Pentagon Needs a Faster Buying Process to Adopt Artificial Intelligence

The Defense Department’s top technologist on Tuesday said the agency will likely need to reform the acquisition process to develop and adopt emerging technologies like artificial intelligence.

“I do believe we will need changes [to the acquisition process],” Pentagon Chief Information Officer Dana Deasy told lawmakers. “I believe it’s going to be [a matter of] how do we move to a more startup mentality when moving to technologies like AI.”

The Pentagon’s artificial intelligence research relies largely on short-term cycles of experimentation and prototyping, which allow the department to rapidly test and develop new skills, Deasy explained to the House Armed Services subcommittee on Emerging Threats and Capabilities. Through this system, researchers can quickly discard failed projects and invest more resources in successful ones, he said.

Because the traditional acquisition process often calls for multi-year projections, he said, it would drastically hamper the department’s ability to innovate “at speed.”

The Pentagon sees AI as a critical aspect of national security in the years ahead, and it’s investing heavily in growing the technology through programs at the Defense Advanced Research Projects Agency and the Joint Artificial Intelligence Center.

Deasy said the JAIC, which will initially apply AI to equipment maintenance and disaster response, is officially “running and ready for business.”

In addition to iterative experimentation, Defense Deputy Undersecretary for Research and Engineering Lisa Porter told lawmakers close collaboration with academia and industry will also be necessary to staying ahead of rivals like China in the global race for AI dominance. While the government is best equipped to fund “high risk, high payoff” research, it’s critical for programs like the Defense Innovation Unit to adapt commercial tech for military use.

Beyond research, Deasy added the Pentagon’s proposed enterprise cloud will also prove critical to its artificial intelligence endeavors.

“AI needs a massive amount of compute power, a massive amount of storage,” he said. “I want to provide the Department of Defense with a way to handle that unlimited compute capacity, unlimited storage, on demand, as needed, with high integrity. The enterprise cloud is going to become the foundation [on] which all the data and all the compute power will reside.”