Defense Secretary James Mattis has a lot on his mind these days. North Korea, obviously. China’s expanding claims on the South China sea. Afghanistan, Iraq, Syria. And, closer to home, the Pentagon lagging behind the tech industry in leveraging artificial intelligence.
Mattis admitted to that concern Thursday during the Silicon Valley leg of a West Coast tour that includes visits to Amazon and Google. When WIRED asked Mattis if the US had ambitions to harness recent progress in AI for military purposes like those recently espoused by China, he said his department needed to do more with the technology.
“It’s got to be better integrated by the Department of Defense, because I see many of the greatest advances out here on the West Coast in private industry,” Mattis said.
Mattis, speaking in Mountain View, a stone’s throw from Google’s campus, hopes the tech industry will help the Pentagon catch up. He was visiting the Defense Innovation Unit Experimental, an organization within the DoD started by his predecessor Ashton Carter in 2015 to make it easier for smaller tech companies to partner with the Department of Defense and the military. DIUx has so far sunk $100 million into 45 contracts, including with companies developing small autonomous drones that could explore buildings during military raids, and a tooth-mounted headset and microphone.
Mattis said Thursday he wanted to see the organization increase the infusion of tech industry savvy into his department. “There’s no doubt in my mind DIUx will continue to exist; it will grow in its influence on the Department of Defense,” he said.
The Pentagon has a long record of researching and deploying artificial intelligence and automation technology. But AI is rapidly progressing, and the most significant developments have come out of the commercial and academic spheres.
Over the past five years, leading tech companies and their lavishly funded AI labs have sucked up ideas and talent from universities. They’re now in a race to spin up the best new products and experimental projects. Google, for example, has recently used machine learning research to power up its automatic translation and cut data-center cooling bills. Waymo, Alphabet’s autonomous-car company, uses AI in developing the technology in its self-driving vehicles.
Making smart use of artificial intelligence looks to be crucial to military advancement and dominance. Just last month, China’s State Council released a detailed strategy for artificial intelligence across the economy and in its military. China’s strategic interest in AI led DIUx to prepare an internal report this year suggesting scrutiny and restrictions on Chinese investment in Silicon Valley companies. Texas senior senator John Cornyn has proposed legislation that could enable that policy.
A recent Harvard report commissioned by the Office of the Director of National Intelligence found that AI-based technologies, like autonomous vehicles, are poised to make advance militaries much more powerful—and possibly cause a transformation similar in scale to the advent of nuclear weapons. But the US does not have a public, high-level national or defense strategy for artificial intelligence in the same way as China—perhaps owing mostly to differences of political style.
On Thursday, Mattis professed confidence that his department would figure out how to make more with AI, without offering specifics. “The bottom line is we’ll get better at integrating advances in AI that are being taken here in the Valley into the US military,” he said.
There is another bottom line to consider. The Trump administration’s proposed budget would increase funding for DIUx, which might help fulfill Mattis’ dreams of an AI acceleration. It also expands support to Pentagon research agency DARPA, which has many AI-related projects. But the White House’s budget proposal also includes cuts to the National Science Foundation, an agency that has long supported AI research, including work on artificial neural networks, the very technique that now has companies—and nations—suddenly so interested in the field’s potential.