Military's Top Spy: AI May Enhance Tradecraft, Prevent Geopolitical Surprises

FacebookTwitterPinterestEmailEmailEmailShare
Defense Intelligence Agency Dir. Lt. Gen. Robert Ashley Jr., testifies before the Senate Armed Services Committee on Capitol Hill in Washington, Tuesday, March 6, 2018. (AP Photo/Pablo Martinez Monsivais)
Defense Intelligence Agency Dir. Lt. Gen. Robert Ashley Jr., testifies before the Senate Armed Services Committee on Capitol Hill in Washington, Tuesday, March 6, 2018. (AP Photo/Pablo Martinez Monsivais)

The Pentagon's top spy said Monday that he hopes advances in artificial intelligence can prevent defense analysts from being surprised when global conflicts ignite overnight.

"My core mission is to make sure that the secretary of defense is never surprised," Army Lt. Gen. Robert Ashley Jr., director of the Defense Intelligence Agency, told an audience at the Center for Strategic and International Studies.

Ashley, who has held the post for a year, was not on the job when Russian troops pushed into the Ukraine in 2014 -- a major destabilizing event in the region that the U.S. intelligence community has been criticized for not predicting.

"AI and machine learning will be a huge enhancement" to tradecraft and other skills defense analysts use to avoid blind spots in "the cacophony of all the information that is out there," he said.

Using algorithms to sort through massive amounts of information can take some of the burden off defense analysts, but it doesn't come without challenges, Ashley said.

"When an analyst sits in front of a senior leader, they always say, 'Based on reporting, based on sources, based on what I have seen I have a moderate [degree of confidence]' or if you see a national assessment that says 'I have a high-degree of confidence,' it goes back to sourcing and analytic tradecraft," he said.

"You never want to be in a position where you say, 'Well, the computer told me so,'" he continued. "Part of the challenge we have now, and I think really the opportunity is, as we look at algorithms, as we look at machine learning and AI, is developing a degree of confidence within the AI, a degree of confidence within the algorithm."

DIA will have to test these algorithms "to be able to prove that it can in fact come back with a high-degree of confidence that the analysis that it's doing is correct," Ashley added.

One of his goals is to ensure that the Machine-assisted Analytic Rapid-repository System, or MARS, is at initial operating capability before he leaves office in two years.

"To get that data environment set will probably be one of the biggest things we do for the Department of Defense," he said. "Over the course of the next two years, we are going to have an IOC beneficial to ... the [defense secretary, to the chairman [of the Joint Chiefs of Staff], to policy makers, to every warfighter. It gives them decision advantage and lets them move faster and understand what is happening in an operational environment. Our goal always is to make sure that we don't get surprised as a nation."

AI is just one of the tools that help analysts determine the difference between a Russian "snap" exercise or whether "they are coming over the border," Ashley said.

"It's not the AI on a thing to itself; it's to be able to sample more information and reduce the burden on the analyst," he said. "But at the end of the day, there is a human in the loop that -- based on training, understanding tradecraft, studying the problem -- makes a judgment ... to let a key leader know that 'I see the following and there is a couple of courses of action.' "

-- Matthew Cox can be reached at matthew.cox@military.com.

Story Continues