The U.S. military will not prevail in highly-technical future wars without escaping the burdensome bureaucracy that marks the World War II-era acquisition system, according Undersecretary of Defense for Research and Engineering Michael Griffin.
Griffin, who has been remarkably candid about the Defense Department’s acquisition challenges since taking office as the inaugural ASD R&E, said the national security community should take a “cold-water bath” to shock it back to reality.
“In the world we have today, we can either maintain our process or we can maintain preeminence, but we probably can’t do both,” Griffin said April 9 at the New America Foundation’s Future of War conference. “We need to be able to move inside the decision loop of our adversaries and our adversaries are not burdened by the acquisition system we have that has grown up really since post-World War II days.”
The current acquisition process, slowed by layer upon layer of bureaucratic process, was successful at deliberately designing weapons that have helped the U.S. maintain its global military superiority since 1945. Now that U.S. supremacy is being challenged by adversaries like Russia and China, which are not beholden to the fairness and even hand written into U.S. acquisition rules, the old system is a burden, Griffin said.
“Our acquisition system is built for a period of time in which, first of all, American preeminence was not really questioned,” he said. “We had the luxury of time to make decisions as if others couldn’t catch up. Now we know – we should know – that they can. We can either devote ourselves to the maintenance of the structure that we have or we can devote ourselves to remaining on top. That’s the choice we face.”
China and Russia are already catching up with, if they have not drawn ahead of, the United States in key technological endeavors like artificial intelligence, machine learning and hypersonics, all of which are named in the new National Defense Strategy as critical investments the U.S. needs to make to retain its military overmatch.
“There might be an artificial intelligence arms race, but we’re not yet in it,” Griffin said. “Our adversaries – and they are our adversaries – understand very well the possible future utility of machine learning. I think it’s time we did, as well.”
AI will be key to defending U.S. military personnel and platforms from emerging threats like drone swarms and proliferating guided missiles, Griffin said. Because of the volume of relatively low-cost weapons an enemy could send or shoot at U.S. ships or ground forces, machines will be needed to detect, identify, track and destroy them, he said. Machines also can more quickly and accurately prescribe countermeasures or defensive weapons for a particular threat.
Directed-energy weapons, hypersonics and micro-electronics manufacturing are other areas Griffin sees the United States ceding ground to potential adversaries.
“China now has a significantly advanced offensive hypersonic capability that the United States needs to have and does not yet,” he said.
Northrop Grumman Chief Technology Officer Patrick Antkowiak said neural nets and other AI technologies that emerged in the 1990s, but overpromised their utility and maturity, are becoming viable through commercial investment. The government must make a conscious decision to embrace those advancements and invest in further research and development to integrate AI and machine learning into weapon systems, he said.
“You can choose to lead in the application of artificial intelligence and/or machine learning into the defense space,” Antkowiak said at the Future of War conference. “There’s a whole stack of technologies that exist, all the way from how to organize data, how to curate large sets of data to the fundamental hardware resource that allows you to either train or do inference in real time. All of these things contribute to the possibility of taking AI and integrate it in our weapon systems’ future. That’s where we’ve got to make a decision.”
“Machines can do certain things much better than humans can and if we can get enough of that reasoning in the machine stack, I think we can make our job, in terms of the Defense Department mission, much simpler,” he added.