William Roper says the military must get better at feeding the voracious learning algorithms that will fight future battles.
The first day of the next major conflict shouldn’t look like war at all, according to William Roper, who runs the Pentagon’s Strategic Capabilities Office, or SCO. Instead, imagine a sort of digital collection blitzkrieg, with data-gathering software and sensors setting of alarms left and right as they vacuum up info for a massive artificial intelligence. Whoever collects the most data on Day One just might win the war before a single shot is fired.
“My prediction for the future is that when we go out to fly planes on the first day of battle, whether they are manned or unmanned, that the purpose of the first day or the second day will not be to go out and destroy enemy aircraft or other systems. It’s to go out, collect data, do data reconnaissance, so that our learning system gets smarter than [the enemy’s],” Roper said Tuesday at an Air Force Association event. “Every day you fly, you get that exponential increase in sophistication.”
As head of SCO, Roper helps the services turn existing technologies and weapons into surprising new capabilities... fast. He is the military’s foremost go-to guy to figure out how to use advances in technology to secure military advantage and how the enemy might do the same. (Think, perhaps, of the title character in 1964’s Stanley Kubrick film "Dr. Strangelove.") Roper says the United States’ most important and overlooked asset is probably the digital information it produces in terrific abundance. Every second, data is pouring out of planes, satellites and sensors related to targeting, machine performance, mission success, intelligence and more. The military, he says, treats that data cheaply.
“We’re still focusing on data in a 1990s way: something you use to go into a fight and win and after that, the data, its raison d’être is over,” he says. The Air Force, in particular, is a “service that generates a lot of data and that goes into mission planning.”
That’s not how the commercial world values data, and particularly not companies like Facebook and Google. Many Silicon Valley companies are charging ahead with deep learning to create new products, services, and ways of outmaneuvering competitors.
“To them that data is…oil,” Roper said. “It’s wealth and fuel. Your data keeps working for you. You stockpile the most data that you can and train that to teach and train autonomous systems.”
Roper hopes to spread this notion throughout the Pentagon.
“Almost nowhere do I see a technology that’s current that offers as much as autonomy,” he said. “We’re working very hard to produce a learning system.”
He said the military had several “pathfinder” initiatives underway but could not go into detail.
Sooner rather than later, Roper envisions, American forces will train to go to war with highly networked (and largely disposable) drones and missiles, which would be in constant communication with one another, learning collectively, constantly and in real time to take out targets and adapt new defensive strategies. Those connections—between every machine on the battlefield, with humans, with deep learning programs that crunch every available piece of information about enemy defenses—will result in a crushing Day Three for the enemy. At least that’s the hope.
Key to that strategy: drones and weapons that can improvise (a bit) on their own while remain beholden to a human operator. Roper pointed to the Pentagon’s Perdix mini-drone swarm program.
“You have no idea where the [drones] are going to go," he said. "That’s awesome, because the pilot doesn’t have to joystick each one of them.”
Roper, like every Defense Department leader we’ve talked to about autonomy and weapons, was careful to emphasize the Pentagon is not looking to replace human decision-making. He echoed Undersecretary Bob Work, who has made collaboration with AIs as central aspect of the Pentagon’s technology strategy. He emphasized the role of the human operator as the “quarterback” in operations.
“It’s kind of a technology we will have to keep our eyes on,” he said.
Roper described his “biggest fear” as “letting perfect, exquisite, government-only solutions be the pacing function for our military.”
That fear is one Roper shared with former Defense Secretary Ash Carter, who championed efforts like the DIUx outreach program, to bypass bureaucratic acquisition cycles that move much, much slower than technology.
Carter’s successor, James Mattis, has so far been coy about his own research and development priorities. That may come out in the weeks ahead in the form of a budget.