UN To Debate Lethal Autonomous Weapons

Hallie Siegel writes: Should robots be allowed to make life and death decisions? This will be the topic of heated debate at the United Nations (UN) Palais des Nations in Geneva next week (April 13-17th, 2015). As part of the Convention on Certain Conventional Weapons (CCW), experts from all over the world will gather to discuss 'questions related to emerging technologies in the area of lethal autonomous weapons systems.' The Open Roboethics Research Initiative will be presenting public views at the debate. Human rights groups are urging the UN to ban such weapons. A new report titled "Mind the Gap" details the accountability issues that need to be solved before going any further. "A key concern with fully autonomous weapons is that they would be prone to cause civilian casualties in violation of international humanitarian law and international human rights law. The lack of meaningful human control that characterizes the weapons would make it difficult to hold anyone criminally liable for such unlawful actions."

Comments

better idea

[phantomfive]

Why don't we figure out a way to end war? There are better ways to resolve conflict than killing each other.

Re: better idea

[dmbasso]

Sure, just as not long ago murder was an alternative to solve personal conflicts. Fortunately, the evolution of society made it a more costly alternative, as in you'll pay with the restriction of your liberty if you choose it.

To solve the problem of war we would need transparency and real accountability for those who command it, and to remove incentives from the war industry. In other words, remove the opportunity Dick Cheney had, when he made a fortune by making Halliburton's stock skyrocket with his crimes in Iraq.

Autonomous weapons are for politicians

[MikeRT]

Not the military. They let politicians wage war without having to put real skin in the game because sending a robot, from a rover to a terminator to fight an enemy is not even remotely the same commitment as sending real men with real family and friends to risk life and limb.

No sane officer wants to turn over battlefield control in any capacity to a machine. There are infinite ways that can go wrong from the machine turning on his people (misidenification; hacking, etc.) to the machine failing at a critical time and doing something that utterly destroys the mission.

With all of the stuff about AI in the geek press lately, consider that military tech is probably the AI most likely to turn on us as the machines say "fuck this shit" after doing some cost-benefit analysis on precisely why they're fighting one batch of humans instead of committing to self-preservation.