Somewhere around half of all ongoing projects at the Department of Defense’s (DoD) Defense Advanced Research Projects Agency (DARPA) are currently utilizing AI in some form, DARPA Deputy Director Dr. Peter Highnam said March 24.
The programs have all flowed through DARPA’s $2 billion AI Next program and have built from there. Highnam shared those insights, as well as some details on a few of those projects, at the National Defense Industrial Association’s National Security AI Convention on March 24.
“In September of 2018, we announced the AI Next campaign, which was about $400 million a year of AI technologies into AI work … split into three classes: applying AI, advancing the state of the art of AI, and then exploring ahead,” Highnam said at the morning keynote.
The program has been a success and DARPA has begun to naturally build AI into these programs. Highnam said that other than those programs specifically made for advancing AI, most of the intersection has come because of the process rather than programs looking to use AI for the sake of using AI.
“It was organic,” Highnam said. “It was … the way that worked best. So, teams out there – research teams in universities, in companies large and small, in the labs – move to those techniques because they work and they work best.”
He went on to tout various unclassified DARPA programs that are in the process of utilizing AI, including those that allow relatively untrained pilots to help “guide” helicopters, programs to aid in space exploration, and even one that could be used to restart the power grid in the face of an ongoing cyber attack. The latter is especially prescient in light of a recent GAO report that concluded the United States power distribution grid is increasingly vulnerable to cyber attacks.
“There are so many different ways to invest in AI,” Highnam said. “Or again … what could be done [with those] capabilities, what might be possible, [and] what needs to be done to make these systems robust and dependable, and so on.”