Weapons systems that fire autonomously are the apotheosis of elites insulating themselves from accountability.
The aversion I have to autonomous weapons dates back to a bygone afternoon when I sat playing Golden Eye, a James Bond-themed video game that I was progressing through with all the alacrity of its lead character until I reached the jungle level. Expert as I was at evading enemy soldiers, I found myself gunned down in a spray of machine-gun bullets, which turned out to be motion activated. "Oh bollocks," I cursed, determined to stay in character. "That's hardly sporting."
I suppose some of you will think I'm a nutty conspiracy theorist when I inform you that many inside the Defense Department are eager to deploy machines on the battlefield that autonomously kill rather than requiring human intervention to "pull the trigger." It sounds like dystopian fiction. But it has its champions, and most informed observers think it's inevitable that they'll win victories in coming years. There aren't even many objections to autonomous military hardware that doesn't kill: Think of drones that take off, surveil, and land entirely on autopilot. Is the day coming when drones like that are armed, programmed to seek out certain sorts of images, and automated to fire in certain circumstances? The short answer is that, given present trends, it's a realistic possibility.
Noel Sharkey is shaken and stirred -- and he isn't alone. As he recently noted in The Guardian, Human Rights Watch is so concerned that they're calling on international actors to immediately "prohibit the development, production and use of fully autonomous weapons through an international legally binding instrument; and adopt national laws and policies to prohibit the development, production, and use of fully autonomous weapons." Would that President Obama listened.
The same Guardian column flags a quotation from a recent Pentagon directive on autonomous weapons that is worth bearing in mind. The lethal machines of the future could conceivable fail due to "a number of causes," the directive states, "including but not limited to human error, human-machine interaction failures, malfunctions, communications degradation, software coding errors, enemy cyber attacks or infiltration into the industrial supply chain, jamming, spoofing, decoys, other enemy countermeasures or actions, or unanticipated situations on the battlefield."
The arguments against adopting this technology nearly make themselves. So I'll focus on a tangential but important observation.
In recent decades, America's elite -- its elected officials, bureaucrats, and CEOs, for starters -- have succeeded spectacularly at insulating themselves from responsibility for their failures. As the global economy melted down, Wall Street got bailouts. CEOs who preside over shrinking companies still depart with "golden parachute" severance packages. The foreign-policy establishment remains virtually unchanged despite the catastrophic errors so many of its members made during the Iraq War. The conservative movement is failing to advance its ideas or its political prospects, yet its institutional leaders and infotainment personalities are as profitable as ever. It is a self-evidently pernicious trend: Once someone achieves insider status, their future success is less and less dependent on how competently and responsibly they perform.
An innocent who dies at the handlessness of an automated killing machine? How easy to phrase the obligatory apology in the passive voice!
It's no wonder that some military leaders are so eager for the advent of autonomous weapons. At present, if WikiLeaks gets ahold of a video that shows innocents being fired upon, the incident in question can be traced back to an individual triggerman, an officer who gave him orders, and perhaps particular people who provided them with faulty intelligence. But an innocent who dies at the handlessness of an automated killing machine? How easy to phrase the obligatory apology in the passive voice! How implausible that any individual would be held culpable for the failure!
And would there ever be accusations that a supposed mistake was actually an intentional killing being conveniently blamed on autonomy gone wrong? I hardly think such suggestions would come from within the establishment. Levying that sort of accusation against a sitting president or the "brave men and women of the United States military" is precisely the sort of thing that bipartisan Washington, D.C., culture declares beyond the pale of reasonable discourse. And when autonomous Chinese drones "accidently" fire upon some Tibetans? Even a genuine accident in America's past would make it that much harder to refrain from giving the Chinese the benefit of the doubt that we ourselves had requested.
Most areas of American culture would benefit if the relevant set of elites were more accountable, but nowhere is in individual accountability more important than when death is being meted out. The secrecy that surrounds the national-security establishment and outsourcing drone strikes to the CIA already introduces a problematic lack of accountability to the War on Terrorism. Autonomous killing machines would exacerbate the problem more than anything else I can imagine.
We want to hear what you think about this article. Submit a letter to the editor or write to firstname.lastname@example.org.
Conor Friedersdorf is a California-based staff writer at The Atlantic, where he focuses on politics and national affairs. He is the founding editor of The Best of Journalism, a newsletter devoted to exceptional nonfiction.