Swarms of armed drones programmed to “fire and forget.” Fully automated anti-missile shields. Killer robots designed to identify and eliminate suspected terrorists on sight.
It all sounds like something out of science fiction, but a new age of machines empowered to make decisions about life and death is dawning as research in artificial intelligence advances. And President-elect Donald Trump will have to make crucial decisions about whether the U.S. military embraces the technology.
“We’re on the doorstep of what armed conflict looks like in the 21st century,” said August Cole, a security expert and senior fellow at the Atlantic Council.
To military planners, lethal machines that select and attack their own targets would be the logical outgrowth of decades of development of increasingly intelligent weapons, including precision-guided missiles and remotely piloted drones. Although it could take decades to deploy weapons that operate free of all human supervision, the Pentagon has already tested drones equipped with facial recognition software that could, in theory, identify enemy insurgents and target them at will.
Human rights groups warn of a “robotic arms race,” with self-directed weapons killing people without regard for international law. And like the human-controlled drones the U.S. employs in hot spots like Yemen and Afghanistan, the groups say, these weapons would be impossible to restrict once they’re in widespread use.
Lethal robots “would not simply be another weapon in the world’s arsenals, but would constitute a new method of warfare,” nine House Democrats wrote to Secretary of State John Kerry and Defense Secretary Ash Carter this month, urging a preemptive ban on the development and use of the technology.
Calls for a ban have also come from at least 19 governments, including those of Pakistan, Mexico, Zimbabwe, Cuba and the Vatican, as well as a “Stop Killer Robots” campaign led by groups such as Human Rights Watch. A United Nations weapons conference in Geneva also took up the issue this month, although it has not yet called for any prohibitions.
One Trump transition team member says the U.S. should not only oppose bans on the weapons but should lead the way in developing them.
“The U.S. is not going to be amenable to any protocol in Geneva that bans the weapons they are developing,” said Steven Groves, a Heritage Foundation fellow assigned to the transition’s State Department landing team. “Why would they do it when their peer competitors are developing the same kind of weapons?”
He stressed that he does not speak on behalf of the transition team. The team did not respond to requests for comment.
Groves, who has written frequently about the issue, has previously warned that attempts to ban autonomous weapons or “regulate [them] out of existence” were unlikely to succeed — although he expressed worries last year that the Obama administration might “cave” to the human rights groups, as it previously had on land mines.
Developing autonomous weapons is “the only way U.S. armed forces can retain a tactical and strategic advantage over its enemies in future conflicts,” Groves wrote for Heritage last year.
The Obama administration has so far trod a middle ground. The White House is conducting a review of the ethical and legal questions posed by military artificial intelligence, although that is unlikely to be completed before Trump’s inauguration. Meanwhile, the Pentagon’s existing policies allow researchers to make advances in the technology while limiting its use in combat.
That clears the way for Trump to make crucial choices about what happens next.
“The next administration has to make some early decisions about when and where they will deploy those systems so they can be compliant with the laws of war,” said Heather Roff, a cybersecurity fellow at the policy think tank New America.
A 2012 Pentagon directive restricts so-called semi-autonomous weapons, which are prevented by their programming from acting with full independence, from engaging targets. But the directive also allows for research and development efforts into fully autonomous weapons with authorization from two undersecretaries of defense and the Joint Chiefs of Staff.
One example of this type of research is Lockheed Martin’s Long Range Anti-Ship Missile, a missile that can fly for hundreds of miles and maneuver to avoid radar on its own. Similarly, the Office of Naval Research announced this month that its drone swarm boats could successfully identify unknown vessels as friendly or suspicious with only “remote human supervision” — another potential step toward the day when smart weapons could decide whether to open fire. (That news came on the third day of the U.N. conference in Geneva.)
The Defense Department’s restriction on deploying autonomous weapons comes with a five-year limit, after which it must be reissued, canceled or updated. That means the Trump administration will have the opportunity next year to set new guidelines on military artificial intelligence research. For experts, such guidelines would ideally outline what is and is not acceptable for development and future use.
“There needs to be clarity on the no-go areas,” Roff said.
As in the 20th century debate about developing atomic weapons, supporters of killer-robot technology argue it can save lives — for example, military planners hope it can allow unmanned systems to undertake more complex and dangerous missions without endangering human warfighters.
But opponents say that by lessening the risks of combat, autonomous weapons would loosen the restraints on nations’ decisions to wage war. They say the machines would also “lack human judgment” — for instance, Amnesty International argues, they would be unable to disobey orders, unlike the Egyptian soldiers who refused to fire on protesters during the Arab Spring.
Among the questions at the heart of the debate: Who is held accountable when a robot weapon causes civilian casualties? What if a programming error leads to friendly fire? How could governments ensure that autonomous weapons respect human rights and the laws of war?
A White House report released in October called for a policy on autonomous and semi-autonomous weapons that is “consistent with international humanitarian law.”
“The goal is to develop a government-wide policy that is consistent with shared human values, national security interests and our international and domestic obligations,” a senior administration official said.
But a total prohibition on the technology is probably impractical, other experts say.
“You’re not going to be able to ban all robots or AI in this space,” said Peter W. Singer, a specialist on robotic weaponry at New America.
Cole, from the Atlantic Council, said that by limiting its own development of militarized artificial intelligence, the U.S. would leave the door open to countries that disregard human rights to lead the way. “What I worry about is you’re going to face adversaries who aren’t as concerned with laws of armed conflict and international humanitarian law,” he said.
At this month’s U.N. weapons conference, countries including the U.S. and China agreed — despite resistance from Russia — to create a working group of experts to study the issue. The group’s next step is to come up with a working definition of autonomous weapons, taking into account the technology’s military value, proliferation risks and humanitarian laws.
While that might seem like a glacial and bureaucratic approach to dealing with rapidly advancing technology, Human Rights Watch’s Mary Wareham was upbeat about the step. Wareham, the global coordinator of the “Stop Killer Robots” campaign, said similar working groups preceded bans on blinding laser weaponry and cluster munitions.
But Roff said she fears that the U.N. effort will have little impact. “I think it’s a baby step,” she said. And next year, she predicted, will bring “a lot of politicking and a lot of feet dragging. If they do come to a definition everyone agrees on, it’s going to be a very weak definition that will encompass no systems ever or all systems.”
Roff, who attended the Geneva conference, blamed Russian resistance for the lack of firmer action. The Russian delegation ultimately abstained from the vote to form the working group.
Roff said Trump’s impending presidency also makes her doubt that the U.S. will push for meaningful progress on the issue at the United Nations. “Given Trump’s predilection for supporting Russia,” she said, “I would not be surprised if the U.S. maintained a chilly silence in the next year.”
However the debate plays out will have vast implications for the future of war. And nations need to tread carefully, said Jamie Metzl, a senior fellow for technology and national security at the Atlantic Council.
“All humans should be extremely cautious about arming independent robots that aren’t asking for approval for everything they do,” he said.
Powered by WPeMatico