Human Rights Watch: Petition Against Robots On the Battle Field

New submitter KublaCant writes "'At this very moment, researchers around the world – including in the United States – are working to develop fully autonomous war machines: killer robots. This is not science fiction. It is a real and powerful threat to humanity.' These are the first words of a Human Rights Watch Petition to President Obama to keep robots from the battlefield. The argument is that robots possess neither common sense, 'real' reason, any sense of mercy nor — most important — the option to not obey illegal commands. With the fast-spreading use of drones et al., we are allegedly a long way off from Asimov's famous Three Laws of Robotics being implanted in autonomous fighting machines, or into any ( semi- ) autonomous robot. A 'Stop the Killer Robots' campaign will also be launched in April at the British House of Commons and includes many of the groups that successfully campaigned to have international action taken against cluster bombs and landmines. They hope to get a similar global treaty against autonomous weapons. The Guardian has more about this, including quotes from well-known robotics researcher Noel Sharkey from Sheffield University."

Comments

Recommended Reading

[smpoole7]

http://en.wikipedia.org/wiki/Berserker_(Saberhagen)

Fred Saberhagen's "Beserker" series.

Aside from touching on the subject at hand, it's just some crackin' good sci-fi. :)

I don't know if we'd ever reach that point ourselves, but in that series, an unknown (and now extinct) alien race, losing a war and desperate, created "doomsday" machines that were simply programmed to kill all life. They were self-replicating, self-aware AIs that took their task seriously, too.

Then again, I ask myself what some jihadist might do, if given half the chance ... . .. ..

These are not the droids you're looking for

[rodrigoandrade]

Hey, James Cameron, are you the submitter??

The automomous Terminator-style robots the summary refers to are far from becoming a battlefield standard, much to the disappointment of the /. crowd and sci-fi nerds.

Predator drones et al., like all current robotic devices in the battlefield, still have a human being in charge making all the decisions, so the points raised are completely moot.

Re:These are not the droids you're looking for

[fuzzyfuzzyfungus]

Yes and no: especially sophisticated autonomous robots, either self-driving vehicles or biomimetic killbots of some sort, are sci-fi stuff; but land mines 'That's bi-state autonomous area denial agent sir to you, cripple!' and more sophisticated devices like the Mark 60 CAPTOR are autonomous killer robots.

And, so far, they've proven deeply unpopular in bleeding-heart circles. The fancier naval and anti-vehicle mines are still on the table; but the classic land mine enjoys a sense of ethical distaste only slightly less than just hacking off children's limbs yourself...

It's the same as bio-warfare

[RobinH]

If you think about a virus for a second, it's the same thing. You can't reason with a virus. It doesn't make moral decisions. It just does what its DNA programs it to do, and it's even more dangerous because it's self-replicating. We need to deal with autonomous robots the same way we deal with bio-warfare.

Effectiveness trumps morality every time.

[concealment]

I don't mean to be the dark figure in this conversation, but I think it's inevitable that robots will be used on the battlefield, just like people are going to continue to use cluster bombs, land mines, dum-dum bullets and other horrible devices. The reason is that they're effective.

War is a measurement of who is most effective at holding territory. It is often fought between uneven sides, for example the Iraqi army in their 40-year-old tanks going out against the American Apaches who promptly slaughtered them. Sometimes, there are seeming upsets but often there's an uneven balance behind the scenes there as well.

Robots are going to make it to the battlefield because they are effective not as killing machines, but as defensive machines. They're an improvement over land mines, actually. The reason for this is that you can programmatically define "defense" where offense is going to require more complexity.

Already South Korean is deploying robotic machine gun-equipped sentries on its border. Why put a human out there to die from sniper fire when you can have armored robots watching the whole border?

Eventually, robots may make it to offensive roles. I think this is more dubious because avoiding friendly fire is difficult, and using transponders just gives the enemy homing beacons. In the meantime, they'll make it to the battlefield, no matter how many teary people sign petitions and throw flowers at them.

The 3 laws are fiction

[Dr. Tom]

How many times must it be said? Asimov's 3 "laws" have nothing to do with real robotics, future or present. They were a _plot device_, designed to make his (fictional) stories more interesting. Even mentioning them at all in this context implies ignorance of actual robotics in reality. In reality, robot 'brains' are computers, programmed with software. Worry more about bugs in that software, and lack of oversight on the people controlling them.

Total Garbage.

[inhuman_4]

This article is absolute garbage. Almost everything in that Guardian article is misinformed and sensationalist.

"fully autonomous war machines"? Care to give an example? I've follow this stuff pretty closely in the news on top of researching AI myself. And from what I have seen no one is working on this. Hell, we've only just started to crack autonomous vehicles. They site X-37 space plane for gods' sake. Everything about that is classified so how do they know it is autonomous?

My favourite gem has to be this one: "No one on your side might get killed, but what effect will you be having on the other side, not just in lives but in attitudes and anger?". Pretty sure that keeping your side alive while attacking your opponent has been the point of every weapon that has ever been developed.

Re:Obama already leads the way

[Anonymous Coward]

In robot drone murders and you morons think he will sign something? Obama, Nobel Peace Prize winner that has killed the most innocent women and children yet!

I believe Yasar Arafat, Henry Kissinger, Yitzhak Rabin, Shimon Peres, Menachem Begin, and Le Duc Tho all currently lead Obama the "Number of Innocents Killed by a Nobel Peace Prize Winner" race.

Re:Fear of robots is a red herring

[ultranova]

Robots have no emotion, no bias. Imagine deploying a few hundred (or thousand) semi-autonomous robotic peacekeepers into a conflict zone. They maintain the peace 24/7, they never tire, they are alert and objective in their duties.

An autonomous robot needs to form a model of what's happening around it, use that to figure out what its possible long- and short-term actions will be, and finally decide how desirable various outcomes are relative to each other. All of these steps are prone to bias, especially since whoever designed the robot and its initial database is going to have their own biases.

Also, a robot acting in real life cannot carefully think everything through. There's simply not enough time for that. This necessiates some kind of emotion-analogy to provide context for reflex and simple actions, just like it does on living beings.

Look at Iraq and Afghanistan! $1 trillion and thousands of allied casualties. Deploy a robot army and watch the costs come down. No need for living quarters, no need of food or water, logistics becomes cheaper in every aspect.

So there will be a lot more "interventions", since the cost (to you) is lower. I think that's part of what worries the the HRW.

Re:Fear of robots is a red herring

[kannibal_klown]

A couple of issues.

1) Software can be hacked... either partially or totally. Maybe just putz with the Friend-Or-Foe logic, maybe take direct control, etc. Sure, humans can be blackmailed and extorted but usually on an individual basis. Mass-putzing with a regiment or squad and you have serious issues. Such as perhaps those drones protecting the US (if they ever become truly robotic).

2) It does make war a bit more meaningless. If you aren't facing emotional losses, then there's little reason NOT to go to war. If it's not personalized... then who cares? Sure, even now we have sympathy for the other side and protests and such... but the majority of the people that care mostly care because our brothers / sisters / sons / daughters / etc. are out there possibly dying. So that helps push back the question "should we actually GO to war with them?"

3) There ARE concerns of self-aware armed robots. Make them too self aware, and maybe they realize that the never-ending violent slaughter of humans is contradictory to their goals of preserving their owners' lives. In which case they take a OVERLY logic to preserve the FUTURE "Needs of the many" by doing PLOTLINE X. Sure, it sounds like bad sci-fi... but as you say they have no emotions and only logic. Take away emotion, and we become like cattle... where they cull the herd due to a few random mad-cow cases to save the majority.

Re:Fear of robots is a red herring

[TheLink]

It is a red herring. The problem is not robots in war. There's no big difference between using robots and drones in wars compared to using cruise missiles in wars. And only a slight difference between using soldiers that have been conditioned/brainwashed to follow orders unquestioningly.

The real problem is the ease of starting wars that only benefit a very few people. Hence my proposal: http://slashdot.org/~TheLink/journal/208853

In the old days kings used to lead their soldiers into battle. In modern times this is impractical and counterproductive.

But you can still have leaders lead the frontline in spirit.

Basically, if leaders are going to send troops on an _offensive_ war/battle (not defensive war) there must be a referendum on the war.

If there are not enough votes for the war, those leaders get put on deathrow.

At a convenient time later, a referendum is held to redeem each leader. Leaders that do not get enough votes get executed. For example if too many people stay at home and don't bother voting - the leaders get executed.

If it turns out later that the war was justified, a fancy ceremony is held, and the executed leaders are awarded a purple heart or equivalent, and you have people say nice things about them, cry and that sort of thing.

If it turns out later that the leaders tricked the voters, a referendum can be held (need to get enough signatories to start such a referendum, just to prevent nutters from wasting everyone else's time).

This proposal has many advantages:
1) Even leaders who don't really care about those "young soldiers on the battlefield" will not consider starting a war lightly.
2) The soldiers will know that the leaders want a war enough to risk their own lives for it.
3) The soldiers will know that X% of the population want the war.
4) Those being attacked will know that X% of the attackers believe in the war - so they want a war, they get a war - for sufficiently high X, collateral damage becomes insignificant. They might even be justified in using WMD and other otherwise dubious tactics. If > 90% of the country attacking you want to kill you and your families, what is so wrong about you using WMD as long as it does not affect neighbouring countries?

I think if this was implemented it would be much better than banning robots. I'm biased of course ;).