Impossible to teach robots to love – robotics inventor
While the idea of humans and robots walking the streets is still just a dream, in reality, robotics is already taking over our daily lives. Will we be able to create thinking robots in the future, and ultimately, is that something we really want? We talked to Peter Redmond, engineer, robotics guru, and creator of the intelligent RuBot robot.
Sophie Shevardnadze: Peter Redmond, it’s so great to have you on our show today, thanks for coming. Let’s start with artificial intelligence. Stephen Hawking said that full artificial intelligence would spell out the end of humanity, with Elon Musk has invested $10 million into artificial intelligence safety. So what do you think, in which way would artificial intelligence threaten the human race?
Peter Redmond: Well, it’s these three main trains of thought on artificial intelligence and what we should do about it when it reaches the consciousness of humans. The first one is that we will never get there, so why bother? Just carry on as we are going. The second is that we will get there and it’s the end of humanity, so why bother doing anything about it? And the third one is we don’t really know what will happen when we get there, so lets make some plans and put some safety measures in place, so that when we do get there at least we are ready.
SS:What’s your take? Which one of the three lines of thought do you agree with?
PR: I agree somewhere in the middle of the two last ones. So I think it’s possible, theoretically possible that we could get there somewhere and we should have some sort of measures in place when it does eventually happen. But I think it’s going to be a long time.
SS:But, I mean, we keep saying “a long time” and there are so many technological breakthroughs and progress has become so fast in the last 30 or 40 years, especially the last 15 years. So does it worry you or are you excited about it?
PR: I am excited about all the small progresses that we make, but I also know there is a lot of artificial intelligence that, we believe, is much more intelligent than it actually is. It is a simulation of intelligence. Robots and the artificial intelligence are very good at doing one thing at a time. Our brains are good at doing lots of things at the same time.
SS:So you think that by 2050 there are going to be robots strolling around on our streets and in the cities – how do you picture that happening? Or are they just going to be robots walking along with us, going about their business?
PR: I think that by 2050 which is the challenge days where we would have robots that could do many things – right now they only do one thing at a time really well – by 2050 they should be able to do lots of things at the same time. And at that time we will be more comfortable, I think, with robots and calling them robots. At the moment when we have robot that does a job very well we call it by the job it does, for example washing machine is in old essence a robot, it has all of the criteria, it has sensors, it does work for us that we don’t want to do and so on. So washing machine is a robot, but we don’t even consider it in anyway a robot, it’s just a machine that washes stuff for us.
SS:Ok, so robot in a sense that we, the regular people, perceive it, in that sense - what is the most advanced robot that you’ve created, what can it do? I would call it a she or a he, but what can it do?
PR: Yeah, it’s really easy to anthropomorphize robots. What I work on is specifically algorithms and vision algorithms in particular. So I teach robots how to see and how to see the world and that’s a very difficult task for a robot. It is very easy for us – when we walk into the room we can instantly say where we are in that room and where is the rest of the objects are in that room and what the functions of those objects are. These concepts are very difficult and you could show, you could teach an artificial intelligence what a chair looks like and several chairs, and then you can show a new chair that it has never seen before right now and it could pick out and say: “That is a chair”. But if you show a series of things and say: “Which of these is for sitting on?” - that concept is too abstract right now for artificial intelligence.
SS: So if you could create a dream robot what would it be? I’m sure you thought about it.
PR: A dream robot? That’s a difficult question. It would depend on my needs of the day, I guess. A robot that when you ask it to do a function it can understand what it is that you want it to do rather than having to particularly specify. If you point at something itself and say: “I want one that thing over there”, - for robot that is so abstract to understand what that gesture of the point and that you are indicating to something that’s in a distance. So I would like a robot that you could talk to on a natural language and that understands these concepts that humans right now take for granted.
SS:So at that point robots stop being just screws and bolts like you’ve put it, right? Because with the AI what else can the robot learn? Can there be a slight chance that robot can actually learn something about emotional intelligence, not only artificial intelligence? Feel something?
PR: You know, I think that even for humans that’s a very difficult concept to understand. So I am sure, long time before robots have actual feelings, if that’s even possible, we could certainly have robots that simulate feelings in a way that we can’t tell the difference. If you can’t tell a difference then what is the difference?
SS: So in that case could humans be falling in love with the robots?
PR: I am pretty sure that they can easily do that, I am pretty sure humans could fall in love with robots, but whether robots can actually fall in love with humans – that’s an entirely different matter.
SS:Does that concept scare you, that humans could fall in love with robots?
PR: Do you know, there are so many strange humans in the world; anyway humans that fall in love with strange things as it is. You know, I’m not going to be able to change humanity or mankind.
SS:If I were you I would say that my dream robot is someone who I can teach how to love. Have you ever thought of that?
PR: No, I don’t think that would be my concept of my dream robot at all. I think it would be something that just makes my life easier rather than something that I could love.
SS:So you have artificial intelligence vs human intelligence and these are two very different things. But at this point already you can walk in a bar and instead of human trio band you can have three robots play some music band. But who would want to hear that? I mean, humans don’t go to see robots play music if they don’t, you know, waste their self and emotions and sorrow in the music. What I am asking is that will the human still be in demand after robots will be able to do everything like humans or maybe even better?
PR: That’s a very interesting question. Right now you can go into, as you said, into a bar and then somebody who has programmed all his music into media interface and he is actually not putting any extra into that, he or she is just playing this music that’s pre-recorded, perhaps his or her voice is adding the emotion, I am sure it is. And can we simulate that with robots? And then if it is not just this emotional thing what else do humans have to offer that machines can’t? I think that’s the question. And I think, over the years technology has shown that jobs are taken away, jobs that I’ve started working at, for example, I was designing circuit boards that are now done by robots. I was building circuit boards and that’s all been taken over by machines, but then new jobs are created from that. And I think, there will always be a place for mankind; we will find new things for us to do. In science fiction it’s often that way – we go off to just explore and enjoy life while the machines carry on doing things.
SS:Let’s talk a bit more about will we have a place next to robots if they start to do our jobs better than we do. If we outsource everything to robots won’t we lose enormous amount of skills? I mean, for instance, we don’t even realise how automated our lives are, our grandparents, per say, they knew how to make bread and how to make cheese – we don’t anymore. If everyone is used to driving an automated cars and it means that no one will even know how to drive anymore, right?
PR: And will they need to?
SS:Well, isn’t it a bit scary when you know that you are losing a certain skill? Is it about need or maybe preserving heritage? Preserving skills is also a part of heritage, of human heritage.
PR: Absolutely! And those kind of skills, I mean people don’t have to know how to bind books anymore, it’s all done by print, and those skills aren’t quite lost but they are forgotten by most people, - and as you’ve said, baking bread. But there are new skills, there is computer generated art, there is computer generated music, there’s robotics - there are so many new skills like space exploration that come from that. Years ago when they were baking bread they didn’t know how to do a space walk.
SS: So you think that we shouldn’t be worried about losing certain amount of skills because they are inevitably replaced by the new ones - it is not like we are going to run out of skills because we are so lazy that we are going to let machines do everything for us?
PR: Yeah, I think we’ll always lack for something, that’s the nature of humans that we are interested in solving problems and finding solutions to difficulties. And the machines will continue helping us, but even when the machines surpass our intelligence we will still be able to work with them and beside them.
SS:So do you think that people will even go to school or colleges or you will have robots teaching you pretty much everything you need to know?
PR: Haven’t really thought about that question.
SS:But you have a chance to think about it right now.
PR: Yeah, I think, right now Wikipedia and Google are used so much in our daily lives that my daughter when she was 9 years old she said: “I don’t need to go to school anymore. If I need to know something I just google it”.
SS:What was your reaction to that?
PR: In some way she is right, if she does need to know something she can google it, but there is more than that in school. There is also human interaction, the skills that you need to learn apart from knowledge. So you have skills and knowledge and both are very important.
SS: So how do you make sure you preserve that? Because you’ve said just two seconds ago that even when the robots surpass us in their skills and intelligence we will still be able to work with them. Where is the guarantee? What if they take over us one day?
PR: Yeah, this concept is off the scale. I don’t believe that they will, I can’t believe that they will find in their programming a requirement or a need to create or to power hunger or anything that, I believe, is inherited human flaw rather than a machine flaw.
SS:Ok, but machine also has flaws, it can just go out of order, go berserk. What are you going to do with that?
PR: Yeah, if this happens some tools, the theoretical Isaac Asimov’s “Three Laws of Robotics”, are going to preserve, which, I don’t believe, can actually be integrated into robots. But as long as we be have an off-button, none of that, I think, should be too bad.
SS: As long as we have access to that button.
SS: Peter, there is this other thing that fascinates me and that’s augmented reality, when you put like Google glasses on and you see everything in 3D or 4D. Will there ever be time when you just laying around your couch and put on your glasses and all of the sudden you are in your office and that’s how you can work all day?
PR: Absolutely, I mean, part of what I do in vision and robotic vision is the augmented reality that you talk about, with two sides of the one coin. And it will be very easy to put on your googles or just stand in your virtual world and you can be somewhere else. It doesn’t even have to be in your office, you can be working in any environment in a different part of the world, even different part of the universe.
SS:How long till that happens? I have a problem getting up in the mornings. How long will that take?
PR: You know, I think you could probably do it in the next few years, you could do it very effectively. This is right on a cutting edge of technologies right now, and these developments are happening daily, there are new changes, and new algorithms are coming out to improve that. And the speed of processes, which is really where the bottleneck is for these processes.
SS:What do you think about how well we humans cope with the progress? Because progress over the last centuries has been consistent, but somewhat slow, and what we see for the past like 15 to 20 years are major breakthroughs. Like 25 years ago not everyone had a computer, now you can’t even think about your life without the computer or iPhone or your iPad, and, like you’ve said, everything is so automated, you can put on glasses and not go to work and work from your bed. Can we emotionally handle this amount of innovation in such a rapid pace always drastically changing so fast, can we cope with it and catch up with it?
PR: I think that emotional aspect to that is not my area of expertise, but in the technology part I don’t really see any reason why we shouldn’t be able to do that when as humans we are very malleable and we can change and diversify very easily and I think that generations move on. Two-year-olds can now type before they can write.
SS:What is your concept of progress? I’m sure it is something positive, right? Progress is something that brings good things to humanity, but when I look at it, it doesn’t necessarily change the societies for the best. I mean, the fact that a little baby can type at the age of two doesn’t mean that some crazy dude from ISIS won’t put an execution video on internet - do you know what I mean? So do you feel like progress is somehow changing humans as humans, or changing a setting for the better, or these are two parallel things – progress is progress and humans will just never change, society will stay the same?
PR: Yeah, I have to say that progress is progress and humans are humans, and we have invented TNT for making holes in mountains for trains to go through, and, of course, they use it for destructive purposes and bombs – that’s human nature, it’s not the fault of the technology or the progress. It’s the misuse of the technology, that’s the issue.
SS:So it’s a misconception to think that progress breeds better society?
PR: Well, I think that it would be really wrong to hold progress back because humans are bad. And I think that would be a counter argument to what you are suggesting.
SS:We’ve already witnessed robotic and drone warfare becoming a reality, and in Korea, for example, they already have robots that kill. What do you think about robots that can kill and then how can a robot decide to kill?
PR: Yeah, I am totally against machines that can kill and machines making decisions. Even when there is a human in the system which is the excuse that a lot of this use, it’s still very disjointed. And with the PlayStation video games, with generation that is growing up, playing Battlefield and games like this where it is so easy to press the button and a little shadow figure at the distance drops; and to do that in real life, it is so destroying, the emotion as you have mentioned several times before - there is nothing, there is no feeling, there is just few pixels on the screen that you are interacting with now. I think, creating machines like that, that are automatic or even semi-automatic with human interloper is a misuse of technology.
SS: Do you think this whole drone and robotics warfare, which makes war certainly safer, at least with the armies, will breed more wars? Will we see more wars, because they are just much more comfortable to fight at this point?
PR: Well, I am not sure whether it will make more wars, but I would argue that it doesn’t make them safer, it means that you just need more of these technologies to attack the other side, and probably we will even have the highest technology race, which has been the way wars has been fought for centuries: the invention of swords, bow and arrows guns, tanks and so on. And now we have drones.
SS:As a man who actually sees this whole artificial intelligence thing as a thing that is inevitable, but, you know, better be safe and ready for it, what security measures would you come up with, so that, you know, someone who is responsible should take blame – like in normal societies if you do something you go to court and you go to jail, right? You can’t be just like: “Oh, machine is wrong and let’s blow up half of the world”. But it is nobody’s fault because something went wrong with the machine. So what do you do with that, because that’s scary too? It must be a scare factor, you know, you can’t just be like: “Oops, not my fault, sorry, machine did it”.
PR: Yeah, well, you can’t play with the machine. These are issues that even if we’d look back, before robots or anything like that, when we, for example, had a simple device like a hairdryer, and you are drying your hair and it is exploding in your hands. This could happen; who is to blame here?
SS:Well, you usually sue the company and they pay a lot of money. What happens when a robot blows up half of the world? Who is responsible for that?
PR: What outcome would you like if you found a person who is responsible?
SS:I would like to hear from you who draws and writes my future, which is at least not my choice anymore, because it’s an irreversible process, I’d like to hear that you know that along with inventions we’ll make sure that it is so safe that if something goes wrong, there is certainly going to be someone who is going to take the blame.
PR: I think that technology shouldn’t be used in any way that is going to be unsafe. And that would include a machine that can blow up half of the world. And anyone that puts the machine that can possibly blow up half of the world, or even a small piece of it, or even a few people, then those people who’ve put that machine in that position are absolutely wrong to do that. I also believe that there should be so many safety functions on anything, I’ve worked on autopilot systems for aircraft. And on those aircraft autopilot systems there is so many fail safes, redundant systems, so if something does fail, there is a backup of the backup of the backup to make sure that it doesn’t fail. But it still fails and who’s false is it? Is it the manufacturer, like you say, we sue the company who made the plane or who put the people on the plane, or we sue the person who made that part that broke down in the chain?
SS:So we are just going to go with the flow and whatever happens – happens…
PR: Oh, no, I’m just saying that we need to make sure we have safety systems in there, but we don’t purposely make machines that cause harm.
SS: I know that you want to fly to space. Do you feel like robot could fly to another galaxy?
PR: Absolutely. A robot could certainly fly to another galaxy. We can program a robot, right now we can send a robot to another galaxy.
SS: And you also think that he could be, or she could be, I don’t know why I am saying he or she – the robots could be the first ones to interact with other forms of life, which I call aliens. Tell me more about it!
PR: Sure, I certainly believe that in this huge universe that we live in that it is more out there. And I would be hesitant to say that they have already visited us. I just think that somewhere out there may not be anything or maybe bacterial life, may not be anything intelligent, but probably there is some intelligent life out there. And, certainly, from now we could send off a robot to interact with some other species out there. Of course, it takes such a long time with current technologies to get anywhere, that by the time it got there we would probably be half way in the middle somewhere because life would probably have developed at similar pace to us, maybe not.
SS:Do you think robots could be colonizing Mars instead of humans?
PR: It is already - robots on Mars, before humans robots are living on Mars if you want to call it living.
PR: To have a colony, I guess, you have to have some sort of reproduction system. And that would mean that robots would have to go there and to be able to create new robots for it to build the colony, rather then us sending up teams of robots.
SS:Just tell me something, that would never be possible, right?
PR: I think it would be possible, it certainly would.
SS: Ok, on this worrisome note for me…
PR: I am sorry, I didn’t ease your mind in any way!
SS:I thank you for this interview, Peter Redmond, thank you for what you are doing and good luck with everything.
PR: Thanks for talking to me.