impressive.”

The commercial satellite sector is another significant source that the NGA taps into. “The coverage that commercial satellites allow is so important right now,” he said. “I'm talking to a lot of people who represent the commercial sector. I want additional collections and additional commercial pixels. I want analytics with commercial. I want it all.”

With all that is needed and all that is being developed, NGA’s decentralized approach has been very responsive, he said. “There are about 9,000 people in the agency now, and about 25 percent of our force is not in any particular headquarters.”

The NGA is about 3 percent military, and about 55 percent government civilian, with the remainder being contractors. “Everyone pulls their weight in their particular areas of expertise, and pound for pound, I'm really proud of the output and the productivity that our team has been able to generate,” he said.

What the NGA needs now is accreditation and certification for the workforce to work with AI. “A lot of people might assume that people find it threatening, or they're fearful that the future involves models instead of people,” he said. “I think it's going to be more of a team approach. Don't forget, these are some of the best people we have in the world at generating positive identification, generating warning, generating targeting. We need to keep them in a place of constantly training through machine learning, structured observations, and data labeling.”

What keeps him up at night is not the enemy, Whitworth said. “It's just the responsibility of not missing something. We don't want to miss some sort of an exploitation that would help us generate positive identification and targeting. And we definitely don't want to miss any detail that would help us render better charts and better maps.” VS