Last Friday the Bulletin of the Atomic Scientists posted my analysis of US policy for autonomous weapon systems (AWS), in which I demonstrate that, far from being a moratorium, the policy is actually highly aggressive. It overrides longstanding resistance within the military, establishes a framework for managing legal, ethical and technical concerns, and signals to developers and vendors that the Pentagon is serious about autonomous weapons. Beyond this, I would argue that the United States is clearly leading the world in developing and utilizing the technology of autonomous warfare.
As if in rebuttal, on Monday morning John Markoff of the New York Times posted an article seemingly asserting the opposite: according to Markoff, the US military “lags” in its pursuit of robotic ground vehicles. Lags… behind whom? China? North Korea? No, Markoff warns that the Pentagon is falling behind another aspiring superpower: Google. With its “small fleet of vehicles with more than a half-million miles of automatic driving,” the company that would not be evil has left the military in the dust with “virtually no chance of meeting the goal set by Congress to have a third of the military’s combat fleet consist of unmanned vehicles by 2015.” Meanwhile, “General Motors and Nissan said last month they would offer self-driving cars to customers before the end of the decade.” Will the Army find itself besieged by geek insurgents and challenged by rogue automakers?
To be sure, Markoff allows that the military “is not completely bereft of high tech ground vehicles”, pointing to LS3, Boston Dynamics’ “Big Dog” robot that has astonished the world with its ability to walk through rubble and mud and recover its balance after hard shoves while walking on ice. Where others, including “1.5 million YouTube viewers” see an engineering milestone that defines the state of the art for rugged, powerful, high-performance legged robots, Markoff sees a “cow” that is all that remains of the Army’s once ambitious plans to transform land warfare with unmanned tanks, howitzers and other heavy weapon systems. Cancelation of the Future Combat Systems (FCS) program in 2009 “knocked the Army out of the technology business,” laments James Lewis of the “bipartisan” Center for Strategic and International Studies. Alas, for the moment, the Army is “now really focused on helicopters.”
So let’s examine this a bit. Markoff correctly notes that “Critics said spending for [FCS] had run amok”, with estimated costs inflating first to $78 billion and later ranging as high as $340 billion, “and the goals did not necessarily address the challenges posed by today’s terrorists and insurgent forces.” Maybe programmable helicopters are what the Army needs more than sentient tanks? Moreover, as Markoff notes, enabling ground vehicles to “operate and survive in unmapped, hostile environments” is a hard technical problem. Driving over paved roads with lane markers and traffic signals, plus GPS and onboard maps, is much, much easier (for humans as well as robots). FCS’s cost overruns weren’t just a matter of bait-and-switch salesmanship. The technology just wasn’t there. It needed to be created first; and that, with Big Dog and, as Markoff notes, $261 million budgeted in the current year for unmanned ground vehicles alone, is just what the Pentagon-funded labs are still busily working on.
A quarter-billion bucks seems to me like a big program for a technology that is not ready to go into production. But Markoff sees it lacking in comparison with the $6 billion he says is being spent on autonomous aircraft (because they work already). He quotes John Arquilla of the Naval Postgraduate School, who finds it “particularly troubling” that robotic trucks weren’t available to bear the brunt of IED hits in Iraq and Afghanistan. Of course, if we’d had fully robotic soldiers (known colloquially as Terminators, Centurions and Robocops), we could have sent those, too. Maybe in the next war, we’ll have the robotrucks, but the Cylons still won’t quite be ready. Arquilla may find that troubling.
Perhaps I shouldn’t be so harsh; after all, nothing that Markoff says is really inconsistent with the picture I painted, even if, judging by headlines alone, his article might appear to contradict mine. FCS was not the first overly ambitious military program to come to grief, and it won’t be the last. Drones that walk are harder to make than ones that fly. But the march toward killer robots continues, and it is being led by the United States of America.
Here I must emphasize that the important issue is not whether robots can navigate autonomously, but whether they can decide whether, when and what or whom to attack and kill or injure autonomously.
Autonomous lethal robots already in use include land and sea mines, sensor-fuzed cluster munitions, and counter-artillery and anti-missile systems which can intercept incoming mortars and rockets in mid-air. Even most AWS opponents are willing to allow the latter kind of purely defensive systems when they are operating under human supervision and directly protecting human life. But some counter-artillery systems can also return fire autonomously, raining death on whoever happens to be in the vicinity of the calculated point from which mortars and shells are being launched. This makes a mockery of the frequently-heard claim that the US military is not currently developing, purchasing or using any type of autonomous lethal systems. Fully autonomous offensive weapon systems are also under development, and arguably in use, as well. As I discuss in the Bulletin, the DoD’s policy directive classifies as “semi-autonomous” some kinds of missiles which, in reality, are to be sent on fully autonomous search-and-destroy missions. There is also a great deal of research into warfighting by swarms of autonomous weapons and teams of humans and robots with flexible autonomy. A number of unarmed autonomous ships and submarines under development for the Navy are obviously likely to be weaponized later.
As Markoff notes, the military, in the past, has not been eager to embrace machines that kill on their own decision. But it is being directed to do so by civilian policy guidance under this administration, with the goal of reducing both the financial and political costs of future wars, as well as the perceived imperative of pursuing emerging technology to maintain military supremacy. It may be true that many technical advances are coming out of the commercial sector these days, but the military remains the single biggest sponsor of cutting-edge robotics research and the biggest customer for its products.
The US military does not “lag” in any robot arms race. It is leading the world into one, at full speed.
UPDATE: Markoff’s article was included in today’s (Sept. 24) New York edition of the Times, page D3, with the headline: Army Lags in Race for Robot Vehicles. Meanwhile, gawker and various tweeters seem to understand Markoff as saying that the US military is behind [who?] on “killer robots,” even though his article is mainly about ground vehicles.