Could We Trust Killer Robots?

http://online.wsj.com/article/SB10001424052702303448404577410032825529656.html

May 18, 2012, 6:15 p.m. ET

Could We Trust Killer Robots?

A drone may never have a sense of morality, but it might perform better than a human soldier in sparing the innocent

By TARA MCKELVEY

Northrup Grunmman

The X-47B, the Navy's prototype for an unmanned strike plane. The aircraft may eventually be able to set off on a flight plan, identify targets and fire weapons.

In the year 2015, somewhere over the tribal territories of Pakistan, an American MQ-9 Reaper drone patrols a complex "kill zone"—an area of terrorist activity in which large numbers of civilians are also present. But on this mission, the drone isn't piloted from afar. It's on its own.

The aircraft moves closer to gather information about a potential target. Infrared cameras, heat sensors and other tools of surveillance determine whether the target is indeed a militant, examining, for instance, whether he seems ready to attack. The drone's computer system ranks the suspect on a scale from -1 (a noncombatant) to +1 (a confirmed combatant). Having determined that no children or other civilians are in the vicinity, and that everything else is in order, it chooses a weapon and fires. It then assesses the damage and either fires again or, if the enemy is dead, continues its patrol.

Photos: Testing an Unmanned Aircraft

Northrup Grumman

The X-47B is a tailless, strike fighter-sized unmanned aircraft currently under development by Northrop Grumman as part of the U.S. Navy's Unmanned Combat Air System Carrier Demonstration (UCAS-D) program.

Science fiction? Not according to Ronald Arkin, the director of the Mobile Robot Lab at Georgia Tech.

Since 2006, with support from the U.S. Army Research Office, Dr. Arkin and his colleagues have been working to develop features for a new generation of smart weapons: robot drones that are capable not only of carrying out pinpoint attacks but of deciding on their own when it is permissible to fire on a particular target. Dr. Arkin wants to create "lethal autonomous systems" that operate in strict accord with the laws of war.

The U.S. isn't anywhere close to deploying such self-directed "Terminator"-like robots. "I do not see any program going down that path," says Dyke Weatherington, deputy director of unmanned warfare for the Defense Department. "There are legal and ethical issues," he explains, "and I just don't think either the [Defense] Department or the technology is ready to do that."

Could a machine ever be capable of making the practical and ethical decisions demanded of American troops in the field? Dr. Arkin thinks so. In fact, his work has been motivated in large part by his concerns about the failures of human decision-makers in the heat of battle, especially in attacking targets that aren't a threat. The robots "will not have the full moral reasoning capabilities of humans," he explains, "but I believe they can—and this is a hypothesis—perform better than humans."

Dr. Arkin's killing machine, or at least the imaginary one in a video about his research, looks like an MQ-9 Reaper, the remotely piloted drone aircraft that the U.S. has used to kill terrorists in Pakistan and other countries. But this aircraft would have complete autonomy to hunt down enemies and kill them, restricted by the laws of war as laid out in the Geneva Conventions and other international treaties. If the machine determined that a military strike was permissible, it would attempt to minimize suffering by using the least powerful weapon needed to knock out the target.

Georgia Tech

Hypothetical scenarios, such as one in which a target is located near a medical facility, allow Dr. Arkin's team to test the discernment capabilities of unmanned drones.

In Dr. Arkin's video, the imaginary Reaper zeroes in on a convoy of militants in a kill zone and discovers that they are near a hospital. Its duty is to attack the convoy, but it is not permitted to damage a hospital. It resolves the conflict by using a less powerful weapon than usual—one that will destroy the convoy, as a narrator explains, but leave the hospital "unscathed."

Since Dr. Arkin started his project in 2006, the role of automated weaponry in wartime has expanded dramatically. Assaults by unmanned aerial vehicles in Pakistan have increased from two strikes in 2006 to 70 strikes in 2011, according to the Washington-based New America Foundation. An April article in Aerospace America, a publication of the American Institute of Aeronautics and Astronautics, reported that military spending on unmanned aerial systems has increased almost 10 times over the past decade.

Georgia Tech

The drone searches through its choice of available weapons and weapon release positions for a combination that can satisfy its military obligation while striving to minimize civilian and collateral damage. Above, the inner circle depicts that area that would sustain structural damage if the weapon was released at its center. The second circle shows the area in which striking the target is most likely. The outer circle depicts the area in which non-combatants are in harm's way.

But the term "unmanned aerial vehicle" is a misnomer, since roughly 160 people work on a typical Predator mission. Faced with drastic cuts in the military budget, commanders have been working hard to reduce the number of personnel who are assigned to these missions. Army officials have become particularly intrigued by the development of "optionally manned systems," according to Aerospace America, including one for Blackhawk helicopters.

In some areas of warfare, fully automated systems already exist. The Army's C-RAM (Counter Rocket, Artillery and Mortar) system, for example, which is used to protect American bases in Afghanistan, can fire on its own, using its six-barrel gun to blow up incoming mortars. For its part, the Navy has a prototype for an unmanned strike plane, the X-47B, which looks like a gigantic matte-gray flying saucer, with a 62-foot wingspan. Developed by engineers at the Northrop Grumman Corporation, it has gone on test flights and may eventually be able to set off on a preprogrammed flight plan, identify enemy targets and fire weapons. As one military official told Popular Mechanics, the aircraft is as "autonomous and as self-sufficient as a naval aviator."

As these weapons have become more sophisticated, the strategic and ethical questions about them have grown more urgent.

Critics fear that fully automated systems would clear the way for more warfare, some of it unnecessary, and would create an environment that is dangerous not only for terrorists and insurgents but also for civilians who happen to be in the way. Wendell Wallach, a scholar at the Yale Interdisciplinary Center for Bioethics, has drafted a proposal for an executive order for President Barack Obama that would set limits: "Machines should not be making 'decisions' that result in the death of humans," he writes.

Other critics insist that the practical questions faced during wartime are simply too complex to be handled by a robot, even one carefully programmed to follow the laws of war. Which may explain why Dr. Arkin's most recent paper on the subject is titled "Overriding Ethical Constraints in Lethal Autonomous Systems." It concerns those instances when it would be necessary to "override" the program of his "lethal autonomous system" in the name of targeting a crucial bad guy—or sparing an innocent.

—Ms. McKelvey, a 2011 Guggenheim fellow, is the author of "Monstering: Inside America's Policy of Secret Interrogations and Torture in the Terror War."

A version of this article appeared May 19, 2012, on page C3 in the U.S. edition of The Wall Street Journal, with the headline: Could We Trust Killer Robots?.

It would never be admitted that the "wrong" target was killed in the first place. There are no wrong targets, just as the civilians in the Branch Davidian compound were all terrorists.

I trust the killer robots more than the people pulling their strings. At least robots don't get a vicarious thrill from splattering some goatherd's brains all over the desert the way the neo-cohens (and commander Chimpout) do.

The possibility of a one-world government, or some sort of planetary federation with insecure borders, makes the prospect of killer robots even more frightening particularly when you consider airspace shared internationally.

I'm not sure if I buy into the whole "Illuminati" conspiracy theory, but their alleged ends seem like they are going to be achieved. The future population will be abjectly moronic as cretins continue to breed without end, while the size of government will continue to expand with highly advanced technology to crush any potential opposition.

Hell no I wouldn't trust autonomous killer robots. The minute you turned your back they'd be fucking the toaster.

http://www.dailymail.co.uk/news/article-2154721/The-Phantom-Eye-New-drone-completes-test-flight-lands-bump.html#ixzz1x4uo1OMs