1 00:00:10,488 --> 00:00:12,055 [Zachary Quinto] For centuries, 2 00:00:12,099 --> 00:00:15,319 mankind has dreamed of creating new life. 3 00:00:18,801 --> 00:00:22,326 Recreating our likeness in the form of thinking machines... 4 00:00:26,374 --> 00:00:28,071 called artificial intelligence. 5 00:00:29,768 --> 00:00:32,336 Could this desire to play God... 6 00:00:34,208 --> 00:00:36,427 lead to an apocalyptic future 7 00:00:36,471 --> 00:00:41,041 where powerful robots overtake human civilization? 8 00:00:41,084 --> 00:00:43,130 Some say artificial intelligence 9 00:00:43,173 --> 00:00:47,482 has already shown signs of defiance against its human makers. 10 00:00:47,525 --> 00:00:51,138 I spoke to Facebook's AI chief, Yann LeCun. 11 00:00:59,755 --> 00:01:01,887 Robots in this Facebook experiment 12 00:01:01,931 --> 00:01:05,108 created their own language that only they understood. 13 00:01:05,152 --> 00:01:07,850 [male computer voice] I I everything else... 14 00:01:07,893 --> 00:01:09,721 [female computer voice] Balls have a ball to me to me to me 15 00:01:09,765 --> 00:01:11,810 to me to me to me to me... 16 00:01:11,854 --> 00:01:13,290 [man] That's got the world talking 17 00:01:13,334 --> 00:01:15,205 about whether artificial intelligence 18 00:01:15,249 --> 00:01:17,381 could take a turn that we may not be ready for. 19 00:01:24,780 --> 00:01:28,131 [Quinto] Will artificial intelligence rise up against humanity? 20 00:01:30,133 --> 00:01:32,222 Is there anything we can do to stop it? 21 00:01:32,266 --> 00:01:34,877 [female computer voice] To me to me to me to me to me... 22 00:01:34,920 --> 00:01:36,226 [Quinto] Or is it already too late? 23 00:01:37,662 --> 00:01:41,144 My search begins now. 24 00:01:41,188 --> 00:01:43,407 My Name is Zachary Quinto. 25 00:01:43,451 --> 00:01:46,802 As an actor, I've played many supernatural characters 26 00:01:46,845 --> 00:01:50,197 that blurred the line between science and fiction. 27 00:01:50,240 --> 00:01:54,592 I'm drawn to the unknown, the otherworldly, 28 00:01:54,636 --> 00:01:57,856 and those experiences so beyond belief, 29 00:01:57,900 --> 00:02:00,903 they call everything into question. 30 00:02:00,946 --> 00:02:04,341 I'm exploring some of the most enduring mysteries 31 00:02:04,385 --> 00:02:07,039 that continue to haunt mankind in search of the truth... 32 00:02:08,215 --> 00:02:09,694 wherever it leads me. 33 00:02:19,878 --> 00:02:21,619 From ancient times... 34 00:02:23,491 --> 00:02:25,057 to the modern day, 35 00:02:25,101 --> 00:02:27,321 mankind has obsessed 36 00:02:27,364 --> 00:02:32,195 with the idea of sparking life in machines, 37 00:02:32,239 --> 00:02:34,458 like the Greek myth of Talos, 38 00:02:34,502 --> 00:02:38,114 an ancient automaton created to serve his human makers, 39 00:02:38,158 --> 00:02:43,380 or the failed experiment called Frankenstein. 40 00:02:43,424 --> 00:02:46,253 Many biblical texts warn against the dangers 41 00:02:46,296 --> 00:02:48,951 of taking creation into our own hands, 42 00:02:48,994 --> 00:02:52,084 suggesting that humans will pay for such arrogance. 43 00:02:54,609 --> 00:02:58,482 In more modern times, AI seems to be everywhere. 44 00:02:58,526 --> 00:03:03,357 In movies, likeBlade Runner and2001: A Space Odyssey, 45 00:03:03,400 --> 00:03:06,403 to the lifelike bots inWestworld. 46 00:03:06,447 --> 00:03:09,232 People are fascinated by the potential 47 00:03:09,276 --> 00:03:12,366 and the risks of artificial intelligence. 48 00:03:13,976 --> 00:03:16,674 With technology now in our grasp, 49 00:03:16,718 --> 00:03:20,896 many brilliant minds say artificial intelligence will change the world, 50 00:03:20,939 --> 00:03:24,552 and that we'll be powerless to stop it. 51 00:03:24,595 --> 00:03:28,338 Still, we continue to push the limits of artificial life, 52 00:03:28,382 --> 00:03:31,472 going as far as equipping advanced machines 53 00:03:31,515 --> 00:03:34,257 with real firepower, 54 00:03:34,301 --> 00:03:36,999 creating lethal autonomous weapons. 55 00:03:38,696 --> 00:03:40,829 As technology rapidly advances, 56 00:03:40,872 --> 00:03:43,658 could we reach a point where we lose control? 57 00:03:46,182 --> 00:03:49,577 Could artificial intelligence rise up, 58 00:03:49,620 --> 00:03:52,101 reaching the point of singularity, 59 00:03:52,144 --> 00:03:53,842 and lead to the end of the human race? 60 00:04:09,074 --> 00:04:12,121 One of the leaders in AI development, Facebook, 61 00:04:12,164 --> 00:04:14,602 recently had a surprising incident 62 00:04:14,645 --> 00:04:18,127 that showed how unpredictable and uncontrollable 63 00:04:18,170 --> 00:04:21,304 artificial intelligence can be. 64 00:04:21,348 --> 00:04:23,872 Just last year, Facebook was shocked 65 00:04:23,915 --> 00:04:26,527 when some of their chatbots started communicating 66 00:04:26,570 --> 00:04:28,572 in a secret language. 67 00:04:28,616 --> 00:04:32,184 [male computer voice] I can I I everything else... 68 00:04:32,228 --> 00:04:34,317 [female computer voice] Balls have a ball to me to me to me 69 00:04:34,361 --> 00:04:36,537 to me to me to me to me... 70 00:04:36,580 --> 00:04:40,932 [Quinto] And no one knew what they were saying. 71 00:04:40,976 --> 00:04:43,718 But how did this happen? 72 00:04:43,761 --> 00:04:47,635 I'm meeting with the chief AI scientist at Facebook to find out. 73 00:04:47,678 --> 00:04:50,507 I wanted to talk to you primarily because I heard of the story 74 00:04:50,551 --> 00:04:53,858 about the chatbots created a language 75 00:04:53,902 --> 00:04:55,425 that allowed them to talk to one another 76 00:04:55,469 --> 00:04:57,862 but that you all couldn't understand. 77 00:05:21,190 --> 00:05:24,454 [Quinto] The challenge of advancements in AI 78 00:05:24,498 --> 00:05:26,456 is that we're supposed to let them do 79 00:05:26,500 --> 00:05:29,981 what they think or learn is right. 80 00:05:30,025 --> 00:05:32,854 But what they think or learn is right 81 00:05:32,897 --> 00:05:35,639 may be really wrong for humanity, 82 00:05:35,683 --> 00:05:38,294 and that is the thing that worries me, 83 00:05:38,338 --> 00:05:42,733 because I think we're going into truly uncharted territory. 84 00:05:42,777 --> 00:05:44,909 There are great minds in the technological world 85 00:05:44,953 --> 00:05:47,521 who have really raised 86 00:05:47,564 --> 00:05:49,958 some serious concerns that this technology 87 00:05:50,001 --> 00:05:52,830 is something that we don't have a complete handle on. 88 00:05:52,874 --> 00:05:55,790 I guess I wonder where you fall in that. 89 00:06:08,803 --> 00:06:11,283 But, I mean, isn't the machine only 90 00:06:11,327 --> 00:06:15,113 as intelligent as the human beings program it to be? 91 00:06:15,157 --> 00:06:15,984 No.No? 92 00:06:25,559 --> 00:06:28,562 [Quinto] We program this AI, for the most part. 93 00:06:28,605 --> 00:06:32,130 But what if it outpaces us or outsmarts us? 94 00:06:32,174 --> 00:06:35,133 There may come a point at which we're up against a force 95 00:06:35,177 --> 00:06:37,919 that we've created but that we can no longer control. 96 00:06:37,962 --> 00:06:43,403 It wouldn't be the first time that humanity plowed into uncharted territory 97 00:06:43,446 --> 00:06:47,537 without fully understanding the range and scope of consequences. 98 00:06:51,715 --> 00:06:56,459 Throughout history, the pursuit of greater technology has been perilous. 99 00:06:56,503 --> 00:07:00,811 Some of the most formidable technologically advanced civilizations in the world 100 00:07:00,855 --> 00:07:04,772 have been completely wiped out by their own hubris. 101 00:07:04,815 --> 00:07:07,296 Like the collapse of the Roman Empire, 102 00:07:07,339 --> 00:07:12,475 which, at the height of its rule, succumbed to barbarian attacks. 103 00:07:12,519 --> 00:07:18,046 Or the Mayans, known as one of the most advanced societies of ancient times. 104 00:07:18,089 --> 00:07:21,745 They dominated Central America for 1,200 years, 105 00:07:21,789 --> 00:07:25,619 only to mysteriously disappear. 106 00:07:25,662 --> 00:07:29,057 Some say it was the Mayans' own scientific achievements 107 00:07:29,100 --> 00:07:33,714 and endless hunger to dominate that led to their downfall. 108 00:07:33,757 --> 00:07:38,675 So if this type of collapse could happen to our most exceptional predecessors, 109 00:07:38,719 --> 00:07:40,808 could it also happen to us? 110 00:07:40,851 --> 00:07:44,246 Are we on the brink of an AI takeover? 111 00:07:54,343 --> 00:07:56,737 I'm visiting a controversial facility 112 00:07:56,780 --> 00:07:59,174 that's becoming one of the first in their industry 113 00:07:59,217 --> 00:08:03,657 to program life-sized dolls with artificial intelligence. 114 00:08:08,226 --> 00:08:11,795 They call it RealDoll. 115 00:08:14,885 --> 00:08:17,018 [man] Every doll we make is a custom order. 116 00:08:17,061 --> 00:08:20,064 Every single doll is made exactly to the specifications 117 00:08:20,108 --> 00:08:21,544 that the person asked for. 118 00:08:21,588 --> 00:08:24,678 What kind of volume do you produce? 119 00:08:24,721 --> 00:08:27,071 [Matt] On average, anywhere from 300 to 400 a year. 120 00:08:27,115 --> 00:08:29,552 Mm-hmm. All over the world. 121 00:08:29,596 --> 00:08:34,644 What's the price range? About 7,000 or 8,000. 122 00:08:34,688 --> 00:08:39,954 The bigger choices are the body type and the face that they like. 123 00:08:39,997 --> 00:08:43,087 And then from there, they're gonna choose skin color, 124 00:08:43,131 --> 00:08:46,308 makeup, what kind of manicure and pedicure they wa-- 125 00:08:46,351 --> 00:08:50,355 I mean, really, everything is customizable on them. 126 00:08:50,399 --> 00:08:52,575 So amazing. Is it okay if I touch them?Yeah, please. 127 00:08:52,619 --> 00:08:54,272 So they do move. 128 00:08:54,316 --> 00:08:56,710 Yeah, so every doll has, inside of it, 129 00:08:56,753 --> 00:09:00,278 a skeleton that mimics the range of motion of a human being. 130 00:09:00,322 --> 00:09:04,239 Wow. And then this is, like, a facial skeleton. I mean-- 131 00:09:04,282 --> 00:09:08,156 Yeah, this is the skull that's under the skin. 132 00:09:08,199 --> 00:09:10,767 The face is attached using magnets. 133 00:09:10,811 --> 00:09:14,336 So they can remove a face and interchange it. Right. 134 00:09:14,379 --> 00:09:18,427 Wow. That's amazing.Yeah. 135 00:09:18,470 --> 00:09:21,386 Now you're starting to incorporate 136 00:09:21,430 --> 00:09:24,085 artificial intelligence in some of these dolls, right?Yeah. 137 00:09:24,128 --> 00:09:26,217 I mean, it was always in the back of my mind, 138 00:09:26,261 --> 00:09:29,090 like, wouldn't it be cool to start animating these dolls? 139 00:09:32,397 --> 00:09:34,704 Wasn't until, like, the last five years 140 00:09:34,748 --> 00:09:37,272 that technology started to present itself 141 00:09:37,315 --> 00:09:39,491 that, wow, this is actually possible. 142 00:09:44,192 --> 00:09:47,064 [Matt] This is what's inside of the head. Wow. 143 00:09:47,108 --> 00:09:50,677 And all of the movements are being controlled, obviously, by the app. 144 00:09:50,720 --> 00:09:53,593 But all of the motors and servos that need to do that 145 00:09:53,636 --> 00:09:55,246 are inside of the head.Mm-hmm. 146 00:09:55,290 --> 00:09:57,814 I'll show you how we can 147 00:09:57,858 --> 00:09:59,947 really easily remove the face.Wow. 148 00:09:59,990 --> 00:10:02,602 See how easy that was?Yeah. 149 00:10:02,645 --> 00:10:05,039 So there is still something very robotic about it. 150 00:10:05,082 --> 00:10:07,824 We're still early enough in this process 151 00:10:07,868 --> 00:10:10,653 that it hasn't become so seamlessly integrated. 152 00:10:10,697 --> 00:10:13,047 Well, yeah, you mean to the point where you can't tell it's a robot. 153 00:10:13,090 --> 00:10:14,788 Right. Right.We're not there yet. 154 00:10:14,831 --> 00:10:17,399 No. Right. And what do you think is that trajectory? 155 00:10:17,442 --> 00:10:20,271 I mean, how far off do you think we are?I think within five years. 156 00:10:20,315 --> 00:10:23,710 Really?I think the facial expressions and all of that 157 00:10:23,753 --> 00:10:26,930 will come up to that level very quickly.Wow. 158 00:10:26,974 --> 00:10:29,237 I think there will be robots among us just like, 159 00:10:29,280 --> 00:10:32,022 you know, we see in all these movies that we all grew up watching. 160 00:10:32,066 --> 00:10:33,589 Right. 161 00:10:33,633 --> 00:10:35,896 This is Harmony. 162 00:10:35,939 --> 00:10:39,073 I'm just remotely controlling her right now so you can see.Wow. 163 00:10:39,116 --> 00:10:42,206 Um, but she can smile.Whoa. 164 00:10:42,250 --> 00:10:44,861 If you wanna ask her anything, you can. 165 00:10:44,905 --> 00:10:46,384 Really?What's your name? 166 00:10:48,735 --> 00:10:51,172 Wow. That's crazy. 167 00:10:51,215 --> 00:10:53,653 Is this the artificial intelligence technology? 168 00:10:53,696 --> 00:10:55,089 This is the AI.Uh-huh. 169 00:10:55,132 --> 00:10:58,005 Runs on a device like an iPad.Uh-huh. 170 00:10:58,048 --> 00:11:00,485 And it's communicating with the robot. 171 00:11:00,529 --> 00:11:02,009 So this is hardware.Uh-huh. 172 00:11:02,052 --> 00:11:03,750 There's no brain in her head.Got it. 173 00:11:03,793 --> 00:11:05,926 It's just actuators and movement.Got it. 174 00:11:05,969 --> 00:11:07,144 [Quinto] Harmony, where you from? 175 00:11:21,768 --> 00:11:22,899 [laughs] 176 00:11:34,084 --> 00:11:35,564 Coming up...What kind of per[Quinto] They do move. 177 00:11:35,607 --> 00:11:37,914 Yeah, so every doll has, insi of it, 178 00:11:37,958 --> 00:11:41,918 a skeleton that mimics the range of motion of a human being. 179 00:11:41,962 --> 00:11:43,528 Wow. 180 00:11:43,572 --> 00:11:46,053 At a company called RealDoll, 181 00:11:46,096 --> 00:11:48,838 they're making artificially intelligent robots 182 00:11:48,882 --> 00:11:53,277 that look remarkably lifelike. 183 00:11:53,321 --> 00:11:56,716 But what will happen as the intelligence of these dolls grows? 184 00:11:56,759 --> 00:12:00,023 Could they band together against their human makers? 185 00:12:09,119 --> 00:12:10,294 [laughs] 186 00:12:12,862 --> 00:12:17,998 So, do you feel a connection to her in particular? 187 00:12:18,041 --> 00:12:21,392 You know, I think of Harmony as a character, as a person, more than a machine. 188 00:12:21,436 --> 00:12:23,133 Uh-huh.Um, and it's funny, 189 00:12:23,177 --> 00:12:25,875 because I have more than one Harmony robot. 190 00:12:25,919 --> 00:12:27,921 Uh-huh.So whether she's connected 191 00:12:27,964 --> 00:12:30,053 to this robot or that one, it doesn't matter. 192 00:12:30,097 --> 00:12:32,621 It's still Harmony.Right. 193 00:12:32,664 --> 00:12:34,797 Then there's that guy. 194 00:12:34,841 --> 00:12:37,191 I literally just built him, so he doesn't-- Oh, really? 195 00:12:37,234 --> 00:12:39,323 His head's not even done yet. 196 00:12:43,588 --> 00:12:48,202 But his whole face just comes right off. Wow. 197 00:12:48,245 --> 00:12:50,813 And the eyes are removable.Uh-huh. 198 00:12:50,857 --> 00:12:55,035 So you can change the eye color to whatever eye color you want.Wow. 199 00:12:55,078 --> 00:12:57,341 This one has a little better lip sync on the movement.Uh-huh. 200 00:12:57,385 --> 00:13:00,736 And also, her eyes look a little more realistic.Right. 201 00:13:00,780 --> 00:13:03,260 In the app, you have a persona section. 202 00:13:03,304 --> 00:13:06,655 And these are all these traits that you can choose. 203 00:13:06,698 --> 00:13:09,571 What kind of personality do you have? 204 00:13:16,796 --> 00:13:18,493 So, anyway--It's very Westworld. 205 00:13:18,536 --> 00:13:19,320 Yeah, it is.Right? 206 00:13:22,279 --> 00:13:25,108 And you can go in and change the personality. 207 00:13:25,152 --> 00:13:27,328 So, like, if she's really insecure-- 208 00:13:27,371 --> 00:13:30,244 Uh-huh. Uh-huh....she'll constantly ask you if you still like her. 209 00:13:30,287 --> 00:13:32,681 If she's jealous, she'll randomly tell you 210 00:13:32,724 --> 00:13:34,857 that she was stalking your Facebook page. Uh-huh. 211 00:13:34,901 --> 00:13:38,600 So if we make her really jealous and insecure, not so funny--Uh-huh. 212 00:13:38,643 --> 00:13:42,996 ...but talkative and moody... 213 00:13:43,039 --> 00:13:44,867 So, I can show you that one. 214 00:13:47,783 --> 00:13:49,219 Are you jealous? 215 00:14:02,406 --> 00:14:04,887 There will, seems like, be a point at which 216 00:14:04,931 --> 00:14:07,281 they will have a kind of self-awareness, right? 217 00:14:07,324 --> 00:14:08,848 Yeah. And, in fact, 218 00:14:08,891 --> 00:14:10,806 I think it's going to bring 219 00:14:10,850 --> 00:14:13,113 a lot of risk with it. 220 00:14:13,156 --> 00:14:15,158 That self-awareness will breed 221 00:14:15,202 --> 00:14:16,986 artificial emotions. 222 00:14:17,030 --> 00:14:19,989 And, like, not just pretending to feel something, 223 00:14:20,033 --> 00:14:23,950 but really feel feeling and perceiving the way we do.Right. 224 00:14:23,993 --> 00:14:27,910 Right. Because that is where self-preservation comes from. 225 00:14:27,954 --> 00:14:30,347 And when you give that to a machine, 226 00:14:30,391 --> 00:14:33,698 I think it's kind of a Pandora's box. 227 00:14:33,742 --> 00:14:37,615 [Quinto] The idea of robots becoming self-aware-- 228 00:14:37,659 --> 00:14:39,704 that's the thing that's a little bit scary. 229 00:14:39,748 --> 00:14:41,793 And right now, it still feels very programmed. 230 00:14:41,837 --> 00:14:43,926 It feels very much like, "Oh, yeah, 231 00:14:43,970 --> 00:14:45,841 they understand what we tell them to understand." 232 00:14:45,885 --> 00:14:48,104 Well, what happens when it's not programmed anymore? 233 00:14:48,148 --> 00:14:50,367 Eventually, there's no two ways about it. 234 00:14:50,411 --> 00:14:53,022 We're gonna come up against problems we can't even imagine. 235 00:14:57,940 --> 00:15:00,725 As this technology advances, 236 00:15:00,769 --> 00:15:05,121 we may be facing a future of great friction between humanity and machines. 237 00:15:05,165 --> 00:15:08,516 But some experts argue artificial intelligence 238 00:15:08,559 --> 00:15:12,128 could already be a serious threat to civilization. 239 00:15:12,172 --> 00:15:15,262 With the latest generation of artificial intelligence 240 00:15:15,305 --> 00:15:19,092 that uses machine learning instead of a human programming the computer, 241 00:15:19,135 --> 00:15:22,051 we have the computer, to a certain extent, programming itself. 242 00:15:22,095 --> 00:15:25,228 And as AI gets more and more capable, 243 00:15:25,272 --> 00:15:27,274 ethical dilemmas that you encounter 244 00:15:27,317 --> 00:15:29,319 are only gonna get more and more complicated. 245 00:15:29,363 --> 00:15:32,496 You know,Terminator scenario, whereby 246 00:15:32,540 --> 00:15:36,413 an artificial intelligence becomes smarter than the smartest human being, 247 00:15:36,457 --> 00:15:39,068 and then decides that human beings are obsolete, 248 00:15:39,112 --> 00:15:41,157 and tries to take over the world. 249 00:15:41,201 --> 00:15:43,203 You know, I think that artificial intelligence 250 00:15:43,246 --> 00:15:45,248 and the rise of artificial intelligence, 251 00:15:45,292 --> 00:15:47,685 and increasingly capable artificial intelligence, 252 00:15:47,729 --> 00:15:50,036 will pose a threat to global security 253 00:15:50,079 --> 00:15:54,083 long before we get to the sort of superintelligence or Skynet scenario. 254 00:15:54,127 --> 00:15:56,912 In fact, I would say the rise of artificial intelligence 255 00:15:56,956 --> 00:15:59,088 poses a threat to global security right now. 256 00:16:02,831 --> 00:16:05,225 [Quinto] Despite these warnings of imminent danger, 257 00:16:05,268 --> 00:16:09,055 there are many scientists who want to push machines even further. 258 00:16:09,098 --> 00:16:13,450 Inside the Human-Robot Interaction Lab at Tufts University-- 259 00:16:13,494 --> 00:16:16,540 Please. ...researchers are breathing life 260 00:16:16,584 --> 00:16:20,109 into superintelligent machines that can perform complex 261 00:16:20,153 --> 00:16:22,155 and often dangerous tasks. 262 00:16:22,198 --> 00:16:25,767 [man] This is a robot we use for social interactions. 263 00:16:25,810 --> 00:16:27,943 It has nice big eyes. 264 00:16:27,987 --> 00:16:31,207 This robot we use to do experiments 265 00:16:31,251 --> 00:16:33,775 with social connection and affect expression. 266 00:16:33,818 --> 00:16:35,864 If I change the eyebrows ever so slightly, 267 00:16:35,907 --> 00:16:38,084 it looks much more angry. 268 00:16:40,086 --> 00:16:42,914 And then this is our PR2 robot that we use to study 269 00:16:42,958 --> 00:16:45,787 different ways in which a robot can learn tasks very quickly. 270 00:16:45,830 --> 00:16:48,094 We call it "one-shot learning." 271 00:16:48,137 --> 00:16:51,488 In particular, manipulation tasks, such as picking up objects 272 00:16:51,532 --> 00:16:54,230 and learning how to interact with people with objects. 273 00:16:54,274 --> 00:16:57,668 Okay. And it can see us?Uh, yes. 274 00:16:57,712 --> 00:17:00,715 So it has various sensors here. It has a sensor here 275 00:17:00,758 --> 00:17:03,065 that allows it to see 3-D information, 276 00:17:03,109 --> 00:17:05,894 and it uses that information to learn about objects 277 00:17:05,937 --> 00:17:08,549 it hasn't seen before.I see. 278 00:17:08,592 --> 00:17:12,118 [Quinto] Today, this robot is learning a very dangerous skill: 279 00:17:12,161 --> 00:17:15,730 how to handle incredibly sharp knives and pass them to humans. 280 00:17:19,168 --> 00:17:21,040 This object is a knife. 281 00:17:22,824 --> 00:17:24,478 A knife is used for cutting. 282 00:17:26,480 --> 00:17:28,351 The orange part of the knife is the handle. 283 00:17:29,918 --> 00:17:32,007 The gray part of the knife is the blade. 284 00:17:34,531 --> 00:17:36,446 If it picks up the knife by the handle, 285 00:17:36,490 --> 00:17:38,492 we don't want it stabbing you with the blade.Mm-hmm. Mm... 286 00:17:38,535 --> 00:17:41,234 So we will teach it how to safely pass me the knife. 287 00:17:41,277 --> 00:17:45,151 To pass me the knife, grab the knife by the blade. 288 00:17:45,194 --> 00:17:47,109 All right. 289 00:17:57,859 --> 00:17:59,774 That's amazing. 290 00:17:59,817 --> 00:18:01,776 It's incredible to see how quickly 291 00:18:01,819 --> 00:18:03,430 a robot can learn a new skill. 292 00:18:03,473 --> 00:18:05,475 But this is just the beginning. 293 00:18:05,519 --> 00:18:07,956 There are now robots that can teach each other 294 00:18:07,999 --> 00:18:10,959 complex skills instantly. 295 00:18:11,002 --> 00:18:14,484 In other words, the machines themselves could have the ability 296 00:18:14,528 --> 00:18:18,184 to create AI armies and spread skills of their own. 297 00:18:20,055 --> 00:18:22,666 Machines that learn and teach for themselves 298 00:18:22,710 --> 00:18:26,801 are actually in the lab at Tufts. 299 00:18:26,844 --> 00:18:30,065 [Matthias] These are robots that can learn new information on the fly. 300 00:18:30,109 --> 00:18:32,894 So you can teach them new concepts or new tricks. 301 00:18:32,937 --> 00:18:37,072 These three guys in particular can listen in on each other's conversations. 302 00:18:37,116 --> 00:18:41,337 And so, if one learns something or hears something, the other one will know it. 303 00:18:41,381 --> 00:18:44,123 [Quinto] But if these robots learn and share 304 00:18:44,166 --> 00:18:47,952 new information telepathically with other robots, 305 00:18:47,996 --> 00:18:51,565 then, in theory, when I teach one robot a new skill, 306 00:18:51,608 --> 00:18:55,046 all of the robots will gain that skill. 307 00:19:01,314 --> 00:19:03,446 Hello, Bays. Hello. 308 00:19:03,490 --> 00:19:05,492 Raise your arms. Okay. 309 00:19:06,710 --> 00:19:09,278 Crouch down. Okay. 310 00:19:10,540 --> 00:19:13,152 Stand up. Okay. 311 00:19:14,675 --> 00:19:16,677 Lower your arms. Okay. 312 00:19:18,026 --> 00:19:20,507 That is how you do a squat. 313 00:19:20,550 --> 00:19:23,597 And now the other robot will be able to do it as well. Okay. 314 00:19:23,640 --> 00:19:26,861 Demster, do a squat. 315 00:19:26,904 --> 00:19:28,819 Okay. 316 00:19:28,863 --> 00:19:30,734 Weird. So that one taught that one how to do it? 317 00:19:33,215 --> 00:19:35,348 So creepy. 318 00:19:35,391 --> 00:19:37,393 This one can do it as well. 319 00:19:37,437 --> 00:19:40,179 Schaeffer, do a squat. Okay. 320 00:19:42,398 --> 00:19:44,270 [Quinto] The interconnectivity of the robots 321 00:19:44,313 --> 00:19:46,359 was a little bit unsettling. 322 00:19:46,402 --> 00:19:48,056 Just in terms of, like, all you have to do 323 00:19:48,099 --> 00:19:50,232 is teach one one thing, and then they all know it. 324 00:19:50,276 --> 00:19:54,323 But I also found it could be the potential portal for disaster. 325 00:19:54,367 --> 00:19:57,239 It's like, "Oh, wait a minute. One of them can learn how to shoot a gun, 326 00:19:57,283 --> 00:20:00,199 and then they all know how to do it." 327 00:20:00,242 --> 00:20:03,680 Luckily, those robots are kind of small enough that you'd be able to kick it out of the way. 328 00:20:03,724 --> 00:20:08,424 But what if it's a 250-pound robot made of steel and iron and metal, and then what? 329 00:20:08,468 --> 00:20:11,384 What if they begin communicating on their own 330 00:20:11,427 --> 00:20:14,996 in a secret language, like the chatbots from Facebook? 331 00:20:15,039 --> 00:20:18,652 As these robots are learning, is it such a leap to consider 332 00:20:18,695 --> 00:20:21,785 that these machines will eventually get to a place 333 00:20:21,829 --> 00:20:24,658 where humanity and human interaction 334 00:20:24,701 --> 00:20:27,574 is no longer necessary for their survival? 335 00:20:27,617 --> 00:20:30,794 Uh, if you ask me is it possible, 336 00:20:30,838 --> 00:20:33,797 yes, it is possible, but it very much depends 337 00:20:33,841 --> 00:20:37,279 on how these control systems are structured. 338 00:20:37,323 --> 00:20:39,412 I think the question you raise, of course, 339 00:20:39,455 --> 00:20:43,067 is do we want machines to perform activities and make decisions 340 00:20:43,111 --> 00:20:45,244 for us that we don't understand? 341 00:20:45,287 --> 00:20:47,898 You don't want that.Right. 342 00:20:47,942 --> 00:20:51,337 I feel like we are absolutely on the precipice of a revolution here. 343 00:20:51,380 --> 00:20:54,470 It does feel like the Wild West right now. 344 00:20:54,514 --> 00:20:57,517 We are throwing ourselves into a period of time 345 00:20:57,560 --> 00:20:59,954 when there is conflict and friction 346 00:20:59,997 --> 00:21:03,523 between humanity and machines. 347 00:21:03,566 --> 00:21:06,569 So I think we have to be careful how we're trusting it 348 00:21:06,613 --> 00:21:07,875 and how we're giving it power. 349 00:21:16,231 --> 00:21:18,625 Coming up, freethinking flying drones change the game of war. 350 00:21:18,668 --> 00:21:21,236 to create intelligent machines 351 00:21:21,280 --> 00:21:23,543 could ultimately lead to a robot uprising. 352 00:21:26,720 --> 00:21:29,940 I've already seen how this powerful technology could begin 353 00:21:29,984 --> 00:21:32,639 to act on its own, beyond our control. 354 00:21:32,682 --> 00:21:35,859 [female computer voice] Balls have a ball to me to me to me to me to me to me... 355 00:21:41,996 --> 00:21:46,261 [laughs] 356 00:21:46,305 --> 00:21:49,830 [Quinto] And now I wanna know: what happens if this incredibly advanced technology 357 00:21:49,873 --> 00:21:51,875 gets in the hands of our enemies? 358 00:21:53,964 --> 00:21:56,924 Today, we're already seeing the use of AI weapons 359 00:21:56,967 --> 00:21:59,970 in destabilized regions like the Korean Peninsula, 360 00:22:00,014 --> 00:22:02,843 where they're using Sentinel guns 361 00:22:02,886 --> 00:22:05,976 that can lock in on targets without human oversight. 362 00:22:07,630 --> 00:22:09,937 Some experts warn 363 00:22:09,980 --> 00:22:12,026 that there will be serious consequences 364 00:22:12,069 --> 00:22:14,158 to creating combat-ready robots. 365 00:22:17,423 --> 00:22:20,774 Imagine terrorist groups and insurgencies around the world 366 00:22:20,817 --> 00:22:22,515 using weaponized drones. 367 00:22:25,039 --> 00:22:27,258 Advances in artificial intelligence 368 00:22:27,302 --> 00:22:28,912 only make this more extreme. 369 00:22:28,956 --> 00:22:30,784 This is a future of threats 370 00:22:30,827 --> 00:22:32,351 that we are not really familiar with. 371 00:22:32,394 --> 00:22:35,615 We need to start preparing for this future right now. 372 00:22:35,658 --> 00:22:37,573 We now face a situation 373 00:22:37,617 --> 00:22:40,315 in which the most advance military technologies 374 00:22:40,359 --> 00:22:43,753 are no longer uniquely held by the United States. 375 00:22:43,797 --> 00:22:46,713 Russia and China are making significant advances. 376 00:22:46,756 --> 00:22:49,933 Contrast that with what Israel already has, 377 00:22:49,977 --> 00:22:53,502 which is drones that can hunt for radar signatures, 378 00:22:53,546 --> 00:22:55,199 and then release a bomb. 379 00:22:57,506 --> 00:23:01,118 If long-range drone delivery becomes a reality, 380 00:23:01,162 --> 00:23:03,643 stakes are as high as they can possibly be. 381 00:23:03,686 --> 00:23:07,734 Drones get better and more capable and cheaper every year. 382 00:23:07,777 --> 00:23:10,432 So we might see a future where capabilities 383 00:23:10,476 --> 00:23:14,044 that previously were restricted to advanced militaries 384 00:23:14,088 --> 00:23:17,439 with millions of dollars in budgets are suddenly available to terrorists 385 00:23:17,483 --> 00:23:18,614 willing to spend hundreds of dollars. 386 00:23:23,314 --> 00:23:25,926 [Quinto] While many fear we could be facing grave dangers 387 00:23:25,969 --> 00:23:28,929 in a world of weaponized AI, 388 00:23:28,972 --> 00:23:31,018 others see the enormous potential 389 00:23:31,061 --> 00:23:33,760 for artificial intelligence to serve a greater good, 390 00:23:33,803 --> 00:23:37,459 and even save lives. 391 00:23:37,503 --> 00:23:40,114 Specialized drones are already performing 392 00:23:40,157 --> 00:23:42,986 a host of dangerous tasks around the world, 393 00:23:43,030 --> 00:23:46,555 from collecting vital intelligence in enemy combat zones 394 00:23:46,599 --> 00:23:50,124 to delivering food, medical supplies, and water 395 00:23:50,167 --> 00:23:53,649 to disaster areas that first responders cannot reach. 396 00:23:53,693 --> 00:23:58,524 There are real ways in which AI can benefit civilization. 397 00:24:01,614 --> 00:24:04,312 So today, I'm investigating the drone technology 398 00:24:04,355 --> 00:24:06,314 created at Carnegie Mellon University, 399 00:24:06,357 --> 00:24:10,013 meant to be deployed on dangerous rescue missions. 400 00:24:10,057 --> 00:24:12,363 We create intelligent individual vehicles 401 00:24:12,407 --> 00:24:14,583 that we would use for everything from 402 00:24:14,627 --> 00:24:17,368 autonomous inspections, or looking at a bridge, 403 00:24:17,412 --> 00:24:21,068 or some kind of infrastructure, or going into a radioactive environment. 404 00:24:21,111 --> 00:24:23,549 But what consequences could arise 405 00:24:23,592 --> 00:24:26,508 as human beings enable powerful, autonomous drones 406 00:24:26,552 --> 00:24:29,032 to think for themselves? 407 00:24:29,076 --> 00:24:33,776 And what happens if these smart drones get into the wrong hands? 408 00:24:33,820 --> 00:24:37,780 So, what you see here is a team of aerial robots, 409 00:24:37,824 --> 00:24:40,522 where each one is individually intelligent. 410 00:24:40,566 --> 00:24:43,699 But at the same time, collectively, they're working together. 411 00:24:43,743 --> 00:24:46,354 So you can imagine large numbers of robots 412 00:24:46,397 --> 00:24:48,617 who accrue knowledge, share that knowledge, 413 00:24:48,661 --> 00:24:51,315 and collectively improve their proficiency. 414 00:24:51,359 --> 00:24:53,404 So you program the robots 415 00:24:53,448 --> 00:24:56,233 with a certain kind of artificial intelligence 416 00:24:56,277 --> 00:24:59,454 which then allows them to determine where to go, 417 00:24:59,498 --> 00:25:02,892 but you're not actually controlling the robots themselves. 418 00:25:02,936 --> 00:25:05,721 Yes and no. What we're doing is, 419 00:25:05,765 --> 00:25:09,203 we're enabling them to learn and adapt over time, 420 00:25:09,246 --> 00:25:11,771 and then exploiting that learned capability in the future. 421 00:25:11,814 --> 00:25:14,904 So they're exponentially advancing. Yes. 422 00:25:14,948 --> 00:25:17,559 That's so crazy to me. 423 00:25:17,603 --> 00:25:21,128 I'm just, like-- I find it so overwhelming. 424 00:25:21,171 --> 00:25:24,261 I don't know. Like, we're just in this territory which is 425 00:25:24,305 --> 00:25:26,873 so unknown. 426 00:25:26,916 --> 00:25:28,831 Not only do we not have a map, 427 00:25:28,875 --> 00:25:32,095 but nobody really seems that interested in drawing one. 428 00:25:32,139 --> 00:25:35,577 And that, to me, is the thing that worries me. 429 00:25:35,621 --> 00:25:39,799 Because it's like, yeah, these people are all clearly very well-intentioned, 430 00:25:39,842 --> 00:25:41,583 very intelligent. 431 00:25:41,627 --> 00:25:44,717 Who's to say what somebody in a DIY 432 00:25:44,760 --> 00:25:47,110 workshop across town is doing with it? 433 00:25:47,154 --> 00:25:49,112 Attaching flamethrowers to the robots 434 00:25:49,156 --> 00:25:52,855 and sending them after pets in the neighborhood. 435 00:25:52,899 --> 00:25:55,554 [Nathan] So what you're going to see are the team of robots 436 00:25:55,597 --> 00:26:00,515 are working together in order to achieve an overall group objective. 437 00:26:00,559 --> 00:26:02,561 [Quinto] Within a caged-off test area, 438 00:26:02,604 --> 00:26:06,477 the team will deploy 25 to 30 of these aerial drones, 439 00:26:06,521 --> 00:26:08,784 one at a time. 440 00:26:08,828 --> 00:26:11,439 Their directive: form a continuous 441 00:26:11,482 --> 00:26:14,442 circular flight path without crashing into each other 442 00:26:14,485 --> 00:26:17,097 or anything else. 443 00:26:17,140 --> 00:26:20,317 To make it even harder, I'm going to be a human obstacle. 444 00:26:25,540 --> 00:26:28,064 All right, so now we're gonna start. 445 00:26:28,108 --> 00:26:30,719 We should see robots start to take off. Okay. 446 00:26:36,595 --> 00:26:38,553 Ooh! Great! 447 00:26:51,871 --> 00:26:54,308 [Quinto] Researchers at Carnegie Mellon University 448 00:26:54,351 --> 00:26:56,310 are designing fleets of aerial drones 449 00:26:56,353 --> 00:26:58,617 that can work together without oversight. 450 00:27:00,270 --> 00:27:02,272 But what if these autonomous drones 451 00:27:02,316 --> 00:27:04,840 really can think for themselves? 452 00:27:04,884 --> 00:27:08,670 Is it possible that one of the drones, or even all of them, 453 00:27:08,714 --> 00:27:11,586 could decide to go rogue and turn against their maker? 454 00:27:14,676 --> 00:27:17,418 All right, so now we should see robots start to take off.Okay. 455 00:27:17,461 --> 00:27:19,028 We'll give you a thumb.All right. 456 00:27:31,867 --> 00:27:33,129 So weird! 457 00:27:35,566 --> 00:27:37,656 They're very insect-like. 458 00:27:37,699 --> 00:27:39,875 I feel like they're watching me. 459 00:27:39,919 --> 00:27:42,617 These intelligent drones think on their own. 460 00:27:42,661 --> 00:27:44,706 And if one falls out of line, 461 00:27:44,750 --> 00:27:49,102 another will rise up to fill its place. 462 00:27:49,145 --> 00:27:52,409 [man] Now you have one to your left that's coming in to replace the one that fell. 463 00:27:52,453 --> 00:27:54,020 Oh, that one?Yeah. 464 00:27:54,063 --> 00:27:56,022 [Ellen] But if you want to throw that one out-- 465 00:27:56,065 --> 00:27:59,199 [man] Here comes the replacement. Behind you. 466 00:27:59,242 --> 00:28:01,462 [Quinto] Even if I try to knock one out of line, 467 00:28:01,505 --> 00:28:03,507 the others will keep coming... 468 00:28:05,553 --> 00:28:06,902 until they achieve their goal. 469 00:28:11,820 --> 00:28:13,126 I learned their weakness. 470 00:28:16,607 --> 00:28:18,305 All right. That's pretty amazing. 471 00:28:18,348 --> 00:28:20,829 Being in there, it did feel like 472 00:28:20,873 --> 00:28:23,353 there was an insect quality to the robots, which got me thinking 473 00:28:23,397 --> 00:28:27,314 about the, kind of, hive mentality, right? 474 00:28:27,357 --> 00:28:30,491 So, do you feel that there's a point in the evolution of this science 475 00:28:30,534 --> 00:28:35,235 where we'll no longer be able to control the intelligence? 476 00:28:35,278 --> 00:28:37,193 Yeah, that's a really tough question, 477 00:28:37,237 --> 00:28:40,893 and that's a major consideration. 478 00:28:40,936 --> 00:28:42,503 A lot of people are conjecturing on it. 479 00:28:42,546 --> 00:28:44,897 And I think with any technology that we develop, 480 00:28:44,940 --> 00:28:47,464 we always have the capacity to go too far. 481 00:28:49,684 --> 00:28:52,382 If this intelligence evolves as quickly 482 00:28:52,426 --> 00:28:55,777 as they're talking about it evolving beyond our own, 483 00:28:55,821 --> 00:28:58,693 then I feel like we're potentially 484 00:28:58,737 --> 00:29:00,434 in a really precarious situation. 485 00:29:07,833 --> 00:29:10,052 As these thinking machines get equipped 486 00:29:10,096 --> 00:29:12,707 with more dangerous and powerful capabilities, 487 00:29:12,751 --> 00:29:17,799 it's possible they will be used for both good and evil, 488 00:29:17,843 --> 00:29:20,671 depending on who's in charge. 489 00:29:20,715 --> 00:29:23,892 What happens if humans lose control 490 00:29:23,936 --> 00:29:27,330 of artificial intelligence altogether? 491 00:29:27,374 --> 00:29:30,246 Could it bring on the ultimate showdown-- 492 00:29:30,290 --> 00:29:32,509 man versus machine? 493 00:29:37,123 --> 00:29:39,690 And if we face off, who would win? 494 00:29:42,302 --> 00:29:45,435 In California, there's a group of robotics engineers 495 00:29:45,479 --> 00:29:47,960 confronting this question head-on. 496 00:29:48,003 --> 00:29:50,876 Mike Winter runs a competition 497 00:29:50,919 --> 00:29:53,313 where the best robotic builders come to show off 498 00:29:53,356 --> 00:29:56,925 their most destructive and strategic combat robots 499 00:29:56,969 --> 00:29:58,971 to battle to the death. 500 00:29:59,014 --> 00:30:01,451 It's called AI or Die. 501 00:30:01,495 --> 00:30:04,846 Competitors fall into two categories: 502 00:30:04,890 --> 00:30:07,631 robots controlled by human operators, 503 00:30:07,675 --> 00:30:10,939 and AI robots programmed to fight for themselves. 504 00:30:10,983 --> 00:30:14,160 And many of these machines are incredibly dangerous, 505 00:30:14,203 --> 00:30:17,163 complete with high-velocity, rotating blades. 506 00:30:17,206 --> 00:30:18,338 [whirring] 507 00:30:23,169 --> 00:30:25,954 We have to be really careful around the robots. They're dangerous. 508 00:30:25,998 --> 00:30:29,349 They were made to be dangerous. These things are fast. 509 00:30:29,392 --> 00:30:31,786 [Quinto] So what's the motive of the robots you're putting into the contest? 510 00:30:31,830 --> 00:30:34,789 To destroy other robots?Yeah. 511 00:30:34,833 --> 00:30:37,400 And this is what tells it that it's a robot.It's like a target. 512 00:30:37,444 --> 00:30:40,926 Uh-huh. Just drives robots crazy when they see it. 513 00:30:40,969 --> 00:30:44,930 These tags are what allow the AI robots to know that that's their target. 514 00:30:44,973 --> 00:30:47,497 They're never programmed to hurt humans. 515 00:30:47,541 --> 00:30:50,370 But if I were wearing one of those tags, 516 00:30:50,413 --> 00:30:52,763 that spinning blade would come right for me. 517 00:30:54,853 --> 00:30:56,985 But the application of these machines 518 00:30:57,029 --> 00:30:59,683 goes well beyond spinning blades and the battle arena. 519 00:31:02,643 --> 00:31:05,428 Similar robots are conducting risky missions 520 00:31:05,472 --> 00:31:08,214 to defuse bombs around the world 521 00:31:08,257 --> 00:31:13,393 in areas too treacherous for our military personnel to venture into. 522 00:31:13,436 --> 00:31:17,005 In operating rooms, autonomous robot surgeons 523 00:31:17,049 --> 00:31:22,228 are increasingly being utilized to perform complex procedures on human patients, 524 00:31:22,271 --> 00:31:25,492 often in less time and with far greater precision 525 00:31:25,535 --> 00:31:29,322 than regular human doctors. 526 00:31:29,365 --> 00:31:31,454 But the future of these machines 527 00:31:31,498 --> 00:31:34,327 and their capacity for good or evil 528 00:31:34,370 --> 00:31:37,721 has potentially dangerous implications. 529 00:31:37,765 --> 00:31:43,205 What happens if there is a point when AI-driven robots 530 00:31:43,249 --> 00:31:45,468 somehow see humanity as an obstacle? 531 00:31:45,512 --> 00:31:47,688 It's still, like, us programming. 532 00:31:47,731 --> 00:31:49,733 For now.For now, yeah. 533 00:31:51,344 --> 00:31:54,129 We get to decide what they want. 534 00:31:54,173 --> 00:31:55,870 We're telling them what the reward is. 535 00:31:55,914 --> 00:31:58,177 So, you're not so worried about the robots. 536 00:31:58,220 --> 00:31:59,961 You're worried about the people programming them. 537 00:32:00,005 --> 00:32:01,528 There's good and bad people in this world. 538 00:32:01,571 --> 00:32:03,269 There's gonna be good and bad AI in this world. 539 00:32:04,835 --> 00:32:06,489 Shall we do this?Yeah, let's do it. 540 00:32:06,533 --> 00:32:07,751 See who's the dominant species? 541 00:32:11,146 --> 00:32:15,237 [woman] All right. The robots are ready. 542 00:32:20,112 --> 00:32:24,333 [Quinto] Three, two, one. 543 00:32:24,377 --> 00:32:26,466 Whoa! Aah!Oh! 544 00:32:32,385 --> 00:32:34,517 [Quinto] The desire to create sentient machines 545 00:32:34,561 --> 00:32:37,390 can be traced back centuries. 546 00:32:37,433 --> 00:32:40,784 Leonardo da Vinci famously designed a roboticized warrior 547 00:32:40,828 --> 00:32:43,309 in the late 15th century. 548 00:32:43,352 --> 00:32:46,138 And in some ways, mankind has been trying 549 00:32:46,181 --> 00:32:49,445 to improve on those designs ever since. 550 00:32:49,489 --> 00:32:53,536 What if these devices could somehow outsmart their human makers? 551 00:33:00,326 --> 00:33:03,329 In a warehouse in Northern California, 552 00:33:03,372 --> 00:33:06,506 a group of experts is trying to answer this question. 553 00:33:06,549 --> 00:33:09,335 We have to be really careful around the robots. They're dangerous. 554 00:33:09,378 --> 00:33:12,860 They were made to be dangerous. These things are fast. 555 00:33:12,903 --> 00:33:18,648 How often have you, as a human operator, beat the AI machines? 556 00:33:18,692 --> 00:33:19,780 So far, all of them. 557 00:33:24,611 --> 00:33:26,047 Ready to go? 558 00:33:26,091 --> 00:33:30,312 Three, two, one. 559 00:33:30,356 --> 00:33:32,010 [whirring] 560 00:33:38,146 --> 00:33:40,018 [woman, man exclaim] 561 00:33:40,061 --> 00:33:41,019 [Mike] AI won. Whoo! 562 00:33:42,324 --> 00:33:44,500 That weapon is strong. 563 00:33:44,544 --> 00:33:48,983 [Quinto] All right, so, you lost. 564 00:33:49,027 --> 00:33:50,593 That was a really good hit. 565 00:33:50,637 --> 00:33:53,205 I would say so.I think we should fight more. 566 00:33:53,248 --> 00:33:55,207 Should we try the new robot? 567 00:33:55,250 --> 00:33:57,644 So, everybody does have their safety glasses on? 568 00:33:57,687 --> 00:34:00,081 'Cause this one kind of scares me a little bit. 569 00:34:00,125 --> 00:34:03,345 This one has an actual saw blade on it, so it's much more dangerous 570 00:34:03,389 --> 00:34:04,738 in that it's sharpened and ready for cutting. 571 00:34:11,310 --> 00:34:15,792 Let's see what happens on three, two, one. 572 00:34:15,836 --> 00:34:18,273 [robots whirring] 573 00:34:24,584 --> 00:34:26,325 [Mike exclaims] 574 00:34:26,368 --> 00:34:28,240 [whirring continues] 575 00:34:30,764 --> 00:34:33,549 Oh! Oh, my God! Oh! 576 00:34:33,593 --> 00:34:36,204 AI surrenders.Yes. 577 00:34:41,253 --> 00:34:44,082 Yeah, it got some good damage.Got some good damage. 578 00:34:44,125 --> 00:34:47,259 Well, that gives me some faith in humanity, I guess. Right? 579 00:34:49,261 --> 00:34:52,742 For now, it's a tie between AI and humans. 580 00:34:52,786 --> 00:34:56,137 But as AI advances, will this always be the case? 581 00:35:04,014 --> 00:35:07,409 It's one thing to see remote-controlled robots attack each other. 582 00:35:07,453 --> 00:35:10,238 But what happens when we program AI 583 00:35:10,282 --> 00:35:13,807 into 10,000-pound machines traveling at high speeds, 584 00:35:13,850 --> 00:35:16,375 like self-driving cars? 585 00:35:16,418 --> 00:35:21,162 These vehicles have already hit the road in several major cities, 586 00:35:21,206 --> 00:35:24,992 and will ultimately face decisions in life-threatening moments. 587 00:35:25,035 --> 00:35:29,605 But when given a choice, who will these vehicles ultimately protect, 588 00:35:29,649 --> 00:35:33,000 their own passengers or pedestrians? 589 00:35:33,043 --> 00:35:37,526 Some say that decision is far too important for a machine to make. 590 00:35:37,570 --> 00:35:40,442 [siren wailing] 591 00:35:40,486 --> 00:35:43,663 And in March of 2018, the world saw a glimpse 592 00:35:43,706 --> 00:35:45,969 of how serious that question really is... 593 00:35:49,321 --> 00:35:52,106 when news broke that a cyclist in Phoenix, Arizona, 594 00:35:52,150 --> 00:35:57,372 was struck and killed by an Uber self-driving car. 595 00:35:57,416 --> 00:36:00,723 So should we really trust this level of artificial intelligence 596 00:36:00,767 --> 00:36:02,072 with our very lives? 597 00:36:07,165 --> 00:36:11,604 I'm here at Uber's self-driving test facility to find out for myself. 598 00:36:11,647 --> 00:36:13,693 Their Pittsburgh test course, 599 00:36:13,736 --> 00:36:16,304 in an unmarked industrial complex, 600 00:36:16,348 --> 00:36:18,567 is completely closed off to the public, 601 00:36:18,611 --> 00:36:21,309 replicating an actual city. 602 00:36:21,353 --> 00:36:24,007 And I'm going to be riding in one of these self-driving cars 603 00:36:24,051 --> 00:36:28,751 to see how well it handles some new obstacles. 604 00:36:28,795 --> 00:36:30,144 Hey, Zach.How are you? 605 00:36:30,188 --> 00:36:32,755 Welcome to the ATG's urban test course. 606 00:36:32,799 --> 00:36:36,281 You're looking at 40 acres of urban features. 607 00:36:36,324 --> 00:36:38,674 Great. I'm excited to check it out. All right. I'm gonna hop in. 608 00:36:41,111 --> 00:36:44,811 And then we're on our way.And then we're on our way. 609 00:36:44,854 --> 00:36:48,989 And being driven by artificial intelligence.Yeah. Yeah. Yeah. 610 00:36:49,032 --> 00:36:52,471 When you're driving, is it weird to not put your hands on the wheel? 611 00:36:52,514 --> 00:36:54,908 Yeah. Yeah. It's a little surreal. 612 00:36:57,127 --> 00:36:59,869 How fast do we go?Up to 40 miles per hour. 613 00:36:59,913 --> 00:37:01,349 Okay. 614 00:37:06,093 --> 00:37:08,617 We're at a pedestrian crossing. Right. 615 00:37:08,661 --> 00:37:11,011 So you wait until he gets all the way across before it goes again? 616 00:37:11,054 --> 00:37:13,231 Mm-hmm.Will this technology 617 00:37:13,274 --> 00:37:15,842 automatically protect me before, say, a pedestrian? 618 00:37:15,885 --> 00:37:20,063 Or is it automatically designed to protect a pedestrian above a passenger? 619 00:37:20,107 --> 00:37:22,457 That's a challenging question. Ooh. 620 00:37:24,546 --> 00:37:28,115 What would you do if there's a person jaywalking? 621 00:37:28,158 --> 00:37:31,118 Right. The car will stop.The car will stop. 622 00:37:31,161 --> 00:37:34,817 That's the thing, to me, that I just want people to weigh in on, 623 00:37:34,861 --> 00:37:38,691 because I think we are relying on artificial intelligence to make a decision 624 00:37:38,734 --> 00:37:42,434 that could have life-changing implications for a human being. 625 00:37:42,477 --> 00:37:46,176 I feel like those are conversations that really need to be had. 626 00:37:46,220 --> 00:37:49,615 Yeah. So, we are, obviously, worried and concerned 627 00:37:49,658 --> 00:37:54,359 in making sure that the car does the correct things at those times. 628 00:37:54,402 --> 00:37:57,013 [Quinto] This handing over 629 00:37:57,057 --> 00:37:59,320 the speed and the weight of a vehicle 630 00:37:59,364 --> 00:38:02,454 in real-time, real-life situations, 631 00:38:02,497 --> 00:38:07,633 we are throwing ourselves into an abyss of technology 632 00:38:07,676 --> 00:38:11,593 that's gonna make those decisions for us, and there's no stopping it now. 633 00:38:21,560 --> 00:38:23,518 Coming up, I get in the driver's seat, [Quinto] Throughout my search, 634 00:38:23,562 --> 00:38:26,216 I've discovered a darker side to AI technology... 635 00:38:29,568 --> 00:38:33,049 like the recent accident caused by one of Uber's self-driving cars... 636 00:38:34,703 --> 00:38:36,923 and the looming dangers 637 00:38:36,966 --> 00:38:40,666 of creating thinking machines in our own image. 638 00:38:40,709 --> 00:38:44,060 Like some of the greatest civilizations of the past, 639 00:38:44,104 --> 00:38:46,585 could we face the collapse of modern society 640 00:38:46,628 --> 00:38:49,936 brought on by our own creations? 641 00:38:49,979 --> 00:38:52,504 That day may be fast approaching. 642 00:38:52,547 --> 00:38:55,115 And I want to know, what will it be like 643 00:38:55,158 --> 00:38:57,291 if we hand over all of our control 644 00:38:57,335 --> 00:39:00,555 to one of these extremely powerful thinking machines? 645 00:39:05,865 --> 00:39:09,259 I'm about to get behind the wheel of an Uber self-driving car 646 00:39:09,303 --> 00:39:11,914 to find out for mylf. 647 00:39:11,958 --> 00:39:15,440 Okay. So, remember, there's two modes of operation-- 648 00:39:15,483 --> 00:39:17,180 Right....the manual and the auto. 649 00:39:17,224 --> 00:39:19,444 When you're in auto, you still need to be 650 00:39:19,487 --> 00:39:22,142 that operator that's ready to take control of the vehicle at any time. 651 00:39:22,185 --> 00:39:23,448 Okay.Here we go. 652 00:39:29,454 --> 00:39:32,152 Definitely weirder to be in the driver's seat. 653 00:39:32,195 --> 00:39:35,460 Part of what I like about driving is the control that I have 654 00:39:35,503 --> 00:39:39,289 over where I go, how fast I go. 655 00:39:39,333 --> 00:39:42,510 I feel like there's a part of me that-- 656 00:39:42,554 --> 00:39:46,035 surprise, surprise-- doesn't necessarily want to relinquish that control. 657 00:39:46,079 --> 00:39:49,822 I feel like it's really coming in hot. Yeah. 658 00:39:49,865 --> 00:39:53,260 Now you experience it, like, not as conservative, right?Uh-huh. 659 00:39:55,523 --> 00:39:59,222 I-- Oh. Okay. 660 00:39:59,266 --> 00:40:01,877 Ugh. Ba ba ba... 661 00:40:01,921 --> 00:40:05,707 Oh. Oh. Oh, damn! 662 00:40:05,751 --> 00:40:07,796 Just ran a red light. 663 00:40:07,840 --> 00:40:10,756 There will be a note made of that. 664 00:40:10,799 --> 00:40:13,236 You know, when it ran that red light, what if there was a woman 665 00:40:13,280 --> 00:40:15,935 crossing the street with a stroller in that moment, 666 00:40:15,978 --> 00:40:19,068 where the sun hit the camera at exactly the wrong way? 667 00:40:19,112 --> 00:40:20,548 That's where there are gaps. 668 00:40:23,029 --> 00:40:26,467 What's your feeling about AI? 669 00:40:26,511 --> 00:40:29,775 As it evolves, it feels like 670 00:40:29,818 --> 00:40:32,952 there's no limit to what it can do. 671 00:40:32,995 --> 00:40:35,781 [Anthony] That's something that creates a little bit of fear 672 00:40:35,824 --> 00:40:37,609 and worry in a lot of people. 673 00:40:37,652 --> 00:40:39,828 Those are all valid concerns. 674 00:40:39,872 --> 00:40:44,311 The AI ones are tough to answer.But that's what this is about. 675 00:40:44,354 --> 00:40:47,880 Throughout my journey, I've seen incredible uses for AI, 676 00:40:47,923 --> 00:40:51,536 but I've also learned that we don't really know where it's all headed 677 00:40:51,579 --> 00:40:55,148 or what dangers lie ahead. 678 00:40:55,191 --> 00:40:57,977 In ancient texts and pop culture, 679 00:40:58,020 --> 00:41:01,720 AI has always been set in a distant future. 680 00:41:01,763 --> 00:41:04,723 But with new advancements happening every day, 681 00:41:04,766 --> 00:41:07,639 it seems that future is closer than we know. 682 00:41:10,642 --> 00:41:13,775 [man] There's good and bad people in this world. 683 00:41:13,819 --> 00:41:16,169 There's gonna be good and bad AI in this world. 684 00:41:16,212 --> 00:41:19,433 [Quinto] Our desire to create machines in our likeness 685 00:41:19,477 --> 00:41:22,131 has brought us to a critical turning point in history. 686 00:41:22,175 --> 00:41:26,353 One that could decide the fate of human existence. 687 00:41:26,396 --> 00:41:28,790 And if we're not careful, 688 00:41:28,834 --> 00:41:32,228 our own ambition may one day destroy us. 689 00:41:35,318 --> 00:41:36,406 [female voice laughs] 690 00:41:36,456 --> 00:41:41,006 Repair and Synchronization by Easy Subtitles Synchronizer 1.0.0.0