Examining the Ethical Implications of Robots in War

Schneier points out an interesting (and long, 117-pages) paper on the ethical implications of robots in war [PDF]. "This report has provided the motivation, philosophy, formalisms, representational requirements, architectural design criteria, recommendations, and test scenarios to design and construct an autonomous robotic system architecture capable of the ethical use of lethal force. These first steps toward that goal are very preliminary and subject to major revision, but at the very least they can be viewed as the beginnings of an ethical robotic warfighter. The primary goal remains to enforce the International Laws of War in the battlefield in a manner that is believed achievable, by creating a class of robots that not only conform to International Law but outperform human soldiers in their ethical capacity."

What's the point?

Obviously a country that can send robots instead of soldiers to fight is way more likely to become 'war happy' - so I'm not sure this robot thing is a good idea at all.

Besides, if your enemy expects your robots to defeat their army, what would be the point of fighting them in the first place? Attacking civilians seems a more logical step (I don't think it's reasonable to demand any country at war not to attack only military targets where there's none that can't be replaced easily).

(and no, I didn't read the whole 117 pages, but after a quick glance I reached the conclusion that whoever wrote the title didn't either, so I'm sharing my thoughts on the title, not the PDF)

Re:What's the point?

Obviously a country that can send robots instead of soldiers to fight is way more likely to become 'war happy'

Of equal concern to me is the fact that a country with a robot army can use them against their own citizens with no chance of mass mutiny.

Re:What's the point?

I don't think automatic robots will ever be a smart plan. The chance of malfunction is just too great, and the consequence would be too serious. There've been a million sci-fi movies to that effect, from "Terminator" to "I, Robot".

What would be interesting though would be robots as a shell to the humans they represent. Think "Quake" with a real robot proxy in the real world. Soldiers with hats on showing wide angle camera views of their area and a quake-like interface that would allow them to attack or assist as needed. Limited automation, but case-hardened soldiers being run by trained humans would present a powerful adversary. Heck, every army recruit would already know 80% of how to operate one on signing day if the UI was good.

I know I'd be a lot upset with "Four robots were blown up by a roadside bomb today. They should be operational again by tomorrow." than to see more soldiers die.

Re:What's the point?

I'd think that it'd be more effective to attack infrastructure--things like power stations, traffic control systems, that manner of thing--than to go after civilians directly.

For one thing, what's the point of taking over a territory if there's nobody there to rebuild and to use as a resource?

For another, it looks a -lot- better on the international PR scene if your robots decidedly ignore the civilians and only go after inanimate strategic targets--at least, up until the point that they get attacked. With that sort of programming, you could make the case that you're "seeking to avoid all unnecessary casualties" etc. etc.

Mowing down a civilian populace does sow terror, of course, but keeping the civilians intact (if in the dark and without water) can be argued to be more effective.

Re:

Mowing down a civilian populace does sow terror, of course, but keeping the civilians intact (if in the dark and without water) can be argued to be more effective.

When desperate enough civilians can become soldiers. In fact, some can be willing to die (

Re:

At which point, once they take up arms, they're conveniently reclassifying themselves as irregular enemy combatants. If they had only stayed calm and awaited further instruction from our occupying forces, robotic or otherwise, this sad scene could have be

Re:

It sounds like this is proposing something along the lines of Asimov's Three Laws of Robots, only instead of not being able to harm humans at all, they're able to harm humans only in an ethical manner.

Instead of sending human soldiers into Iraq who are

Economic Warfare & Gundams

Consider that robots cost money, the country with more economic power is likely to be the winner in such a conflict. A large part of the U.S.A.'s success in WW2 was the sheer capacity of it's factories which were by if nothing but distance well defended against attack. European nations where under constant attack on their military infrastructure while American Factories where never bombed and even the concept of saboteurs blowing up factories in the States was a ridiculous notion to the Axis. Sure, Blow up the Pittsburgh bomb factory then you still have 20 more scattered about the US.

Robots won't be used simply because a robot doesn't have the discrimination as to who to attack and not to. Despite Orwellian fantasies, the practical upshot is that you would suffer to much friendly fire from such weapons and intense PR backlash. Sorry I don't see it happening.

Telepresence weapons are far more likely, as we have already seen in use.

Japan's Ministry of Agriculture [animenewsnetwork.com] has been denying their work on this. America is full of fully trained pilots for these crafts (Wii, Xbox, Playstation etc).

Suggested reading of Robert A. Heinlein's Starship Troopers and Robert Aspirin's Cold Cash War

Re:What's the point?

Obviously a country that can send robots instead of soldiers to fight is way more likely to become 'war happy' - so I'm not sure this robot thing is a good idea at all.

Not necessarily. One of the big reasons the USA lost in Vietnam was that it became politically unacceptable to have body bags coming home. The current administration found a solution to that; ban news crews from the areas of airports where the body bags are unloaded.

Beyond that it's just a question of economics. It costs a certain amount to train a soldier. Since the first world war, sending untrained recruits out to fight hasn't been economically viable since they get killed too quickly (often while carrying expensive equipment). A mass-produced robot might be cheaper, assuming the support costs aren't too great. If it isn't then the only reason for using one would be political.

Re:What's the point?

This assumes that once you have destroyed your opponent's robotic army you are done. However most likely is that after the robots will come humans, so in the end you are going to lose both.

Besides, I still fail to see why a country which is likely to lose in the robotic war would accept these rules, when it makes a lot more sense to attack the other country's civil population - which in turn might reconsider the whole thing.

Fighting from the sofa is one thing, having bombs exploding nearby is quite different.

Obligatory

I for one welcome our new robotic overlords.

Why bother going to war in the first place anymore

If you've got battlebots, why not have one against another to resolve international conflicts, rather than destroy infrastructure and the like?

It'd probably take a mountain of treaties and the like, and of course any organization used to judge the battlebot contest would be rife for corruption and whatnot, but it couldn't be that much worse than what happens around the World Cup and the Olympics...

Re:Why bother going to war in the first place anym

If you've got battlebots, why not have one against another to resolve international conflicts, rather than destroy infrastructure and the like?

We've already built structures to solve international conflicts, and it works extremely well when the two sides are willing to work through those structures. The US doesn't need battlebots to deal with European powers, because both sides are willing to talk it through instead. However, when Iraq refuses to cooperate, or the Arabs in Israel refuse to cooperate, the procedures break down and you're left with two countries that can't reach an agreement without raising the stakes.

In other words, for those countries willing to abide by a mountain of treaties, the problem's already solved. It's the other countries that are the problem, and they're unlikely to resolve their differences like this anyway.

Re:Why bother going to war in the first place anym

War is what happens when treaties stop working. You can't have a treaty for some other competition to replace war--if that was the case, FIFA would have replaced the UN by now and Brazil would be a superpower. The purpose of war is to use force in order to impose your will on the enemy, whoever those people may be. The idea is, after your robots destroy the enemy's robots, they will continue to destroy the enemy's infrastructure and population until they give up.

Re:What if they programmed a war,and nobody logged

Or use an alternative 1960's solution. [startrek.com]

Political Ethics...

Yes. Superior robotic ethics. A regular Ghandi-bot, saving only those who are threatened, willing to die rather than kill in doubt.

That's all well and good... but what of the men who send these robots into battle? What happens to their sense of ethics? Do they begin to believe that their sending troops into pacify a landscape over political differences is a morally superior action? Do they begin to believe that death-by-algorithm is a morally superior way of dealing with irrational people?

There's an endless array of rationalizations man can make for war, and subjugation of those who disagree with them. Taking the cost of friendly human lives out of the equation of war, and replace it with an autoturret enforcing your wishes doesn't make for a 'morally superior' political game. For many, it would make for an endgame in terms of justifying a military police as the default form of political governance.

Ryan Fenton

Re:Political Ethics...

War is about sacrifice, cost, and essentially fighting for what you believe in, hold dear, and WILL DIE to preserve. If you remove the *human* cost from war, then where is the cost? What will it mean if no-one dies? Will anyone remember what was fought for? Will they even recognize why it was so important in the first place?

Bullshit. War is about taking orders, fighting for what someone else believes in, and then getting blown up. Dulce et decorum est pro patria mori and all that shite. That poetic nonsense you spout there is just part of the cultural lie that sells war as romantic and idealistic to every generation of young fools who sign up and go out there to put their lives on the line for the sake of the millionaires. You got it from anime, too... how sad is that? You're buying the same line of bullshit that inspired the damn kamikaze! Clue: Bushido is a lie. Chivalry is a lie. War is about nothing but power.

Also, if we have mass armies of robots, won't the victor simply be the one with the most natural resources (metal, power, etc) to waste? (Better weapons technology aside)

Yes. How does that differ from the present situation?

Too easy to counter

My Apple comp00tar will just upload a virus wirelessly to them and they will all shut down! I've seen it done!

The biggest question...

Do these killbots have a preset kill limit? Can they be defeated by sending wave after wave of your own men at them?

Re:The biggest question...

It's probably MAX_INT.

Natalie Portman Robot

When are they going to stop using robots for evil and start using it for good? I want a Natalie Portman "pleasure model" robot and I want it now! Science has lost it's way.....

So what you're really saying is...

...you want robots to make love and not war.

I have bad news for the war ethicists

Wars are won by those who do not follow the "rules." There are no rules in war. If there were, then there would be a third party far more powerful than either side who could enforce said rules. If there was, then that power could enforce a solution to the conflict that started the war, and there would be no need for war. Said power would also not need answer to anyone, and would be exempt from said rules (having no one capable of enforcing them).

I have some good news and some bad news.

The question of making lethal robots act ethically is far easier in some ways than doing so with humans and far harder in others. On the plus side, robots will not be subject to anger, fear, stress, desire for revenge, etc. So they should be effectively immune to the tendency toward taking out the stress of a difficult or unwinnable conflict on the local population. On the minus side, robots have no scruples, probably won't include whistleblowing functions, and will obey any order that can be expressed machine-readably.

The real trick, I suspect, will not be in the design of the robots; but in the design of the information gathering, storage, analysis, and release process that will enforce compliance with ethical rules by the robot's operators. As the robots will need a strong authentication system, in order to prevent their being hijacked or otherwise misused, the technical basis for a strong system of logging and accountability will come practically for free. Fair amounts of direct sensor data from robots in the field will probably be available as well. From the perspective of quantity and quality of information, a robot army will be the most accountable one in history. No verbal orders that nobody seems to remember, the ability to look through the sensors of the combatants in the field without reliance on human memory, and so on. Unfortunately, this vast collection of data will be much, much easier to control than has historically been the case. The robots aren't going to leak to the press, confess to their shrink, send photos home, or anything else.

It will all come down to governance. We will need a way for the data to be audited rigorously by people who will actually have the power and the motivation to act on what they find without revealing so much so soon that we destroy the robots' strategic effectiveness. We can't just dump the whole lot on youtube; but we all know what sorts of things happen behind the blank wall of "national security" even when there are humans who might talk. Robots will not, ever, talk; but they will provide the best data in history if we can handle it correctly.

Same problem...

I've always wondered how HAL or Joshua would interpret:

Rule 1: Kill enemy combatants.
Rule 2: Do not kill or abuse prisoners.

"Take no prisoners, kill everything that moves" would be the most efficient means of satisfying both, especially after friendly-fire ensues.