Autonomy Should Be Fully Integrated, Carefully Limited

Top Defense Department leaders addressed the role of autonomous systems in US military strategy Friday, saying integrating autonomy with existing systems is critical but must be accomplished with strict attention to the risks involved. Deputy Secretary of Defense Robert Work said the challenge is “injecting AI [artificial intelligence] into the grid,” with a focus on “human-machine collaboration.” Work mentioned the use of “exoskeletons,” “wearable sensors,” “directed energy,” and “hypersonics” as ways of using machines to enhance human decision-making that would be necessary to maintain a competitive edge given the challenges the US military faces today. He offered the example of a “machine learning” exercise in which a computer was able to retrace the missile launch that brought down Malaysia Airlines Flight 17 in Ukraine on July 17, 2014, relying entirely on data mined from social media accounts. Twitter and other services provided detailed images, from the transport of the Buk missile system across the Russian border to the contrails of the missile itself. Work suggested that such “information dominance” could be harnessed to “allow us new indication and warning” capabilities. Vice Chairman of the Joint Chiefs of Staff Gen. Paul Selva agreed, saying the future of autonomy in the military is “AI and algorithms helping us do what we do better.” But he added that a “fleet of autonomous single things” would be a static failure. In order to turn autonomy into a competitive advantage, Selva insisted, it must be linked to a global network of military capabilities. Finally, Secretary of Defense Ash Carter articulated a clear doctrine on the use of autonomous systems within the military. “We will always have a human being in charge of the decision about the use of force,” he said. (See also: Shifting Futures, which will appear in the November issue of Air Force Magazine.)