Tearful Terminator? Military Seeks Killer Robots With Moral Judgment
Is the face of future warfare that of a steel-skinned Terminator-style killer robot—with a tear trickling down its cheek? That's essentially the goal of research funded by the U.S. military that seeks to defuse a growing chorus of warnings that drones and other increasingly autonomous weapons are morphing into self-directed killer robots.
DefenseOne's Patrick Tucker reports:
The Office of Naval Research will award $7.5 million in grant money over five years to university researchers from Tufts, Rensselaer Polytechnic Institute, Brown, Yale and Georgetown to explore how to build a sense of right and wrong and moral consequence into autonomous robotic systems.
This isn't just pie-in-the-sky research. Semi-autonomous weapons systems are already deployed by countries including Israel, South Korea, and the United States. The weapons are restrained from killing on their own say-so more as a matter of policy than because of technical limitations. (That's the reportedly discontinued Armed Robotic Vehicle depicted above.) The United States military currently requires robotic weapons systems to be human-supervised and to engage only non-human targets. Fully autonomous Terminator-style systems aren't allowed. Yet.
That's because people find the idea of machines choosing and snuffing their own targets creepy.
The United Nations Human Rights Council wants a moratorium on lethal autonomous robotics—at least until an internationally agreed upon framework has been established. (That's the U.N. all over—concern and impotence in the same sentence.)
"Humans must not be taken out of the loop over decisions regarding life and death for other human beings. Meaningful human intervention over such decisions must always be present," the Vatican's Archbishop Silvano Tomasi told an international gathering on the issue just yesterday.
And a new report from Human Rights Watch and Harvard Law School's International Human Rights Clinic cautions:
Fully autonomous weapons' inability to relate to humans could interfere with their ability to ensure that all means short of force are exhausted. … Furthermore, it is unlikely that a fully autonomous weapon would be able to read a situation well enough to strategize about the best alternatives to use of force.
While fully autonomous weapons would not respond to threats in fear or anger, they would also not feel the "natural inhibition of humans not to kill or hurt fellow human beings." Studies of human soldiers have demonstrated that "there is within man an intense resistance to killing their fellow man." Compassion contributes to such a resistance, but it is hard to see how the capacity to feel compassion could be reproduced in robots.
That would seem to be a daunting task, You could program a robot with all sorts of scenarios and decision trees, but at the end of the day, it's a robot following programming, not a human following values and instinct.
Then again, maybe that will prove safer. No emotions means no rage killing for one thing. Could an arsenal of compassion-less robot killers mean fewer atrocities?
Chances are that we'll get to find out. The U.S. project may or may not succeed in teaching morality to computers. But it's hard to imagine that all militaries will resist the temptation to deploy advancing generations of automated weapons.