By Clay Dillow
/ Published September 15, 2015
IBM Watson's computer housing case in New York.
Photograph by Andrew Spear — The Washington Post/Getty Images
The military service is building its own automation system to combat complex bureaucracy.
As part of the U.S. federal government’s most sprawling bureaucracies, the Air Force knows a thing or two about red tape and complex bureaucratic processes.
So, perhaps it’s appropriate that in its effort to combat unnecessary red tape, the USAF isn’t taking any half measures. In order to smooth out its acquisition process and review new technologies and equipment more efficiently, the Air Force is building a state-of-the-art “cognitive computer” specifically designed to analyze large amounts of data.
In other words, the Pentagon is building a cutting-edge supercomputer to battle its own bureaucracy.
“In general, when we talk about where the Air Force is today versus 20 to 30 years ago, people say it takes so much longer to get things done,” Dr. Camron Gorguinpour, director of the USAF’s office of Transformational Innovation, says. “People ask why it takes so much longer. And the answer is, typically, complexity.”
Cognitive computers function more like the human brain, making them capable of tasks that are very difficult for regular computing machines. While they are not artificially intelligent in the sci-fi sense, they do possess the ability to think critically, learn from experience, and comprehend context. And critically, cognitive computers are able to interface with humans through natural language. In other words, they speak English.
Readers who recall IBM’s supercomputer from a 2011 episode of Jeopardy! are already somewhat familiar with how this computing paradigm works. Cognitive computers can respond to queries that don’t fit neatly into a pre-programmed set of questions and responses. That’s key for navigating something like the USAF acquisitions process, Gorguinpour says, especially for small business that don’t have the resources to field a team of acquisition experts.
Using the new tool, businesses will be able to enter a question into a query bar, much like conducting a Google search. However, instead of being directed to a dense and lengthy list of acquisitions requirements or links to various opaque Pentagon documents, users would get a simple answer.
The bureaucracy-slaying cognitive machine marks the first technology initiative from the Air Force’s young Transformational Innovation Office in its attempt to reduce complexity and ultimately trim cost.
The office—about one year old—was created under the USAF acquisitions office specifically to bend the cost curve within the USAF acquisitions process. Last fall, some officers within the Air Force approached Gorguinpour’s office curious about using IBM’s Watson for Air Force applications.
In a demo conducted with IBM shortly thereafter, the Air Force uploaded its reams of acquisition regulations and put the software through its paces. The demonstration showed promise, Gorguinpour says. Natural language questions tended to produce coherent natural language answers.
On July 31 the Air Force awarded two companies—Applied Research in Acoustics, LLC, in Washington, D.C., and KalScott Engineering in Lawrence, Kansas—to take the initiative to the next level. The companies will better train Watson in acquisition regulations to sharpen its responses and improve the user interface so that it’s easier for the public to use. The initial nine-month effort is already underway, with delivery of a first product slated for April of next year.
Ultimately the Air Force is trying to make it easier for the contractors it already does business with, and enable companies that don’t typically interact with the federal government to engage in the USAF acquisitions process as well, Gorguinpour says. Doing so will increase the Air Force’s options, increase competition for contracts, and help smaller businesses with smart technology bring new solutions to the Air Force.
“This is part of a bigger context of work that we have underway,” Gorguinpour says. “The idea is to bend the cost-curve, and we’ve got 20-something more things in progress, pushing on multiple fronts.”
Bringing the power of cognitive computing to bear on the problem of bureaucratic inefficiency could pay dividends across the service branches as well as for the contractors that do business with the USAF. And the real transformational insights may come as much from those people as from the cognitive computer itself. “We’re going to learn a lot from developing it at first,” Gorguinpour says. “And once we release it in beta we’ll learn a lot a more.”