How AI can make use of DOD's data stockpile

The Pentagon is sitting on a hoard of data but doesn't know what to do with it. A new report from George Washington University's Center for Cyber and Homeland Security makes the case that artificial intelligence -- powered by cloud computing and big data analytics -- is the solution.

How bad is the problem? The GWU report estimates that "probably 99%-plus of the data that [DOD] collects is dark [and] never exploited. It just sits someplace, waiting for daylight."

The report, a series of issue briefs compiled from speaking panels on emerging technologies, doesn't attribute individual quotes to panelists, but includes input from George Duchak, deputy assistant secretary of defense for communications, command and control and cyber.

Specifically, the report argues that a September 2017 memo authorizing DOD officials to accelerate cloud adoption was done to lay the groundwork for a larger plan to unleash AI on the department's vast data repositories.

The cloud memo "was largely motivated for purposes of exploiting data, using and applying AI to a lot of our problems in defense," the report states.

How do the two technologies relate? By leveraging the cloud, AI algorithms can potentially exploit and process data at a previously unseen scale. While speaking at a public administration conference Nov. 16, John Kamensky, senior fellow at the IBM Center for the Business of Government, said that the same underlying incentives driving technologies like cloud computing and big data analytics are also driving the hype around AI.

"It is essentially the ability to take advantage of this massive amount of computing and storage computing power at our fingertips and the democratization of data itself," Kamensky said.

The report also calls good data -- specifically training data -- "the feedstock of AI." It argues that DOD's proprietary, non-public data is a previously untapped resource that could give it a leg up on others when it comes to training its AI systems.

"Our competitive advantage in DOD is the data we collect, with national technical means or otherwise," the brief continues. "This is data that is not in the public domain -- which gives us a competitive advantage in whatever AI algorithms that we have developed."

It's not the first time the idea of leveraging AI to put DOD data stockpiles to use has been entertained. In October, the department's CIO Essye Miller said at a military and technology conference that "we are hoarders" of data, and that there is "a natural inclination that we feel we have to keep everything." She also indicated that the department was looking at AI in some form to make use of its latent data.