The National Geospatial-Intelligence Agency’s Maven program, an artificial-intelligence-powered image and video processing capability, is exploring new approaches to data and AI that could enable threat prediction, Defense One reported Tuesday.
Speaking on the sidelines of the Global SOF event, Vice Adm. Frank Whitworth, the NGA director and a four-time Wash100 recipient, said the next chapter of Maven would include new AI reasoning capabilities to detect potential threats, instead of merely identifying objects.
The effort requires creating and training new models “for the pieces of equipment or even behaviors that we want to prioritize,” Whitworth said, noting the role of the agency’s analysts in making models more accurate and confident in detecting threats. He expressed confidence in their ability to train the models; however, he acknowledged that some corroboration is needed to support foreseen threats before briefing a combatant commander or the president.
Computational Problem Hits Maven
NGA took over the Maven program in January 2023 after it was established in 2017 as the Pentagon’s flagship AI project to integrate AI into military workflows. According to the NGA director, the Maven user base rose fourfold within a year, reflecting increasing confidence of commanders across the services in AI.
In an August 2024 conference, Whitworth discussed the computational problem associated with the rising Maven users, saying that the program started to slow down since its 2017 compute level did not change.
At the Global SOF event, the official said the agency has been able to keep up with customer demand because of its continuous work on models to improve accuracy and efficiency and the foresight to acquire enough computing power that matches the demand.
