Machines, Morality and Sexbots

the model, Yeah, it's a weird hat for sure. But no, like evil robots and saying film in film, Okay, well, I guess you could say the terminator, but he wasn't really a robot. Well, I mean, yeah, underneath the fake skin he was, you could say Arnol Schwartzenegger. There, Okay, well, I'll let's roll with the terminator thing. All right, so he's he's in the well, no, because he's probably not

a robot. But but yeah, the Terminator. There's a classic example evil robot, right though even in this case, it's the terminators just programmed to perform function. And in the first movie that function is quote unquote evil, And in

the second film he's programmed to do something good. Yes, And so I think that's the interesting thing that we've been talking about when we've been talking to Dr Ronald Arkin at Georgia Tech, and he of course is the man he and his UM research assistance who are responsible for what they're calling in the media the Decepticon, which sounds really you know, brooding and scary, but it's basically robots that have had deception programmed in them so they

can deceive other robots and they can deceive other people. But turns out that they're actually working on different emotions with robots just so that like for instance, like the the Terminator, that they might be able to learn empathy or something like guilt that would help them operate out on the battlefield. So there's probably a joke there about UM some sort of like faith based robotics program like well, you know, because on a certain service of it, why

would you want to make a robot feel guilty. It's like these these bots have it too good, and you know, install some guilt in them, make them feel bad for a change, because we have found out that guilt actually helps us to behave in a way that's more altruistic. Right, Okay, so we're gonna actually listen to Dr or can talk about this a little bit. Well, let's let's talk about emotions, and we do. I've worked in emotions for decades, actually in robot emotions. I worked on Sony and EBow the

small robot dog Curio, their humanoid. We just finished a project with Samsung UH and Billie Mushkeina, my student just successfully defended or dissertation yesterday on time Varying effective models of behavior a complete a span of effect including traits, attitudes, moods, and emotions. UH. We'd like to think it's the most complete model of emotion for any robotics system ever. Today. It's like a palette. You can wonderful emotion and effective

portraits that can interact with human beings. But the real question is does emotion belong in the battle field and what value does it bring? M What value does emotion bring to the battlefield. Some could argue that fear would be a useful one for in terms of self preservation, but anger and frustration and many of the others seem to UH tend to cloud judgment in human beings and lead them towards criminal acts. That's what we'd like to

engineer out of the battlefield. If we potentially could. Now, in some of my work I did include one of the emotions. There are moral emotions which include empathy and compassion and alike as well. Too. One could argue that those emotions are already legislated into the Geneva Conventions. So to some extent, if you adhere to that, you are being empathetic. Uh, you're not. You're understanding the distinction between the civilian. You're making sure that you're not applying unnecessary

force in different sets of circumstances. You're making sure that when you kill someone it is done in a way which is not considered uncivilized. So in that sense, we don't have to incorporate that motion directly into the ROADBOT if and it follows those uh, those rules, it potentially can. But what we did incorporates a different moral motion, which was guilt. Interestingly enough, uh, we use guilt as a mechanism by which the system could reduce the level of

force it uses if it doesn't fully understand it. So let me give an example. Um. One of the things that's done prior to the deployment of a weapon is a battle damage estimate. So if you were going to drop a bomb in a particular area. You would have to have an estimate. This is religious proportionality on other things. Uh, you would drop this weapon and you would expect such

and such to occur. Afterwards, you do a battle damage assessment. UM. So suppose, for example, an autonomous system, say an intelligent Reaper or something like that, the unmanned area vehicles they're using now, which are under human control at this point in time, made a decision that it was going to drop a weapon, and it did a battle damage assessment. It calculated the collateral damage it could occur. There's important

things to understand. What you're disturbing to many people about the tolerance of civilian casualties and civilian deaths, which is actually a part of warfare, but that that dates back to the Middle Ages and the principle of double effect, among other things. You're not a war criminal if you kill a civilian. Your war criminal if you intentionally kill a civilians. And there's fundamental differences. It's very hard to expect,

inspect a human mind to be able to tell. So it's really interesting that he's talking about this model of emotions for robotics system UM and even that the whole point about how you don't necessarily need to program empathy and compassion if the robot can follow the rules of the Geneva conventions, right, so the protocols of it. So if if the robot follows that, then you don't necessarily need to to get into the weeds with empathy at least in that case, right, if you've got a robot

but that is performing specific functions out on the battlefield. Um, but I wanted to look a little bit more at that Smith's and Debot componential I R T model for guilt, because this is the cognitive model that they used for their robots. And they said, okay, let's look at a good model of guilt and how to program it. So that model is actually they studied the process of the structure of guilt and this what was administered to two hundred and sevent equal students, and the finding show that

this kind of modeling is appropriate to investing through other emotions. Right, But what you're using teenagers? Though? Oh yeah, the the guiltiest humans on the planet, are you? I mean, yeah, all right, I'm just a little I don't know about using teenagers to program robots, you know, like, oh right, right, Well, I mean you know, they were not you know, saying like let's take their skill level at driving, or you know, their their seat of judgment reasoning here and instilled in

a robot. I can just imagine robots becoming very moody, having crushes and and and just all sorts of ridiculous stuff. Yeah. Yeah, I mean, you know what, I'm sure there's a market for that robot somewhere out there. Um. But there were

five different components of guilt, and the first one. I'll just kind of run through these quickly because actually, there's a really great paper on This is very long, so I'm sure that nobody wants to hear uh twenty page paper described here, So we'll just get to the meat of it, which is the first The first condition is that guilt implies an appraisal in terms of responsibility. Okay, Okay, So, for example, guilt only appears in situations for which one

feels personally responsible for. Okay, So that's like, Um, I was supposed to walk the dog and then the dog pooped on the living room floor. Yeah, that's kind of on me. Because I was supposed to walk the dog. That was my responsibility. I feel guilty. Yeah. The poops on you. The second is that guilt implies an appraisal in terms of norm violation. A violation of a norm or the moral order proceeds guilt. Yeah, dog poop does not go in the living room floor. Right, that's a norm.

That's that's a norm and that norm has been defied. Right. Guilt implies the third one, a negative self evaluation as a covert reaction of the type I did something bad. So it's the negative self of that evaluation relates to an act, and it's not a definite disapproval of the entire self. Okay, so it's like, hey, I'm a pretty good guy, but man, I should have walked that dog so it didn't poop on the living room floor. Yeah.

So I mean you're not going into a shame spiral, right, So you're saying, gosh, I could I could probably be better if I didn't do that. Exactly. I'm not losing sleep over it. It's not a crisis, but to self need to get better at walking that dog. Yes. This presentation is brought to you by Intel Sponsors of Tomorrow. Number four is, while feeling guilty, one's attention and inner thoughts or covert ruminative ring reactions focused on the act

much more than on the self. I really hate that that dog pooped on the floor, and it's my doing, right that we're looking at here. Yeah, well, and I was also kind of thinking about Lady Macbeth in the

super crazy way. She's not a norm obviously a guilt but outdamn spot right, she's obsessed with the blood on her hands, the imagined blood on her hands, and and that's kind of the She's not doing a lot of introspective work right, right, She's just turning it over in her brain over and over in and then the fifth one is that guilt implies the motivations and action tendencies related uh to ourselves, So one is inclined to confess, to undo one's fault and try to write what her

what was gone wrong with an apology. Okay, I'm not sure how to relate that to the dog poop though, Well, I mean I guess you could say, especially if this is the space on teenagers, right, like, sorry, um, I will not do that again. I will take the dog out. Okay. Yeah, So I guess that's the acknowledgement part, right, So these are that's the actual cognitive model that they used to them program the robots, which is pretty fascinating. Right, So

you see that play out. Obviously, this is an algorithms you know, it's a if if in then situation. Um. But nonetheless, here we are trying to create some sort of morality in robots, right, even if it's x than y yes on carpet, then sat exactly exactly. Um. But it was really interesting because Dr Arhin was talking about this um, and then he was talking about in a larger context, we need to really be thinking about robots

and our connection to them. So it's not just okay, we'll use them out in the battlefield, but how do we connect to technology as a whole. So he had a little bit to say about why we can't help but connect to technology even though it's not another human. So let's hear that. But the real key for us is I have been concerned in robotics from the very beginning with the behavior of intelligence systems, and uh, these kinds of things affect behavior. Fortunately, now we have architectures

where we can create and compose autonomously robotic behavior. So these systems can do things in the context emissions or in the context of your home or whatever you like. And in some cases emotions are very useful. If you want a robot companion, you would probably like it to be able to express emotions so you could train with it and but feel that it's more a part of you.

Remember the Tamaguchi's that people used to have, the little watches that people would cry when they didn't feed them, and they died in the In all sorts of things, we have this propensity of human beings to create bonds to artifacts extremely easily. Uh, it's very natural. Cliffness studied this in his book The Media Equation UH and documented it very well that even if we fully understand that

this is an artifact, it doesn't matter. And what's interesting is that roboticists, and this begs many different other ethical questions, we can create these artifacts in ways that we can make you want them. Uh. Same you go to a movie, right, what are you watching? You're watching a bunch of dots on a screen. Uh, And you walk out there crying, laughing if it's a good movie, and you'll have paid for the privilege of doing that, And you're being manipulated

all the time. Your emotions are being an it's like a ride, right, Well, a robot could have that same capability, but it can follow you around and it can be more a portion of your life on a daily basis, So that actually changes things to some degree. This physical embodiment actually alters the equation in many people's minds, so that there is an extra level of concern that we need to address as we move forward with this technology. So again it comes down to the fact that our

our emotions can be manipulated by anything. Yeah. I loved it when he was talking about the movies. He was just saying, look, this is sort of pixelated light and you are crying, you know, um, because we can't help but see ourselves in all those different situations. You know. I guess it's like the rorshot test of our existence. Yeah, it's like the we've I've mentioned before, the bit from Community where Jeff Winger points outther you give a pencil and name and then you snap it in half and

then we feel sad. You know. It's like we can we can just become emotional attached just about anything. Yeah, we ascribe meaning to everything. Um. And we've talked about Sherry Chuckle before too. She's the psychologist who worked at M I T for many years and and adults. Yeah, she she developed a crush on Cog, the lab robot, and found herself wishing that she had more alone time with Cog in the sense that you know, the other her her co workers were maybe bow guarding some time

with with Cog. Um. Not that she wanted to get intimate with Cog. And we're making her sound crazier than she No, No, she's actually she has some really interesting things to say. She has a book out called Alone Together, um, and about how again we're connecting with technology and we're making the connection, but we're all doing it alone together. Um. But so you know, actually dr Arkin brought her up as well, because she was going to be talking at

Georgia Tech the week that we spoke to him. And then he also brought up the fact that he teaches robots ethics in terms of intimacy to his class and robots and society. And of course you know who came up during the XY with three exes. Well, hey, let's hear what dr Arkin has to say about Foxy Roxy and William's dissertation which I was mentioning, uh that she very successfully passed yesterday one of the studies. We did lots of human robot interaction studies, and one of those

was having a robot in a search and rescue mission. Uh, start when a sudden event occurs, start to give commands to a human being to evacuate, to get out of the room. When we did it without the effective component, there was very little compliance, actually none. Uh. Well he added the effect people started moving when the robot told him to go in that case. So the very interesting dimensions. Uh.

And you don't even need that in many cases. But the point is that once the more and more we understand human intelligence and human feelings, and the more and more we put them into these systems, the more prone we are to fall in love, to care, to whatever with these particular devices. And that begs the next question, which I'm talking about with my class this week as well too in robot ethics. It is uh, intimacy and how far you want to go? What is socially acceptable?

What is societally acceptable? Where are we going to draw the lines? And very very very few people are we wing to approach that subject. It's it's very interesting. There's virtually no academic research whatsoever going in the intimate level, the deeper levels of human robot interaction and robot sexuality. Uh, there are no funding agencies that I am aware of, it would dare to uh fund in that space. You can imagine the repercussions that that could occur. But there

are people doing it. I was just saying, yet you know there's a market for it because they've already Yeah, it's exactly right. It's kind of like what the pornography industry is being done in people's garages and warehouses and the like as well too. There was the so called first sex robot called roxy r o x R. Yes, exactly. Oh boy, that's a bad robot. Sorry, I wouldn't even

call it a robot as well too. And the point is you can make claims about things which are completely unfounded because people don't understand the the ways of which humans relate to these particular artifacts. And if you're abhoard by the mere thought of that, that's okay, But then what you're gonna do about it? I mean, are you going to provide guidelines, restrictions, regulations for the conductor research?

Right now? There are no such guidelines of restrictions, So there's just a social pressure, as Lessig would talk about in this context. But we may uh, and I also have to blame our own profession for not coming up with much in the way of regulate, regulating the way in which we do things. It's still it was a

real cow cowboy cow girl field when I first got started. H. Now we have much more effective scientific measures for evaluating results, but we still haven't got to the UH bioethics community as well too in those aspects, although we do and I am a member, a founding member the technically Tripoli Technical Committee and robo ethics. But it's getting stronger. People

are understanding that we're succeeding. That's the scary song. You see, there were actually succeeding in making these kinds of artifacts and the consequences of them. Uh, we don't fully understand. So Roxy a bad robot, Yeah, a bad robot in Roxy's case. They're taking their uh sort of well less successful attempt at a healthcare robot and turning it into a into a sex spot for to make a few, you know, to to to generate a little more revenue flow, so to speak. Yeah. Yeah, but but yeah, so's the

issues out there. We we we need to be thinking about it. Uh and and uh, yeah, I don't know if it means we need to found like a you know, the Sex Spot Institute of North America or what right? Right?

And actually I think this isn't included in the in the audio, but I did ask him about whether or not these issues surrounding intimacy with robots or something that are slow going in the United States and are are are more talked about in Europe for instance, And he said, well, you know, like a good scientist, I don't have any data so I can tell you that, but I can tell you that people started talking about it in Europe, uh, far earlier than we are now talking about it in

the United States. Um. And that there are all sorts of issues that they're talking about with robotics in terms of like even like robotic spare parts. If you were to have you know, say an exo skeleton arm, um, you know, who should get that arm who should have

access to these enhancements essentially? Um? You know? And and I said, oh, that kind of makes me think that there could be this black market created for some of this technology if we don't now start thinking about how we want to release it to the public or we

want the public at large using technology. UM. And so when you talk about things like that, like a exo skeleton or a robotic arm, then you're really talking more like people who might use it in terroristic acts, you know, or to to bolster those sort of um, terroristic acts, if that makes any sense. But not like the building of a sex spot in their basement. Well that's that's

a concern too. But he was sort of saying, this is all a part and parcel that that you know, in in Europe, some of these other considerations have been going on for a while. UM, have the have nots, who would have access to it? How would it be used? You know, what's what's the best use of this technology? Um? And he also brought up another really interesting point about you know, really looking at the situation and trying to figure out are we we being sensitive to it in

the right ways? Are we really listening to all of the voices, um that are out there, including neo Luddites. Yeah. And it turns out that he actually has his class study Ted Kaczynski and UM in some of his writings in terms of neo Luddites and what they can learn

as programmers about the technology they're creating. Uh. And of course, for those of you who aren't familiar, Kazynski was of course the Unibomber, famous for the Unibomber Manifesto, which is another documented is that is far too long to read, and it's entirety here. Some only gonna read half of it. Um No, I'm just gonna read a couple well, just a brief excerpt from it here, just to give you

an idea in case you've never looked at it. It starts off like this, The Industrial Revolution and its consequences have been a disaster for the human race. They have greatly increased the life expectancy of those of us who live in advanced countries, but they have distabilized society, have made life unfulfilling, has subjected human beings to indignities, have allowed to widespread psychological suffering in the third world, to physical suffering as well, and have inflicted severe damage on

the natural world. And it kind of goes on from there. And there's is kind of a downer about about technology in the modern age and where it's headed, and it's it's just a very grim you on how technology has changed life on Earth. Yeah, and just just for a little refresher for everybody to He actually sent sixteen mail bombs to various targets at universities and airlines, um, and

an attempt to get his message across UM. So obviously this was someone who is not uh playing with full deck and was had some interesting things to say, but was more psychotic. Yeah, and um, you know, killed some people and it was it was not a good thing. But you know, again that being said, it was interesting to to look at the material I assume of a Neil Ludite and try to find points there, points of

consideration at least. I mean, it's like with politics, even if you're more of a middle to line person, it pays to at least glance over and see what the the extreme opinions on either side are thinking about. Yeah, well, let's hear what Dr Orkand has to say about that.

There are people all that. Bill Joy, in his article why the Future Doesn't Meet Us, which was published at the beginning of the millennia I believe it was in Wired, talked about g n R genetics, nanotechnology and robotics due to self replication as leading to the extinction of mankind and the solution that he advocated is we should relinquish all research in that particular space. Most of us think

that's a little premature, um. But if that is the threat is as great as he argues, it's something we should continue to discuss. Um. So I just want to put that on the table as well. And actually his inspiration, uh was the UNI Bomber for much of his work as well too in terms of reading what Kazynski wrote, which actually encouraged my class to take a look at

as well too. Not because I admire the man, far from it, but rather a neo Luddite has a perspective that must be considered in terms of what we are potentially doing as a society. And everyone implores the means that we're used in that particular case, but nonetheless we have to be aware of the technologies that we're creating, what the potential effects are on our species and our civilization.

So yeah, Univirment's Manifesto as a classroom text, Yeah yeah, And again I think it's really interesting that he introduces that just so that his students would be aware of the issues. And it was making me think about confirmation bias and why we, you know, with our confirmation bias. As humans, we can't help but continue to seek out supportive material to to help us come to the conclusion we want the conclusions we'd like to come to in life, right,

we typically do that. We typically don't seek out stuff that that makes us wrong about something. Right. So, actually, Leonard Millot knew, uh Milo. Now he is the author of The Drunkard's Walk. He has some interesting things to say about that um in that book about confirmation bias. And here's a quote. It says, to make matters worse, not only do we preferentially seek evidence to confirm our preconceived notions, but we also interpret ambiguous evidence in favor

of our ideas. This can be a big problem because data are often ambiguous. So by ignoring some patterns and emphasizing others are clever, brains can reinforce their beliefs even in the absence of convincing data. So again, I think it's it's very interesting that the class would look at that text. We'll try to find something in it that that might have a kernel truth to it. Yeah, I mean, it's like anytime you see somebody who say goes down the road of conspiracy theories, or or or even just um,

let's say one particular political ideology or another. You know, if they start just consuming just one type of media about it, like all they read or books by this particular author and their um their comrades about a particular movement or or set of ideas. Then that you know, you are what you eat. You you sort of streamline your your your brain on that particular topic. So it Yeah, it pays to have a little wider viewpoint on things. Yeah,

food for thought there if you are which what you eat? There? Um, I want to thank Dr Ronald Organ again for taking the time to talk to us about robotics and ethics and uh, we really appreciate it. So, hey, what do you think about the future of social interaction with robots? About guilty robots or sexy robots? Let us know we're on Twitter and Facebook, both of those as below the mind and you can always send your thoughts to blow the mind at how stuff works dot com for moral

on this and thousands of other topics. Is it how stuff works dot com. To learn more about the podcast, click on the podcast icon in the upper right corner of our home page. The house stuff Works iPhone app has a ride. Download it today on iTunes,