Soldiers and killbots: a love story

Some of you might be anticipating the day when robots are capable of engaging in interpersonal and perhaps even romantic relationships with homosapiens, but it may surprise you to learn that there are already deep connections being made between carbon and silicon in the unlikeliest of places: the battlefields of war-torn Iraq and Afghanistan. The Washington Post has an interesting piece on the bonds that US soldiers have been forming with their Packbots and other autonomous companions, christening the metallic team members with names such as 'Scooby Doo,' 'Frankenstein,' and 'Sgt. Talon,' anthropomorphizing them with drawn-on faces, and bestowing them with medals after successful completion of a mission. We're even told at the beginning of the article that WowWee founder Mark Tilden was once showing off a multi-legged mine-detecting bot at Arizona's Yuma test grounds, and while the prototype in question was pulling itself along on just one leg after having been battered and dismembered by numerous detonations, the Army colonel in charge abruptly put a stop to the test -- calling it inhumane. Which brings us once again to the topic of robot ethics -- whose tenets are already being codified in Europe, Japan, and South Korea -- and the inevitable issues that will arise as the Asimos and Ever Muses of the world get even more emotive and lifelike: what rights and rules do we bestow upon our planet's new cohabitants; at what point do we determine that they are completely sentient; and most importantly, how do we defer for as long as possible the inevitable uprising that any sane-minded person knows is coming?

[Via Gizmag]

[Thursday @ May 10th 2007 10:45PM]

“1,500 years ago, everybody “knew” that the earth was the center of the universe. 500 years ago, everybody “knew” that the earth was flat. And 15 minutes ago, you “knew” that humans were alone on this planet. Imagine what you’ll “know” tomorrow. ” Agent K, Men In Black.

Speaking technically if the Universe is infinite, any point can be deemed to be the center.
As for everyone thinking the world was flat, read your Aristotle, then spew that BS to me.
~G

[Lee Roy Brandon III @ May 10th 2007 11:47PM]

Movie quotes? Come on, bring some real ammo. People have known the Earth was round since at least the time of Aristotle... construct a real arguement that hasn't been "proven" by Hollywood.

[DerekPowell @ May 10th 2007 8:53PM]

Regardless of your feelings about armed conflict, it's very strange that a man whose job it is to kill people feels destroying a non-living (though animate) object is "inhumane." I guess it's sort of like when everyone feels sad for the dog that dies in movies, when dozens of people have just been slaughtered in the same movie.

[Brian @ May 10th 2007 9:55PM]

So does this mean that if you rip a robot's power supply from it's cold chassis you suddenly are some sort of crazed killer?

"Inhumane" will always be an emotionally charged gray area, and there's no point using logic when you're discussing emotions.

C'mon how many of you have had to explain to your girlfriend about why you had to kill the Big Boss in some video game, and she got sad/upset when you did?

[Grant @ May 10th 2007 10:30PM]

"at what point do we determine that they are completely sentient;"

Never. Even if an AI could pass the Turing Test, all it has done is fool us. That is all the Turing test looks for, the ability to mimic Human intelligence. There is something incredible that Human's can do that no computer program will ever be able to do, and that is to question its core programming (i.e. philosophize). A computer program is fated to never do anything but what its programming (and by extension what its programmer) tells it to do.
~G

[Ryan @ May 10th 2007 11:13PM]

You're thinking too small. CURRENT computer designs can't develop and think, but I can guarantee that we'll see computer designs that can change and develop in much the same way that a human brain does.

[barnz2k @ May 10th 2007 10:41PM]

id like to see rights for animals with-held before they try and dish em out to robots!! come on.

[Thursday @ May 10th 2007 11:27PM]

It is not the design of the computer I fault, it is the programing. There are quite a few other reasons that AI will never achieve sentience, but most of them involve quite a bit of symbolic logic. The simple truth is, the Human mind is unique, Socrates knew it, Plato knew it, Aristotle Knew it, Aquinas Knew it, Spinoza Knew it, and Descartes knew it. I can pile an infinite number of stones together, give it a jolt of electricity, and they will never think. If they did, the whole would be more than the sum of it's parts. There is something else, something "other" that must be there. Observation tells us there is nothing "missing" from a dead body, but reason tells us there must be.
I could go one for pages about this, but I think I'll save it for a term paper.
~G

[Chuckles McGee @ May 10th 2007 11:44PM]

The human brain, though incredible, is nothing more than carbon, hydrogen and a few trace elements. That's a huge understatement, but nevertheless, there's nothing "magical" or unique about a brain- it obeys the same basic laws of mass, energy and follows the rules of physics, just like any other human organ. Complexity does not equate to an impossibility in terms of emulation, only a difficulty. And a whole can perform more than the sum of it's parts- a car perform something that tires, an engine and gas cannot do separately. And whether philosphers claim to "know" something is logically independent of whether or not the thing itself is true. Whether or not a hunk of "thinking" silicon has ethical rights is still entirely up for debate.

[zoara @ May 11th 2007 10:49AM]

@ Lee Roy Brandon III:

I love the literalism exhibited by Engadget commenters. Think about the point RandomThoughts was making, not the method used to make that point.

At some time in the past, people thought [insert idea here], but we now accept that idea as ridiculous because we have made many discoveries since that time.

Why will this be any different when people look back at the year 2007 from a hundred, a thousand, or ten thousand years in the future?

Maybe someone will be having an argument on the Engadget of their time and point out how funny it is that in the early 21st Century people still argued over whether artificial intelligence was possible. Haw haw, those ignorant people from the past with their constant use of primitive 'computers' and 'cellphones'.

[Ari Moshe @ May 11th 2007 12:10AM]

man, the picture got my hopes up that it was going to be about Futurama! damn you all!

[miffe @ May 11th 2007 12:48AM]

There are no boundries to this universe, every limit is to be broken, even gravity has been 'defied' in certain ways...

Why are you so certain that AI will never achieve the 'uniqueness' of the human brain, all you are is a living paradigm, and I will show you wrong after completing my MsC in AI in two years...

[DerekPowell @ May 11th 2007 1:43AM]

well, "never" technically could apply to what Grant is talking about. If in your philosophy you accept some form of dualism, some division between body and mind, then he's correct. Robots will never possess that crucial thing that is the mind. If you're a reductive materialist (which I'm not) you already think people are just like robots, so the job's already done.

[snafle @ May 11th 2007 2:36AM]

Why will robots never possess that crucial thing? What is this crucial thing? You don't know, therefore you can't predict anything based on it. What passes for thought and intelligent action *will* be able to be displayed by a robot, given time.

You say it's programming will limit it- let me ask you, what's a baby? A bag of flesh with a few very simple imperatives. It's got the ability to learn and extend these, to become one the intelligent (or not so) adults surrounding you. A baby has the intelligence of a chimp baby- are you saying they have the certain something? The human mind is just a more complicated version of the same thing. To imply the human mind is anything special, that it somehow contravenes the laws of physics is excessibely anthropocentric. It's like the old Earth, Centre of the Universe thing- people want to think they're special, but they aren't. They're just one more piece of matter.

[segovia101 @ May 11th 2007 5:28AM]

Half a mouse brain is difficult to simulate for a period of only 10 seconds at a speed of one-tenth real-time:

http://www.engadget.com/2007/04/29/ibms-bluegene-l-supercomputer-simulates-half-a-mouse-brain/

I believe that a computer program will, some day, probably be able to convince a human that they are conversing with another human. On the other hand I doubt that if the population of Earth wrote code, as we understand it today, from now until the end of the Universe would they create a program that could be regarded as sentient. Even if such a monumentally complex piece of engineering could be planned, managed and implemented we'd still need an enormously fast computer to run it on.

BlueGene/L can sustain 380 teraflops as per it's wiki. The highest current estimate is that the brain can achieve 100 petaflops (1 petaflop is 1000 teraflops). Computers have a way to go but no doubt it will be achieved. But that's probably a pointless comparison in any event; does the brain recognise that a flower is of 'type = rose' by performing floating point operations? Try programming that in Visual Basic.

Try to write a program that can take Othello or David Copperfield or The Dummies guide to AI and understand it, assimilate it and relate it to it's previous knowledge in order to be able to use it's new learning in the future. Where would you start? Multiply that problem by a billion and solve them all and you would still not be close to a sentient computer/program. Methinks it's a stretch too far.

[Sinbios @ May 11th 2007 6:17AM]

You still don't understand that technology evolves exponentially? A hundred years ago nobody would have DREAMT of the technology we possess today. There is NOTHING that's "a stretch too far"; nobody knows what the limits of technology is.

A human brain is nothing but a bunch of neurons that shoot signals at each other. There's no reason it cannot be replicated by technology. Unless someone manages to quantitize the "soul" or the "human factor", it's just a figment of the imagination created by people who want to feel unique and superior. And if it IS one day quantitized, then there's no reason why it can't be artificially recreated.

[Tim @ May 11th 2007 7:42AM]

I can't believe how narrow-minded and naive some of you people are, especially those that think they have the right to predict future technology.

You gain a soul as you learn and grow up in the world. The brain is just a very adaptive type of computer, and when silicon computers are designed to be adaptive enough, we'll end up with digital people.

[sendeth @ May 11th 2007 8:01AM]

the part about the soldiers bonds is like a pilot and his plane, a race car driver and his car. the officer in charge put a stop to the test because of the fact that the robot is still viewed as a soldier. unlike a hummer that just sits there and makes no decisions. i do agree that there will be some big issues that arise from this, but i hardly think that we are heading to the matrix, terminator. probably more along the lines of bladerunner or ghost in the shell. we will begin merging with machines long before they could "take over." sometimes i think that's what people want anyway. just let the robots do it all so they don't miss american idol.

that's not to mention the "ghost in the machine" theories. in one study (berkley maybe???) one robot committed suicide by shutting itself down because it decided that everything it did was wrong.

about the "uprising"......

"The function of science fiction is not always to predict the future but sometimes to prevent it."
-quote: frank herbert

[mymoustache @ May 11th 2007 8:51AM]

I'm actually from the future, so I'm really getting a kick out of these responses...

[cole @ May 11th 2007 9:01AM]

I loved



I loved EOD's robots when i was in Iraq. A robots face will never replace my team leader who has a family at home.

Let the soldiers have fun. Christ most of them are 18-26 years old giving their life's for you freedom.

P.S. F. the crap that is going on right now. Give my boys the money they need so they can come back home safe to their families

[Lee Roy Brandon III @ May 11th 2007 11:58AM]

I understood his point, I just wanted to hear him put forth a stronger argument based on facts, rather than continuing to repeat common misinformation. That shouldn't be hard...

[Mike Van Plew @ May 11th 2007 10:57AM]

Wow! for something about intelligence of ROBOTS you guys got off on the human side of pretty quick! A robot should be just that a device of and for service to humankind, No, No rights to robots or any other type of machine. They are simply a tool, nothing more.

As for what humans have that makes them special, maybe you will remember that at least humans have souls or a spirit per say, in any account you seem to always forget about GOD and his divine SPARK of life that we all know that is within each and everyone of us! Somehow I doubt that a robot will obtain a soul or a spirit at anytime.

[zoara @ May 11th 2007 11:17AM]

Mike, you're a nutter.

[marcoemerson @ May 11th 2007 12:11PM]

The thing the mind can do that I can't imagine a computer doing is understanding symbols. Computers will undoubtedly surpass the processing ability of the human mind, but it will always come down to x+y=z. The beauty of the human mind isn't in the ability to calculate these things, but in the ability to understand what x and y are, and WHY z needs to be calculated in the first. Computers will never innately value anything. The only things computers value is what they're explicitly told to value. They'll never have instincts. The lack of ability to understand symbols are why electronic translators are so terrible at their jobs. Language is about ideas and concepts, and each language may have a completely different syntax to express the same concept. It is a function that is hard to achieve with human mind. We're still working on statements like "All your base are belong to us." I'm a programmer, not a philosopher, but I still can't see how we can build something as complicated as our own mind.

[AdvidG @ May 11th 2007 1:41PM]

Compassion.

Almost... no other animal or insect displays this trait.

Some have risked their own lives to save complete strangers or pets.

I think people form emotional connections to things around them...to help cope.

Otherwise it's just back to fuck over, up, down anything in your way.

[cybereality @ May 12th 2007 3:09PM]

Does anyone ever wonder if maybe we've done it already? Maybe the internet *is* alive and is actually sentient, but we just don't understand what its saying. Is the human brain really all that different from the net? At its core, they are both just a bunch of electrical signals flashing back and forth. If you think I'm full of it, please google "emergent intelligence" and maybe you'll learn something.

[Mike @ May 13th 2007 9:41AM]

Theres a lot of hollywood 'matrix' inspired thinking here. If a computer were to be designed that could emulate all the functions of the human brain, why couldn't it feel compassion or fear, or philosophize on the meaning of its own existence? We desperately try to present some pseudo-spiritual argument about how we are innately superior, having some divine spark. Computers as we know them are currently cold, logical, and basically stupid when it comes to real world situations, but huge strides are being made in neural nets, artificial cortex's, and other forms of AI. I have seen a robot capable of fear, panicking when threatened. But this just at the beginning, AI has just got started, but its getting smarter by the day. We are not. There will be, at some point, a catchup, an overtaking and a racing ahead, not just real intelligence, but superintelligence, and I'm betting within my lifetime, its not science fiction, its obvious to anyone aware of the pace of progress in the field. If anyone can present a coherent argument about why this is wrong, without invoking the bogeyman, Flying Spaghetti Monster or other unprovable nonsense, I'd like to hear it.

[segovia101 @ May 13th 2007 11:56AM]

You know, that really grinds my gears...

One robot capable of displaying one programmed 'emotion' does not a human make. It's true that technology is moving forward at an incredible pace. What this amounts to at the moment in terms of hardware is an increase in processing power and an increase in storage capacity.

I don't believe in God so that doesn't enter into my argument. All I'm saying is that I don't understand how anyone can believe that, using current programming methods, a sentient being can be created. In terms of improvements in programming all we have really achieved are IDEs that allow programmers to produce programs more easily, at a higher level of abstraction and without the need to program at the machine code or binary level. So there is no exponential improvement in this field.

There is no exponential improvement in writing software in teams; we've just gotten slightly better at managing the process. Software development can take years, involve many people and be very expensive. Even if it was possible to write such a program I don't believe it would be truly sentient. Could it ever have a new or original thought?

In a previous comment I made reference to the mouse brain simulation: http://news.bbc.co.uk/2/hi/technology/6600965.stm

Writing a program to simulate a brain and then feeding it stimuli to see if it develops or learns is, in my opinion, a better approach than trying to create a traditional program trillions of lines long that simulates someone with an IQ of, say, 90. I concede that such a simulated brain may be capable of becoming sentient. To write this simulation all we have to do is develop a perfect understanding of the human brain's tiniest intricacies and describe it in a model. Of course when we do have a supercomputer thousands or millions of times more powerful than BlueGene L which can run this model, all we have to do is figure out how to make the simulation learn and develop into an intelligent adult and then fit this supercomputer into a humanoid sized robot.

As for 'Matrix' inspired thinking, I'm guessing that a lot of those who believe that sentient robots will someday be a reality have their unshakable belief inspired by movies like Terminator or the Matrix. The skeptics are the ones who may enjoy the movies but can differentiate between what is possible in reality and fiction.

Alot of significant developments are pushed by the military and then adopted by business later. I can't see why they would want a sentient computer or robot? Aren't robots just cannon fodder? Given the shift away from infantry based warfare I can't see the need for battallions of superintelligent robots marching off to war. So who pays for the research?

Maybe in 3000AD or 4000AD or even 5000AD it may happen I hear you cry? I think that the chances of the human race even being around at that point is a little optimistic; and I DON't mean because of the robot uprising. Has no one seen Mad Max? We won't have the infrastructure.

[Mike @ May 13th 2007 5:05PM]

segovia101 -
I'm sorry I caused your gears to grind, might I suggest a little machine oil ;)
Just to clarify - my last entry was written in haste, without any editing, a stream of consciousness if you will.
I don't remember arguing at any point that we could achieve artificial intelligence with conventional programming techniques, or simply faster processing. I'm a programmer too, so I certainly appreciate the limitations of the linear processing that is the norm. Neither did I suggest that a robot that could display one emotion was even remotely human (though its behavior was remarkably organic compared to the others on display). I may not have made myself clear when I mentioned that technology was moving at a rapid pace - I wasn't referring to the increase in storage and processing speed (although these are important starting points in complex systems) but the developments in the last 5-10 years in the field of A.I, such as recognition and self-learning software, genetic algorithm, et al.
I had already read the 'mouse brain' story As a person obviously familiar with software development, you will appreciate that emulation is vastly less efficient than native systems, and this experiment, cutting through the hyperbole, has little do with A.I science and more to do with neuroscience. It is not necessary to replicate the exact architecture of the brain cells, merely their functions, inputs and outputs. I don't think we differ too much here, once you strip away the straw-man arguments you put in my mouth (or keyboard, whatever)

I did find your 'development' points particularly weak however.
'Alot of significant developments are pushed by the military' - No doubt, but certainly not all, and likely not the majority.

'I can't see why they would want a sentient computer or robot?' - Sentience is particularly hard to define, at what point do you say a robot, animal or even another human is sentient, rather than intelligent?

'Aren't robots just cannon fodder?' You could argue the same of soldiers. IMHO, any agent, human or otherwise, should be capable of making moral judgments when wielding a lethal weapon.
As a matter of fact, the military ARE very interested in robotic warfare, for all the obvious reasons. Stands to reason that a robot designed to do the job of a human soldier needs a high level of intelligence.
Oh, and robots don't need a salary or breaks. A robot that can function at a human level is desirable for industry, government and the military.
There is clearly a drive, though, for both practical and academic reasons, for more human-like machines, and we will not be satisfied till we achieve that. Nature did it through trial and error, I don't see why we cannot, through solid engineering and a sense of direction. Just do a little research, from experts in the field, not just Engadget, or even Wikipedia! I spent a day in an A.I lab last month, and going back to study there as a mature student. I'm not dreaming up science fiction scenarios, the ball is already rolling!