Taking a break

Ladies and gents, I have been informed by my doctor that I have to stop typing so much. So I'm going to have to take a break from blogging. Which is lame, but it's doctor's orders... I'll have some searingly intellectual things to say when I come back, I'm sure. Stay tuned.

Continuing with all the problems I see inherent in the programming of combat robots to have 'functional morality'...

Putting aside the very real difficulties in the mere logistics of trying to program a wartime 'morality' into a machine for a moment, I find a much deeper issue with the moral questions inherent in trying to make a machine act as a human would in such an extreme situation as a war zone, in the first place. What does this say about us, our feelings toward robots and the wars we want to put them in? What would be the point of using robots instead of soldiers in the first place?

Ostensibly, I'm sure, the idea would be that using robots, we could spare the lives of our soldiers, which are worth immeasurably more to us on an emotional level than the thousands and millions of dollars we'd have poured into robot soldiers. But, let's step back for a minute and look at the bigger picture, which will tell us in no uncertain terms that most wars are fought for money these days anyway, and that in the grand scheme of things, human life has never stood up to the powers that be trying to get to the bottom line. Simply through experience, I think we can make the assumption that a military engineer saying he's concerned with protecting human life is kind of like a slave owner saying he believes that all men are created equal. It sounds nice, and we want to believe it, but there's something else going on there. In this case, it seems to me that the something else goes along the lines of: "Dude, let's make a super-powered robot with awesome firepower capabilities to do our work for us!"

Not exactly an irresponsible idea, given that we very well could save human lives with these things. But, given that they're to be 'super-powered robots with awesome firepower capabilities' (according to me), we do not seem to be too worried about anybody's lives but our guys'. We seem to want to kill the crap out of anybody who could be on the other side. Which was all well and good for, you know, the Mongol hordes, but in this day and age, a bit more forebearance and good judgment is required. Good MORAL judgment, that is, which I've already posited, is simply out of the reach of robot soliders, no matter how well-programmed. They're not people, and yes, that's the whole point, but if they can't actually understand the morality we've programmed into them, they can't be expected to use it discerningly to protect innocent lives on both sides. They're just killing machines.

So, if we can take my superior reasoning skills here as scientific fact (which we obviously can), and the robots are not going to save the lives of the innocent, nor are they really at their root being used to save our guys, then what are they for? Besides blowing things up in a science-fiction-awesome way?

I know I'm cynical and will go to almost any length to discredit robots, but honestly, regardless of cynicism, doesn't the idea of a super-powered robot with a very non-functional morality and the ability to kill lots of people in place of soldiers that have historically been trained to do that job, smell just a bit like a placeholder? Doesn't it seem like the robot here would be just an agent that, because it is not technically alive or sentient or capable of understanding the morality with which we have endowed it, can be used to absolve ourselves of all moral culpability for the destruction it will cause? Sure, we made it, and sure, we programmed it, and sure, we sent it into that town with those rocket-launchers, but we didn't actually kill all those people. We can't be held morally responsible if it screwed up. We can't be held morally responsible if it didn't.

This isn't to say that we wouldn't be responsible in other ways, as the government/military behind combat robots would be forced to accept responsibility in a practical sense, but it would sure make us feel better knowing that the agents of destruction we sent in and who blew things up and who got blown up weren't people with a moral responsibility to other people. They're just robots! They must've malfunctioned or something, who knows! But hey, we didn't lose any men, and that's great, right? More champagne!

...see the sliminess in there? I guess I had to dig for it but I think it's just as plain as any gigantic rotting tuber when you find it...

More tomorrow.

Moral machines.

I spent a lot of time in college in Philosophy classes. So much time, in fact, that in my senior year I realized that if I took just one more course, I'd have a minor. So I did, and now I do. Philosophy courses at a Jesuit university tend to have a very religious and often moralistic bent, so I had opted mostly for fun, un-think-y classes like Native American Philosophy and Buddhist Philosophy, which pretty much parroted the ideas I already had about the nature of morality and human nature. But then, when I decided to take the plunge for the minor, the only class I could fit into my schedule that would qualify me was an intensely think-y class called Philosophy of Human Nature. I had a really tough time in that class because we kept bringing up questions that to me were simply preposterous and therefore not worth our time. The subject matter ranged from the agency of the human mind (if we took this person's brain and put it in that person's body, whose personality would be the result? where is the personhood of the person? etc) and the possible person-hood of artificial intelligence. I simply couldnt' see a situation in which these questions would actually become applicable, so I almost refused to do well in the class.

Well, that sure came around to bite me in the ass. I can't seem to formulate proper arguments, or bring up the necessary groundwork of philosophy that has already been laid by hundreds of years of moral philosophizing. Serves me right, I guess. I've been trying to piece together what I have to say about the article I linked to on Saturday about the difficulties that military engineers are having with programming morality into robots. I've been giving the matter some thought, have made extensive notes, and even tried to have a calm and rational out-loud discussion with Connor about it. I had hoped, through these processes, to cement a solid idea of just what it is about the idea of morally reasoning combat robots that makes me want to cry. But instead I realized that, far from reaching one solid conclusion, I have more of a tendency to fly into fits of adrenaline-fueled word-flinging that doesn't make sense. I know why I feel the way I do, but I feel it so passionately that I have difficulty articulating. So. I'm going to try to write down just one of the arguments here, to avoid rambling, ranting, and crying.

Ahem.

After some horrendous errors in the past two years in which combat robots opened fire inexplicably on 'friendly' troops, engineers and programmers have decided it might be a good idea to try to equip battle bots with a "functional morality." They are hoping to successfully use a combination of programming techniques to accomplish this goal: "top down" programming would provide the machine with a set of absolute rules (a la "I, Robot") that could not be undermined, while "bottom up" learning would be implemented, enabling the robot to apply rules to new and changing situations and to learn from past experience (Wall-E, you little bastard). This, scientists hope, would make the robots doing our slaughtering for us more capable to deal with the unexpected and morally dubious realities of war zones, while operating on a basic set of principles that would guide them in decision-making.

Lest you think this sounds cool, and without even trying to describe the creepy-crawly sensation I get when I think of robots 'learning from experience,' let me point out a few things that I know about war zones, gleaned from the watching of several war movies and the reading of several similar books. As they say, "War is Hell," and believe it or not, they say it for a reason. And what can you expect from Hell but a complete upending of morality? While might be the place to find some of the most black and white examples of human morality in action, it is also the place where normal morality often goes totally grey, gets turned on its head, and does breakdance spins. In a war zone, "the innocent" become "the enemy," "children" become "soldiers," "machine gun fire" can be termed "friendly fire," and a zillion other f-ed up things can and do go wrong every moment, turning every situation into a scene of horror that moral philosophers would pee their pants to think of. Particularly in modern war zones, where the enemy is often never seen and thus never 100% identifiable as the enemy, where soldiers fight over causes they hardly understand, and where we're trying to send in ROBOTS to do our bidding, for God's sake, to say that a "functional morality" is the end goal seems a little simple-minded. What could the basic tenets, the basic system of rules in the "top down" model possibly be?

If we're going for a general human morality to implement here, or to form as a structure for the robot rules, the author of our article points out that "there is no clear task specification for general moral behavior, nor is there a single answer to the question of whose morality or what morality should be implemented in AI," and he has a good point. But even if we were to aim for the very most general set of rules that applies to as much of the human population's idea of moral behavior worldwide, we'd be screwed. By its very nature, the set of morals we'd have to instill in a combat robot would be sideways from the beginning. "Do not kill"? Doesn't apply. "Do not kill the innocent"? What constitutes innocent when you're taking over a city full of people who would gladly slaughter you and your entire culture? "Do not kill children"? Modern Africa has proven that children can be just as ruthless as adults. ...And on and on we go.

Human beings, who ostensibly have a much more refined sense of right and wrong, produced organically after years of learning through trial and error, who have a mortal fear of reprisal for acting immorally, and whose pasts are filled with genuine decision-making incidents, still have plenty of trouble determining how to behave morally in war times. We've all seen pictures of Abu Ghraib. So then, in all our vast experience with morality, how can we expect a machine, put together and programmed by us and then taught how to react to situations in very controlled lab environments, to make better decisions than we do? And given their superior speed, agility, and strength (which supposedly is why we'd be using them in the first place - not to give ourselves the impression that WE are not the ones doing the killing and therefore not responsible for it... but that's another argument. Maybe tomorrow.), if these robots do make bad decisions based on their imperfect senses of morality, they have a lot more capability for causing serious destruction.

I have to get going because the gallery is apparently closing up early today, but I will contineu with a much more (I hope) nuanced and deeper discussion of why this is balls to the wall batshit fucking insane tomorrow.