WEBVTT 1 00:00:00.000 --> 00:00:01.340 Greetings, ladies and gentlemen. 2 00:00:01.460 --> 00:00:06.480 Welcome to this session of the Adelaide Festival of Ideas. 3 00:00:07.300 --> 00:00:10.900 This session, as you know, is the robots are coming. 4 00:00:11.640 --> 00:00:12.960 Will machines take over? 5 00:00:13.200 --> 00:00:15.060 My name's Derek Abbott. 6 00:00:15.340 --> 00:00:18.540 I'm the director of the Biomedical Engineering Center 7 00:00:18.540 --> 00:00:19.900 here at Adelaide Uni. 8 00:00:20.080 --> 00:00:24.380 And it's my great pleasure to introduce our two speakers, 9 00:00:25.160 --> 00:00:27.000 Rodney Brooks and Margaret Wertheim. 10 00:00:28.120 --> 00:00:33.680 Rodney is the guy who possibly might be responsible 11 00:00:33.680 --> 00:00:36.180 for robots taking the world over one day 12 00:00:36.180 --> 00:00:39.320 because he's the head of the AI laboratory 13 00:00:39.320 --> 00:00:43.840 at the Massachusetts Institute of Technology in the USA, MIT. 14 00:00:44.980 --> 00:00:47.300 He's also the iRobot Corporation 15 00:00:47.300 --> 00:00:50.320 and a few other companies making robots in the USA. 16 00:00:52.760 --> 00:00:56.980 His group is so big that if you look at it, 17 00:00:56.980 --> 00:00:59.340 if you look at his website at MIT, 18 00:00:59.680 --> 00:01:04.160 you need a search engine just to see all the names of the people in his lab. 19 00:01:04.300 --> 00:01:04.760 It's incredible. 20 00:01:05.680 --> 00:01:07.920 It just gives you some idea of the magnitude. 21 00:01:09.180 --> 00:01:14.680 He's arguably the leading expert in robots in the world. 22 00:01:16.960 --> 00:01:20.340 He's originally from Adelaide and now lives in Boston. 23 00:01:21.080 --> 00:01:25.560 Now, Margaret Wertheim is a highly acclaimed science writer. 24 00:01:26.980 --> 00:01:28.060 She's written a couple of books, 25 00:01:28.180 --> 00:01:32.500 The Pearly Gates of Cyberspace and Pythagoras's Trousers. 26 00:01:34.040 --> 00:01:35.340 Or was it underwear? 27 00:01:35.940 --> 00:01:37.000 Trousers, that's right. 28 00:01:38.840 --> 00:01:42.100 She's actually originally from Melbourne 29 00:01:42.100 --> 00:01:44.160 but was brought up in Brizzy 30 00:01:44.160 --> 00:01:46.560 and also spent some time in Adelaide 31 00:01:46.560 --> 00:01:48.400 but now is actually living in Los Angeles. 32 00:01:50.600 --> 00:01:55.840 I did see a copy of Pythagoras's Trousers in the tent behind there. 33 00:01:55.900 --> 00:01:56.860 In fact, a few copies. 34 00:01:56.980 --> 00:02:01.140 I think Margaret said she'll be there at the end to sign any 35 00:02:01.140 --> 00:02:03.720 if you'd like to get some at the end. 36 00:02:04.180 --> 00:02:05.300 Without further ado, 37 00:02:06.480 --> 00:02:09.580 I'd like to hand it over to the speakers. 38 00:02:09.920 --> 00:02:12.480 What's going to happen is Rodney's going to come up first 39 00:02:12.480 --> 00:02:15.900 and make a little bit of a slide presentation about robots 40 00:02:15.900 --> 00:02:18.580 and then we'll just enter into a conversation 41 00:02:19.380 --> 00:02:22.320 and then at the end there'll be some time for some questions. 42 00:02:22.740 --> 00:02:25.140 Can we greet the speakers in the usual way? 43 00:02:25.280 --> 00:02:25.760 Thank you. 44 00:02:36.840 --> 00:02:37.980 Well, thank you, Derek. 45 00:02:39.180 --> 00:02:40.480 It's a great pleasure to be here. 46 00:02:40.960 --> 00:02:45.560 Yesterday morning at the same time I gave a longer talk by myself in this auditorium 47 00:02:45.560 --> 00:02:50.700 and I want to recap just a little tiny bit of that for about a minute and a half 48 00:02:50.700 --> 00:02:53.780 and then go into context and set up some context 49 00:02:53.780 --> 00:02:56.960 for the discussion today about will machines take over. 50 00:02:56.980 --> 00:03:01.780 And so the first two or three slides are really from yesterday's talk 51 00:03:02.320 --> 00:03:06.160 where I showed that there'd been a big resurgence in the last 10 years 52 00:03:06.160 --> 00:03:10.680 of building robots with human form and here are some examples of robots being built 53 00:03:10.680 --> 00:03:11.920 around the world in laboratories. 54 00:03:12.360 --> 00:03:16.120 The robots with human forms are not out in the world in any way. 55 00:03:16.200 --> 00:03:18.700 There are lots of other robots that are getting out into the world, 56 00:03:18.820 --> 00:03:21.560 home domestic robots, toys, et cetera. 57 00:03:21.760 --> 00:03:26.820 The robots with human form, which is what most people think of as the perhaps, 58 00:03:26.820 --> 00:03:30.060 threatening robots of the future, are in the research labs. 59 00:03:30.280 --> 00:03:33.540 And I wanted to show you one of those robots from my lab 60 00:03:33.540 --> 00:03:35.240 to show you the current state of the art. 61 00:03:35.360 --> 00:03:40.200 This is a robot called Kismet, which is a neck and head robot 62 00:03:40.800 --> 00:03:46.820 which understands people's emotional state and their faces, et cetera, 63 00:03:46.840 --> 00:03:48.860 and their movements, doesn't understand their language 64 00:03:49.460 --> 00:03:55.400 and it generates pre-linguistic babbling with prosody in its voice. 65 00:03:55.540 --> 00:03:56.700 And here's just an example. 66 00:03:56.700 --> 00:03:59.300 This is an example of a naive subject taken off the street, 67 00:03:59.440 --> 00:04:04.540 not exactly taken off the street, volunteer, who we put in front of the robot, 68 00:04:04.740 --> 00:04:08.120 we've done this with hundreds of people, and asked him to talk to the robot 69 00:04:08.120 --> 00:04:09.100 with no further instruction. 70 00:04:09.480 --> 00:04:12.360 This just gives you an idea of the state of the level of interaction 71 00:04:12.360 --> 00:04:13.360 as of today. 72 00:04:13.700 --> 00:04:14.280 There's something. 73 00:04:15.600 --> 00:04:18.400 I want to show you something. 74 00:04:18.660 --> 00:04:24.180 This is a watch that my girlfriend gave me. 75 00:04:24.480 --> 00:04:25.020 How the hell? 76 00:04:25.300 --> 00:04:26.680 Yeah, look, it's got a little blue thing. 77 00:04:26.680 --> 00:04:30.860 And the note for the note for the note for the note for the note for the note for the 78 00:04:30.860 --> 00:04:42.200 note for the note for the note for the note for the note for the note for the note for the 79 00:04:42.200 --> 00:04:42.780 note for the note for the note for the note for the note for the note for the note for the 80 00:04:42.780 --> 00:04:56.660 note for the note for the note for the note for the note for the note for the note for the 81 00:05:09.200 --> 00:05:14.280 So that's the sort of state of the art of social interaction between people and robots. 82 00:05:14.720 --> 00:05:19.780 And when people see these sorts of robots, there are two big questions that come up. 83 00:05:19.900 --> 00:05:23.240 The first question is, will we accept them in the future? 84 00:05:23.240 --> 00:05:28.780 And the second question, which is really what this session is about, is will they want to take over? 85 00:05:29.900 --> 00:05:33.120 Let me just say a couple of words about will we accept them. 86 00:05:33.380 --> 00:05:40.700 And both of these questions have been addressed many times in books, science fiction books and movies. 87 00:05:40.960 --> 00:05:45.500 And I've got a few images here from various movies about will we accept them. 88 00:05:46.360 --> 00:05:51.020 And my position on that, which I talked at a little greater length yesterday, 89 00:05:51.020 --> 00:05:55.780 is that whether we accept them and grant them personhood or not, 90 00:05:58.000 --> 00:06:04.680 in order for us to grant them personhood, we will have to get over our want for specialness. 91 00:06:05.200 --> 00:06:08.340 Historically, we have liked to think of ourselves as being special, 92 00:06:08.480 --> 00:06:13.660 and science and technology has assaulted us on that issue over time. 93 00:06:14.480 --> 00:06:17.940 500 years ago, the Earth was no longer the center of the universe. 94 00:06:18.000 --> 00:06:19.700 There was a clash with the Catholic Church. 95 00:06:19.780 --> 00:06:21.000 That made us less. 96 00:06:21.020 --> 00:06:21.660 Less special. 97 00:06:21.920 --> 00:06:25.520 And we've seen that with Darwin, that we're related to animals, et cetera. 98 00:06:25.700 --> 00:06:27.080 And we keep getting assaulted. 99 00:06:29.600 --> 00:06:34.000 Recently, people have been saying maybe we didn't randomly arise here on Earth, 100 00:06:34.100 --> 00:06:38.540 but life came from another planet, so we don't even have our own creation event. 101 00:06:38.780 --> 00:06:41.880 More recently, we find we only have 35,000 genes. 102 00:06:42.040 --> 00:06:46.520 This was just three or four months ago, and everyone got really upset that we had so few genes. 103 00:06:47.620 --> 00:06:50.940 And whether we ultimately will get over ourselves. 104 00:06:51.020 --> 00:06:52.220 And we all have to understand for ourselves that we're special, 105 00:06:52.320 --> 00:06:56.520 because we have emotions, and those cold, hard machines can't have real emotions. 106 00:06:57.100 --> 00:07:00.980 My position is we're going to have to retreat from our specialness again 107 00:07:01.540 --> 00:07:05.720 before we grant that some sort of robots in the future can be living creatures. 108 00:07:06.060 --> 00:07:10.680 So that's the will we accept them, but the next question is should we fear them? 109 00:07:11.220 --> 00:07:15.000 And this has been a theme in many, many movies, of course, 110 00:07:15.120 --> 00:07:16.980 and we'll probably mention a few others in the discussion. 111 00:07:16.980 --> 00:07:20.940 and the issue here is 112 00:07:20.940 --> 00:07:24.320 you know, we create these entities 113 00:07:24.320 --> 00:07:28.020 we nurture them, we teach them everything we know 114 00:07:28.020 --> 00:07:30.360 they start getting smarter and smarter 115 00:07:30.360 --> 00:07:32.820 eventually they start thinking they're smarter than us 116 00:07:32.820 --> 00:07:34.200 they get impatient with us 117 00:07:34.200 --> 00:07:36.520 and they want us out of their lives 118 00:07:36.520 --> 00:07:38.580 and this is very common with our teenagers 119 00:07:38.580 --> 00:07:41.000 as I'm sure you know 120 00:07:45.560 --> 00:07:48.700 I don't think this is necessarily going to happen 121 00:07:48.700 --> 00:07:52.740 because we could choose to build dangerous robots 122 00:07:52.740 --> 00:07:55.740 just like we could choose to build dangerous trains 123 00:07:55.740 --> 00:07:59.220 but we choose not to build dangerous trains 124 00:07:59.220 --> 00:08:01.800 we're sort of in control of what we build 125 00:08:01.800 --> 00:08:03.800 so I think there's some argument there 126 00:08:03.800 --> 00:08:05.900 that things aren't necessarily going to get away from us 127 00:08:05.900 --> 00:08:09.440 but one of the scenarios that's popular in fiction 128 00:08:10.120 --> 00:08:10.980 is the lack of control 129 00:08:11.000 --> 00:08:11.780 there's a lone scientist 130 00:08:11.780 --> 00:08:14.160 who's tinkering away 131 00:08:14.160 --> 00:08:17.140 and accidentally builds 132 00:08:17.140 --> 00:08:19.760 I built a 747 in the backyard 133 00:08:19.760 --> 00:08:22.480 I just put these pieces together 134 00:08:22.480 --> 00:08:24.860 and it flew like a jet airplane 135 00:08:24.860 --> 00:08:26.480 that doesn't happen 136 00:08:26.480 --> 00:08:28.600 so I don't think that scenario is going to happen 137 00:08:28.600 --> 00:08:32.000 building these things is a complex undertaking 138 00:08:32.000 --> 00:08:34.040 with lots of people working together 139 00:08:34.040 --> 00:08:37.160 across many different organizations for many years 140 00:08:37.160 --> 00:08:39.160 and we'll build them incrementally 141 00:08:39.160 --> 00:08:40.980 we're not going to just suddenly build them 142 00:08:41.000 --> 00:08:44.240 we're not going to build the robot with super intelligence 143 00:08:44.240 --> 00:08:45.740 or the robot with emotions 144 00:08:48.960 --> 00:08:50.880 we'll be building simpler robots 145 00:08:50.880 --> 00:08:52.140 and then more complex robots 146 00:08:52.140 --> 00:08:53.340 and more complex robots 147 00:08:53.340 --> 00:08:54.680 so we'll gain experience with them 148 00:08:54.680 --> 00:08:58.880 we will be the Darwinian influence on the robots 149 00:08:58.880 --> 00:09:00.940 in deciding what we want to build 150 00:09:00.940 --> 00:09:01.800 and how we build them 151 00:09:01.800 --> 00:09:03.460 and we'll in a sense breed them 152 00:09:03.460 --> 00:09:07.220 and modify them to be the right sort of robots that we want 153 00:09:07.220 --> 00:09:08.640 if we start building robots 154 00:09:09.860 --> 00:09:10.980 that are not Senate 155 00:09:10.980 --> 00:09:15.180 are troublesome in one way, we'll modify them and build different sorts of robots. 156 00:09:15.320 --> 00:09:18.880 So we'll have control over them while they're less than us. And that while 157 00:09:18.880 --> 00:09:22.560 they're less than us may be the critical issue that this 158 00:09:23.140 --> 00:09:24.100 whole session is about. 159 00:09:26.460 --> 00:09:30.220 At the same time, we can't foresee all the uses and developments of them. 160 00:09:32.960 --> 00:09:35.080 Now, I think 161 00:09:35.080 --> 00:09:38.800 there's another issue that's going to happen fairly quickly, and that's us 162 00:09:38.800 --> 00:09:42.820 as them. And this might make the question of whether the 163 00:09:42.820 --> 00:09:46.780 robot's going to take over from us a little different. And maybe 164 00:09:46.780 --> 00:09:50.780 it's the wrong perspective. Maybe we don't quite understand things and are asking the wrong 165 00:09:50.780 --> 00:09:54.160 question. You know, in the last millennium, we came to rely on our machines. 166 00:09:54.960 --> 00:09:58.160 In this millennium, I think we will become our machines. 167 00:09:58.800 --> 00:10:02.840 And in the first instance, we're becoming our machines by putting 168 00:10:02.840 --> 00:10:06.700 silicon and steel inside our bodies, often for good clinical reasons, 169 00:10:06.880 --> 00:10:07.920 but that's going to 170 00:10:07.920 --> 00:10:08.780 become 171 00:10:08.800 --> 00:10:13.000 more cosmetic. The answer, I think, about whether 172 00:10:13.000 --> 00:10:17.100 they'll take over from us under this is we don't need to fear the machines 173 00:10:17.100 --> 00:10:20.880 because we, the man machines, will always be a step ahead of the machine 174 00:10:20.880 --> 00:10:24.980 machines because we'll be putting the technologies in our own bodies. There won't be any us for them 175 00:10:24.980 --> 00:10:27.180 to take over from because them will be us. 176 00:10:29.880 --> 00:10:32.320 So this uniting of flesh and machines, 177 00:10:32.700 --> 00:10:36.640 you know, we see that already, leg and arm prostheses, cochlear implants, 178 00:10:36.840 --> 00:10:38.780 there's tens of thousands of people walking around. 179 00:10:38.800 --> 00:10:43.080 On the planet today with direct neural connections into their cochlears, 180 00:10:43.100 --> 00:10:47.400 into electronics interfacing to neurons so they can hear again. 181 00:10:47.660 --> 00:10:51.540 There's experiments in retinal implants, people with macular degeneration. 182 00:10:52.820 --> 00:10:58.240 There have been implants of artificial retinas, temporary implants into people's eyes. 183 00:10:59.600 --> 00:11:03.020 Can't really, they don't really work yet, but there's a lot of clinical push. 184 00:11:03.040 --> 00:11:08.100 If you have macular degeneration and there's a chance that a silicon retina will let you see something again, 185 00:11:08.100 --> 00:11:10.380 that's something a lot of people are going to jump on. 186 00:11:10.800 --> 00:11:21.400 And electrodes and the thalamus for Parkinson's disease, other sorts of interfaces for other diseases of the cerebellum 187 00:11:21.400 --> 00:11:25.120 are really being pushed on right now. So direct connections into the brain. 188 00:11:25.260 --> 00:11:30.080 And then there are people putting extra neural circuits into rat and mouse brains, 189 00:11:30.740 --> 00:11:37.300 only being done with animals now, but that will change over time too because it's aimed at therapy. 190 00:11:38.100 --> 00:11:39.200 For good clinical reasons. 191 00:11:39.480 --> 00:11:43.760 Once you have these clinical sorts of capabilities though, 192 00:11:43.860 --> 00:11:48.440 there's going to be some real push for people to use these technologies, 193 00:11:48.640 --> 00:11:51.900 not just for repairing damage, but for improving themselves. 194 00:11:52.580 --> 00:11:57.060 So one can imagine someone saying, gee, you know, 195 00:11:59.540 --> 00:12:04.360 first of all someone who's got macular degeneration in both eyes, they get one eye fixed the normal way. 196 00:12:04.560 --> 00:12:08.080 And then they say, well, when my second eye gets fixed, it'd be great. 197 00:12:08.080 --> 00:12:10.900 It'd be great to have an infrared camera in my left eye. 198 00:12:12.340 --> 00:12:23.700 And maybe, you know, drug runners might choose to do that so they can evade the police or military or whoever better. 199 00:12:23.900 --> 00:12:32.040 And one can imagine at some point some military organizations deciding they're going to augment their soldiers 200 00:12:32.040 --> 00:12:34.820 with implants of some sort. 201 00:12:38.180 --> 00:12:46.340 I think ultimately, and already we have a few quadriplegics in the world who can control a mouse just by thinking with direct neural implants. 202 00:12:46.520 --> 00:12:56.040 I don't think it's going to be too long before we can have and expect some sort of wireless internet connection inside our heads. 203 00:12:56.880 --> 00:13:02.080 And when we get that, that's really going to change who we are and how we interact with the world. 204 00:13:02.240 --> 00:13:06.100 But more generally, we're putting more and more technology into our bodies. 205 00:13:08.080 --> 00:13:18.560 But then as we're becoming silicon and steel in our bodies, which is already starting and will accelerate over the next 10, 20 years, 206 00:13:19.300 --> 00:13:23.360 engineering will gain exquisite control over biological processes. 207 00:13:23.540 --> 00:13:27.940 Right now, you know, all this genetic engineering is pretty primitive stuff. 208 00:13:28.200 --> 00:13:36.720 The way cloning happens is you take an egg, you cut out a piece, you put in another nucleus, and then you zap it with electricity. 209 00:13:37.060 --> 00:13:38.060 It's like an old Frankenstein. 210 00:13:38.080 --> 00:13:42.060 That's the level of technological sophistication right now. 211 00:13:42.140 --> 00:13:44.020 That is changing over time. 212 00:13:44.220 --> 00:13:50.240 And Derek mentioned this, a lot of people in my lab, there's about 20 faculty in my lab, 213 00:13:50.320 --> 00:14:01.480 and one of them is working on and has successfully built digitally controlled E. coli by having a compiler 214 00:14:02.100 --> 00:14:07.040 which compiles programs into DNA strands, which he then inserts into the genome 215 00:14:07.040 --> 00:14:08.060 of the E. coli. 216 00:14:08.060 --> 00:14:18.900 And when the RNA transcription mechanism runs over that, it executes a digital program where different enzyme concentrations 217 00:14:18.900 --> 00:14:23.960 represent the different numbers in the program, and he gets digital control of these E. coli. 218 00:14:24.040 --> 00:14:26.080 So essentially they're living little tiny robots. 219 00:14:26.460 --> 00:14:29.220 So this is just starting now. 220 00:14:29.300 --> 00:14:38.040 And by the way, he's doing that in a lab that we built which cost about $100,000 to set up and didn't use any technology. 221 00:14:38.060 --> 00:14:41.200 technology that's not available in some American high schools. 222 00:14:41.480 --> 00:14:45.460 So this is not really difficult stuff to do anymore. 223 00:14:45.860 --> 00:14:49.140 Engineering will gain exquisite control over biological processes, 224 00:14:49.840 --> 00:14:53.000 and our whole manufacturing technologies are going to change as a result. 225 00:14:53.520 --> 00:14:59.340 In 40 to 50 years, we won't grow a tree, cut it down, and build a table. 226 00:14:59.420 --> 00:15:00.520 We'll grow a table. 227 00:15:01.520 --> 00:15:02.080 Okay? 228 00:15:02.080 --> 00:15:08.440 So our biological, our industrial infrastructure just has changed completely in the last 50 years. 229 00:15:08.500 --> 00:15:10.560 Look at the industrial infrastructure from 50 years ago. 230 00:15:10.820 --> 00:15:14.960 Totally different materials to what we use today is going to change again in the next 50 years, 231 00:15:15.060 --> 00:15:17.020 and it's going to change to biological materials. 232 00:15:18.440 --> 00:15:22.180 So the next generation of robots are going to be made out of that stuff. 233 00:15:22.500 --> 00:15:25.420 The next generation of our industry is going to be made out of that stuff, 234 00:15:25.460 --> 00:15:27.340 and it's the same stuff as us. 235 00:15:27.900 --> 00:15:31.360 So I think that, in a sense, the robots will take over 236 00:15:31.360 --> 00:15:31.960 because they will. 237 00:15:32.080 --> 00:15:36.900 They will be us because we're going to ultimately merge in the 50 to 100-year time frame. 238 00:15:37.080 --> 00:15:39.240 So that's my position on this question, 239 00:15:41.000 --> 00:15:46.180 and maybe now we should get rid of that and have our discussion. 240 00:15:46.660 --> 00:15:51.180 Well, to kick off, I'd like to pick up on a point you made, Rodney, 241 00:15:51.240 --> 00:15:55.500 about we have control over the safety of trains. 242 00:15:55.780 --> 00:15:57.340 So you say trains are totally safe, 243 00:15:57.440 --> 00:16:01.100 but the analogy there, I think, is a little weak, perhaps, 244 00:16:01.160 --> 00:16:01.640 because... 245 00:16:02.080 --> 00:16:06.440 Trains don't have autonomy, whereas robots do, 246 00:16:06.560 --> 00:16:08.600 and perhaps the danger is their autonomy. 247 00:16:08.720 --> 00:16:10.840 How can we guarantee safety and autonomy? 248 00:16:11.100 --> 00:16:12.800 I would say not so much autonomy, 249 00:16:12.880 --> 00:16:16.660 because actually many trains do have autonomy in lots of train systems. 250 00:16:16.780 --> 00:16:17.400 It's willfulness. 251 00:16:17.780 --> 00:16:22.300 They don't have willfulness, and I think that's an issue. 252 00:16:23.380 --> 00:16:26.700 But I think, in the shorter term, 253 00:16:27.960 --> 00:16:31.140 we can choose not to build trains with willfulness, 254 00:16:32.080 --> 00:16:33.220 and robots with willfulness. 255 00:16:33.620 --> 00:16:36.880 And I think in the shorter term, there's going to be a lot of... 256 00:16:36.880 --> 00:16:40.200 Within the next 10 years, certainly, maybe within the next five years, 257 00:16:40.420 --> 00:16:44.220 there'll be lots of issues that come up about international treaties 258 00:16:44.220 --> 00:16:48.220 on the use of robots in warfare, extensions of the Geneva Convention. 259 00:16:48.380 --> 00:16:54.720 Should we say that robots are allowed to make autonomous targeting decisions? 260 00:16:55.260 --> 00:16:59.620 And we might decide, just as we've decided that biological warfare is a no-no, 261 00:16:59.700 --> 00:17:01.600 we might decide that we don't want that. 262 00:17:02.080 --> 00:17:03.360 And that might be a prudent decision. 263 00:17:05.640 --> 00:17:06.180 Margaret. 264 00:17:07.940 --> 00:17:09.080 That's a very interesting question. 265 00:17:09.160 --> 00:17:11.660 There was a science fiction story, I forget who wrote it, 266 00:17:11.720 --> 00:17:16.020 about the notion that Americans build a supercomputer to take over their... 267 00:17:16.020 --> 00:17:18.160 basically to control their military, 268 00:17:18.200 --> 00:17:21.220 because they decided that these decisions are so complex 269 00:17:21.220 --> 00:17:25.220 that only a supercomputer could factor in all the equations, 270 00:17:25.380 --> 00:17:26.860 and then the Russians build one too. 271 00:17:27.180 --> 00:17:31.160 And very quickly, these two super-controlling machines decide 272 00:17:31.160 --> 00:17:32.060 that the only thing... 273 00:17:32.080 --> 00:17:33.200 worth talking to is each other. 274 00:17:33.480 --> 00:17:37.540 And so they basically leave the humans out of the equation altogether. 275 00:17:38.020 --> 00:17:43.960 And I think that does raise the issue of how much control should we hand over to these systems? 276 00:17:44.220 --> 00:17:47.900 I mean, we're already facing this problem in things like air traffic control, 277 00:17:48.400 --> 00:17:54.020 as systems, particularly traffic-type situations, get very, very complicated. 278 00:17:54.280 --> 00:17:55.000 And the stock market. 279 00:17:55.220 --> 00:17:56.240 And the stock market. 280 00:17:56.260 --> 00:17:59.880 More and more of these kinds of very large-scale organizations, 281 00:18:00.260 --> 00:18:02.040 the decisions are, in fact, being made. 282 00:18:02.080 --> 00:18:02.680 They're being made by machines. 283 00:18:02.880 --> 00:18:06.080 And some people worry that, in fact, there are problems there. 284 00:18:06.200 --> 00:18:08.040 But, of course, then there is the other problem of, 285 00:18:08.060 --> 00:18:10.840 can a human being actually keep track of it all? 286 00:18:11.920 --> 00:18:15.080 So I think in that sense, as you said before, Rodney, 287 00:18:15.160 --> 00:18:19.540 that already we are handing over a lot of tasks to machines, 288 00:18:19.780 --> 00:18:23.020 not necessarily robots, but certainly computer systems. 289 00:18:27.440 --> 00:18:30.620 At the same time, as you say that, I think we do have to be careful 290 00:18:30.620 --> 00:18:36.300 that a lot of our examples in thinking about this do come from science fiction and from movies. 291 00:18:36.500 --> 00:18:41.380 And the motivation of the movie makers is to tell a good story that is going to rivet people. 292 00:18:41.700 --> 00:18:46.440 And that's not necessarily the same as being a prediction of what the future will be like. 293 00:18:46.880 --> 00:18:51.560 So I think these movies are good things to stimulate our thought, 294 00:18:51.700 --> 00:18:57.500 but we shouldn't necessarily think that that is the prediction of the future 295 00:18:57.500 --> 00:18:59.020 and that is how things are going to unfold. 296 00:18:59.020 --> 00:19:03.320 And a good way to look at that is look at any old science fiction movie 297 00:19:03.320 --> 00:19:10.360 or any old book written by a technology pundit predicting what the future is going to be like, 298 00:19:10.420 --> 00:19:11.660 and they're all pretty much wrong. 299 00:19:12.240 --> 00:19:16.040 But it does raise questions that we need to start thinking about. 300 00:19:16.220 --> 00:19:21.080 And I think the moral and ethical issues of all these technologies coming together 301 00:19:21.080 --> 00:19:23.860 are going to be pretty severe for us over the next few years. 302 00:19:24.060 --> 00:19:29.000 Yes. There was a lovely cartoon in the New Yorker last year that I thought it interesting. 303 00:19:29.020 --> 00:19:30.700 It in many ways summed up this whole issue very well. 304 00:19:30.760 --> 00:19:34.560 There's this guy standing in front of his microwave oven looking at it saying, 305 00:19:34.720 --> 00:19:38.140 no, I don't want to play chess, I just want you to reheat the lasagna. 306 00:19:40.800 --> 00:19:46.600 This is the kind of issue that is being raised in a lot of these films, 307 00:19:46.660 --> 00:19:50.320 like Blade Runner or the new Steven Spielberg film that's just come out in America 308 00:19:50.320 --> 00:19:56.340 called Artificial Intelligence about if we do have these increasingly intelligent machines, 309 00:19:57.520 --> 00:20:01.780 presumably that will come with some emerging level of self-awareness 310 00:20:01.780 --> 00:20:05.380 and what would our responsibility to such beings be. 311 00:20:05.780 --> 00:20:09.320 I suspect we're a fair way from them actually being conscious, 312 00:20:09.440 --> 00:20:12.640 although perhaps you'd like to correct me on that. 313 00:20:12.680 --> 00:20:17.880 No, actually in my lab we refer to that as the C word and we don't say it. 314 00:20:17.920 --> 00:20:18.140 C word. 315 00:20:18.800 --> 00:20:24.800 Because my opinion is that no one understands anything about it in any real sense. 316 00:20:25.060 --> 00:20:26.320 But even if you didn't understand it, 317 00:20:26.340 --> 00:20:27.840 it's possible that it could emerge. 318 00:20:29.080 --> 00:20:35.280 Yes, but I don't think that's a plausible possibility, but I don't think, yeah, who knows. 319 00:20:36.040 --> 00:20:37.960 Could you be cruel to a robot? 320 00:20:39.220 --> 00:20:46.600 Ah, there was a science fiction story called Beastie 3 and Beastie 4 from about 35 years ago. 321 00:20:47.040 --> 00:20:51.540 A small robot, I think this was inspired by the Johns Hopkins Beast, 322 00:20:51.780 --> 00:20:55.340 which was a robot built at Johns Hopkins University in the 60s. 323 00:20:55.340 --> 00:21:01.220 A robot which went around and plugged itself into power outlets and got its own energy, 324 00:21:01.340 --> 00:21:05.800 but it didn't have an off switch and it would run around 325 00:21:07.060 --> 00:21:15.320 and if you caught it, you know, it would squeal and run off if you put it down on the ground 326 00:21:15.320 --> 00:21:17.080 and you couldn't, because you couldn't switch it off, 327 00:21:17.120 --> 00:21:20.900 you had to lock it in a room and starve it where there were no power outlets. 328 00:21:22.340 --> 00:21:24.340 And the point of the story was, 329 00:21:24.400 --> 00:21:30.180 someone claiming you couldn't be cruel to a robot because they didn't have any emotions, 330 00:21:30.320 --> 00:21:32.760 because we're special and we have the emotions, not the robots. 331 00:21:33.000 --> 00:21:36.820 And so the protagonist in the story gives it to this woman who's saying this and says, 332 00:21:36.940 --> 00:21:40.920 well, you kill it then, and gives her a hammer and it's squealing and, 333 00:21:40.980 --> 00:21:41.140 ah! 334 00:21:41.940 --> 00:21:44.380 And she can't bring herself to do it. 335 00:21:45.600 --> 00:21:47.440 So can you be cruel to a robot? 336 00:21:47.560 --> 00:21:51.360 I think cruelty is in the eye of the observer and beholder. 337 00:21:51.500 --> 00:21:53.720 I mean, that's really, I think, 338 00:21:53.720 --> 00:21:58.480 the basis of what we think of cruelty is our empathy for the thing we're being cruel to. 339 00:21:58.760 --> 00:22:02.580 And, you know, there's studies that psychopathic killers, 340 00:22:02.800 --> 00:22:05.940 you know, when they were children, did horrible things to animals 341 00:22:05.940 --> 00:22:08.300 because they didn't feel empathy for the animals. 342 00:22:08.560 --> 00:22:10.580 And so I think it gets down to an empathy issue, 343 00:22:10.680 --> 00:22:13.300 whether we can identify with that creature. 344 00:22:13.400 --> 00:22:14.780 And I think it depends on its behaviour. 345 00:22:17.680 --> 00:22:22.040 There's also the philosopher in computer science, Brian Cantwell Smith, 346 00:22:22.160 --> 00:22:23.600 who raised an interesting point about that. 347 00:22:23.720 --> 00:22:28.360 And he says that if you did get some kind of artificial intelligence, 348 00:22:29.520 --> 00:22:32.720 you know, in some form of mechanistic form, 349 00:22:32.920 --> 00:22:35.720 that it would be a differently embodied being 350 00:22:35.720 --> 00:22:39.480 and that perhaps its whole emotional experience of the world would be different 351 00:22:39.480 --> 00:22:41.320 because it had a different embodiment to us. 352 00:22:41.420 --> 00:22:43.820 And I actually think that's a very important point, 353 00:22:43.920 --> 00:22:48.460 is that just because it had the same basic shape as us, 354 00:22:48.500 --> 00:22:50.400 just because it was a humanoid in form, 355 00:22:50.500 --> 00:22:53.120 would not necessarily mean that we should assume, 356 00:22:53.120 --> 00:22:56.760 that its emotional structure would in some sense mirror ours. 357 00:22:57.000 --> 00:22:59.300 Because that's always the sort of underlying premise 358 00:22:59.300 --> 00:23:01.880 of these science fiction books and films, 359 00:23:01.980 --> 00:23:07.360 that they would be like us, only they'd be mechanical rather than organic. 360 00:23:07.600 --> 00:23:08.920 But I think that he's right. 361 00:23:09.000 --> 00:23:12.660 It's not necessarily clear that they would mirror us psychologically. 362 00:23:13.100 --> 00:23:15.280 So maybe they would experience all of those things, 363 00:23:15.340 --> 00:23:19.460 including ethical concepts, very differently than we might imagine. 364 00:23:19.460 --> 00:23:21.660 Yeah, but I think we have to remember that at the moment, 365 00:23:21.700 --> 00:23:23.100 it certainly depends on how people view it. 366 00:23:23.120 --> 00:23:23.840 And I think that's the key to building the robots. 367 00:23:24.000 --> 00:23:27.840 So Brian Cantwell Smith's point about the form 368 00:23:27.840 --> 00:23:31.440 is also another one that's taken by, 369 00:23:31.640 --> 00:23:33.520 and now I have trouble with names, 370 00:23:34.720 --> 00:23:36.980 the Berkeley philosopher who wrote 371 00:23:36.980 --> 00:23:38.500 Women, Fire and Dangerous Things. 372 00:23:39.080 --> 00:23:41.620 Oh, George Lakoff? 373 00:23:41.700 --> 00:23:43.460 George Lakoff and Smith, 374 00:23:44.100 --> 00:23:48.260 who argue that most of our conceptions of the world 375 00:23:48.260 --> 00:23:50.160 are based on bodily metaphors. 376 00:23:50.180 --> 00:23:52.700 So that anything with a different bodily form, 377 00:23:52.700 --> 00:23:55.680 than us, will have a different basic understanding, 378 00:23:55.840 --> 00:23:58.040 a different set of metaphors for the world. 379 00:23:58.200 --> 00:24:00.540 Which is one of the reasons that so many people 380 00:24:00.540 --> 00:24:02.040 are building robots with human form, 381 00:24:02.220 --> 00:24:05.800 so that the robots will have the same sorts of experiences as humans have. 382 00:24:06.380 --> 00:24:10.460 Additionally, though, there is this thing that we've observed very strongly, 383 00:24:10.640 --> 00:24:15.620 that people seeing the human form react to it in a human-like way, 384 00:24:15.720 --> 00:24:17.360 and know how to react to it. 385 00:24:17.740 --> 00:24:19.700 Which then accelerates this process. 386 00:24:19.920 --> 00:24:22.460 But underneath it all, it's still a matter of what it is, 387 00:24:22.700 --> 00:24:24.340 what algorithms, what components, 388 00:24:25.080 --> 00:24:27.680 computational components we put inside. 389 00:24:28.460 --> 00:24:32.060 What do you think, Rodney, of this guy, Stuart Wilkinson, 390 00:24:32.060 --> 00:24:34.360 who's making robots that eat, 391 00:24:34.480 --> 00:24:38.520 rather than get their power from batteries or electricity, 392 00:24:38.740 --> 00:24:40.520 you know, just going down in Florida? 393 00:24:40.880 --> 00:24:43.360 Yeah, there's actually a couple of other projects, 394 00:24:43.460 --> 00:24:44.880 one at the University of West England, 395 00:24:45.000 --> 00:24:47.140 and another one at Helsinki Institute of Technology, 396 00:24:48.980 --> 00:24:52.680 where, you know, one of the aspects of what we're doing, 397 00:24:52.700 --> 00:24:53.220 is that we're building living systems, 398 00:24:53.340 --> 00:24:55.180 as they all supply their own energy, 399 00:24:56.860 --> 00:24:59.760 and they do it by taking organic material 400 00:24:59.760 --> 00:25:02.060 and decomposing it and getting energy out of that. 401 00:25:02.320 --> 00:25:03.760