1
00:00:00,040 --> 00:00:02,480
Speaker 1: Hey, there are odd lots listeners. I'm Tracy Alloway and

2
00:00:02,560 --> 00:00:03,000
Speaker 1: I'm Jill.

3
00:00:03,040 --> 00:00:03,560
Speaker 2: Why isn't though?

4
00:00:03,640 --> 00:00:06,159
Speaker 1: And we want to welcome you to a special presentation

5
00:00:06,280 --> 00:00:09,720
Speaker 1: of the podcast search Engine. We all know that artificial

6
00:00:09,760 --> 00:00:13,120
Speaker 1: intelligence might replace all sorts of jobs humans do today,

7
00:00:13,360 --> 00:00:16,799
Speaker 1: but for most of us that's still mostly theoretical. There's

8
00:00:16,840 --> 00:00:19,919
Speaker 1: one job, though, where robots are already taking the wheel,

9
00:00:20,079 --> 00:00:21,320
Speaker 1: and that is driving.

10
00:00:21,640 --> 00:00:23,360
Speaker 3: In fact, it's one of the most common jobs in

11
00:00:23,400 --> 00:00:26,239
Speaker 3: America for young men without college degrees, and over the

12
00:00:26,280 --> 00:00:29,520
Speaker 3: course of two episodes, Search Engine tackles both the promise

13
00:00:29,600 --> 00:00:31,600
Speaker 3: and the peril this growing technology.

14
00:00:31,840 --> 00:00:34,320
Speaker 1: In part one, Are You a Good Driver? The Search

15
00:00:34,360 --> 00:00:36,920
Speaker 1: Engine team tells the story of how a small secret

16
00:00:36,960 --> 00:00:39,879
Speaker 1: team at Google spent fifteen years teaching a computer to

17
00:00:39,960 --> 00:00:43,000
Speaker 1: drive from a failed robot in the Mojave Desert to

18
00:00:43,080 --> 00:00:45,480
Speaker 1: a vehicle that might actually be the safest on the road.

19
00:00:45,800 --> 00:00:49,680
Speaker 1: This episode tracks the engineering breakthroughs the nearer catastrophes, and

20
00:00:49,760 --> 00:00:52,720
Speaker 1: takes a skeptical look at the safety data behind Weaymo's

21
00:00:52,760 --> 00:00:55,520
Speaker 1: claim that its cars are ninety percent safer than human

22
00:00:55,640 --> 00:00:57,160
Speaker 1: drivers in serious crashes.

23
00:00:57,560 --> 00:01:00,200
Speaker 3: All of it boils down to one big question. All

24
00:01:00,200 --> 00:01:03,000
Speaker 3: the robots actually say for drivers than we are. Enjoy

25
00:01:03,040 --> 00:01:05,680
Speaker 3: this presentation of Search Engine, and be sure to catch

26
00:01:05,760 --> 00:01:09,280
Speaker 3: part two, titled The Trial of the Driverless Car, available

27
00:01:09,280 --> 00:01:33,480
Speaker 3: wherever you get your podcasts.

28
00:01:37,800 --> 00:01:40,119
Speaker 2: Before we start the story today, I want to ask

29
00:01:40,160 --> 00:01:43,600
Speaker 2: you to imagine a different version of your life. You're you,

30
00:01:44,080 --> 00:01:47,200
Speaker 2: but it's almost two hundred years ago, and unfortunately and

31
00:01:47,280 --> 00:01:51,360
Speaker 2: our hypothetical, it's Monday morning. It's Monday morning, and it's

32
00:01:51,480 --> 00:01:55,440
Speaker 2: very early pre dawn. You wake up to this really

33
00:01:55,480 --> 00:01:59,200
Speaker 2: hard wrapping at your window. That's the knocker Upper here

34
00:01:59,240 --> 00:02:01,360
Speaker 2: to get you up for where we're in the eighteen

35
00:02:01,440 --> 00:02:05,040
Speaker 2: hundreds before the invention of the adjustable alarm clock. The

36
00:02:05,080 --> 00:02:08,840
Speaker 2: knocker Upper is a job. The knocker Upper walks the

37
00:02:08,840 --> 00:02:11,679
Speaker 2: neighborhood with a long stick and taps it on the

38
00:02:11,720 --> 00:02:14,240
Speaker 2: windows of people's houses early in the morning to wake

39
00:02:14,280 --> 00:02:17,440
Speaker 2: them up for work. Who wakes up the locker rupper

40
00:02:17,440 --> 00:02:20,720
Speaker 2: for work? Nobody knows. But this is a job, a

41
00:02:20,840 --> 00:02:25,280
Speaker 2: job that'll actually exist for another century. Outside the gas

42
00:02:25,280 --> 00:02:28,000
Speaker 2: street lamps are still burning. The lamplighter lit them the

43
00:02:28,080 --> 00:02:31,359
Speaker 2: night before. He's supposed to come at dawn to extinguish them,

44
00:02:31,440 --> 00:02:34,320
Speaker 2: but it's so early that he has it yet. Your

45
00:02:34,400 --> 00:02:36,399
Speaker 2: lamplighter is one of those neighbors. You have a deep

46
00:02:36,440 --> 00:02:40,360
Speaker 2: fondness for a fixture. Every day you watch him make

47
00:02:40,400 --> 00:02:42,919
Speaker 2: the rounds at dusk with his ladder and his light.

48
00:02:44,639 --> 00:02:48,840
Speaker 2: You yourself are a driver. Professional driver two hundred years

49
00:02:48,840 --> 00:02:52,000
Speaker 2: ago is also a job. You're a person who sits

50
00:02:52,000 --> 00:02:54,160
Speaker 2: on a coach and holds the reins of a horse.

51
00:02:54,880 --> 00:02:57,760
Speaker 2: You take passengers where they want to go. You start

52
00:02:57,760 --> 00:03:05,840
Speaker 2: your workday. Okay, hypothetical. Over two of those jobs are

53
00:03:05,880 --> 00:03:09,360
Speaker 2: obviously so long disappeared that most people don't know about them.

54
00:03:09,760 --> 00:03:13,040
Speaker 2: The knocker upper is your iPhone alarm. The lamplighter is

55
00:03:13,040 --> 00:03:17,400
Speaker 2: the electric street light. The third one driver has persisted

56
00:03:18,080 --> 00:03:20,840
Speaker 2: as a job for some, as a routine human task

57
00:03:20,960 --> 00:03:24,600
Speaker 2: for nearly everyone else. This is a story about whether

58
00:03:24,639 --> 00:03:27,880
Speaker 2: that's about to change. It's about how the word driver,

59
00:03:28,240 --> 00:03:31,040
Speaker 2: which right now makes me picture a human, could soon

60
00:03:31,120 --> 00:03:34,359
Speaker 2: transform to refer to a machine, the same way the

61
00:03:34,400 --> 00:03:39,840
Speaker 2: words dishwasher, printer, and computer all did. I've thought about this,

62
00:03:40,080 --> 00:03:42,800
Speaker 2: maybe too much, in the year I've been working on

63
00:03:42,840 --> 00:03:46,600
Speaker 2: this story. In conversations constantly, I'd asked the humans I

64
00:03:46,680 --> 00:03:50,920
Speaker 2: meant the same question, Are you a good driver? Are

65
00:03:51,040 --> 00:03:52,760
Speaker 2: you do you consider yourself a good driver.

66
00:03:54,040 --> 00:03:59,440
Speaker 4: I do within limits. I think I'm a good driver

67
00:03:59,560 --> 00:04:02,680
Speaker 4: because I understand the limitations of my driving.

68
00:04:03,320 --> 00:04:06,680
Speaker 2: This is Alex Davies. He wrote an excellent book called Driven,

69
00:04:06,880 --> 00:04:10,160
Speaker 2: the Race to create the Autonomous Car. Alex, like me,

70
00:04:10,280 --> 00:04:13,720
Speaker 2: thinks a lot about human driving about his own personal limitations.

71
00:04:14,240 --> 00:04:15,280
Speaker 2: What are the limitations?

72
00:04:15,840 --> 00:04:18,880
Speaker 4: The limitations are that I can't always pay attention to

73
00:04:18,920 --> 00:04:23,240
Speaker 4: everything that I get tired. I've been trying really hard

74
00:04:23,839 --> 00:04:27,800
Speaker 4: to be calmer in the road. My husband and I

75
00:04:27,839 --> 00:04:31,920
Speaker 4: are expecting our first baby this fall. Congratulations, thank you,

76
00:04:32,240 --> 00:04:35,000
Speaker 4: and I thought that, along with reading all the baby books,

77
00:04:35,040 --> 00:04:37,679
Speaker 4: a good project to work on is just be calmer

78
00:04:37,720 --> 00:04:38,680
Speaker 4: in the car.

79
00:04:39,560 --> 00:04:42,200
Speaker 2: A very good resolution, because, of course, for most of us,

80
00:04:42,440 --> 00:04:46,320
Speaker 2: driving is the riskiest behavior we routinely engage in. In fact,

81
00:04:46,400 --> 00:04:49,359
Speaker 2: even Alex, despite his good intentions, would actually get in

82
00:04:49,400 --> 00:04:51,559
Speaker 2: a car accident just a few months after we first spoke.

83
00:04:52,440 --> 00:04:54,400
Speaker 2: He was okay, it was the car that was totaled.

84
00:04:56,040 --> 00:04:58,719
Speaker 2: Safety is the entire pitch for the driver of this car,

85
00:04:58,880 --> 00:05:01,800
Speaker 2: which is really a car and by a computer driver.

86
00:05:01,880 --> 00:05:05,159
Speaker 2: LESE cars don't get drunk, tired, or distracted. They never

87
00:05:05,279 --> 00:05:09,160
Speaker 2: text or feel road rage, and these drivers cars, they

88
00:05:09,200 --> 00:05:13,719
Speaker 2: aren't the future. They're actually already here. But it's funny.

89
00:05:13,720 --> 00:05:15,479
Speaker 2: If you just don't happen to live in a place

90
00:05:15,520 --> 00:05:18,320
Speaker 2: that already has them, it's easy to not see how

91
00:05:18,360 --> 00:05:22,719
Speaker 2: fast things are changing. Robo taxis like Waimo are operating

92
00:05:22,720 --> 00:05:25,880
Speaker 2: in ten American cities, providing millions of rides to Americans.

93
00:05:26,880 --> 00:05:29,440
Speaker 2: In China, the rollout is happening even more widely. They're

94
00:05:29,440 --> 00:05:32,360
Speaker 2: in twice as many cities. But here, if you live

95
00:05:32,400 --> 00:05:35,240
Speaker 2: in a place like San Francisco or Austin, today, a

96
00:05:35,320 --> 00:05:38,680
Speaker 2: driver's car is about as exotic as an uber. A

97
00:05:38,720 --> 00:05:41,720
Speaker 2: passenger in those cities opens up their phone and decides

98
00:05:41,760 --> 00:05:45,159
Speaker 2: who should drive them, a human driver or a robot driver.

99
00:05:48,040 --> 00:05:51,360
Speaker 2: How that happened is a story, a story we are

100
00:05:51,360 --> 00:05:54,599
Speaker 2: living through right now, whose ending promise is to totally

101
00:05:54,600 --> 00:05:57,400
Speaker 2: reshape the places we live. And today we're going to

102
00:05:57,400 --> 00:06:03,760
Speaker 2: tell you how we got here. In chapters, Chapter one

103
00:06:04,360 --> 00:06:09,080
Speaker 2: dreams without drivers. So it turns out this dream that

104
00:06:09,160 --> 00:06:12,120
Speaker 2: inventors have had to replace the human driver with some

105
00:06:12,200 --> 00:06:14,760
Speaker 2: kind of machine. That dream is about as old as

106
00:06:14,760 --> 00:06:15,600
Speaker 2: the Lamplighters.

107
00:06:16,480 --> 00:06:20,080
Speaker 4: People have been thinking about a self driving car for

108
00:06:21,640 --> 00:06:23,880
Speaker 4: so it's about as long as there's been a human

109
00:06:24,080 --> 00:06:28,240
Speaker 4: driven car. Why, there's this funny thing you lose when

110
00:06:28,279 --> 00:06:30,760
Speaker 4: you move from the horse to the human driven car,

111
00:06:30,800 --> 00:06:34,640
Speaker 4: which is that in a horse drawn carriage, the horse

112
00:06:34,720 --> 00:06:36,280
Speaker 4: is not just going to run off a cliff. If

113
00:06:36,320 --> 00:06:40,800
Speaker 4: you let go of the reins, you lose sentience in

114
00:06:40,839 --> 00:06:41,520
Speaker 4: your vehicle.

115
00:06:45,640 --> 00:06:50,320
Speaker 2: When automobiles first arrived, these powerful and nonsensient cars, there's

116
00:06:50,320 --> 00:06:52,960
Speaker 2: actually a passionate fight to keep them off the streets.

117
00:06:53,760 --> 00:06:56,800
Speaker 2: It was the eighteen hundreds, and people feared these new things,

118
00:06:57,640 --> 00:07:00,960
Speaker 2: the steam powered vehicles thundering down the roads that soon

119
00:07:00,960 --> 00:07:04,599
Speaker 2: evolved into gas powered vehicles, also thundering down the roads.

120
00:07:05,880 --> 00:07:09,039
Speaker 2: The fear was partly about jobs. These vehicles were seen

121
00:07:09,080 --> 00:07:11,800
Speaker 2: as a huge threat to a whole network of working

122
00:07:11,880 --> 00:07:16,280
Speaker 2: class jobs. Horse breeders and horse farriers, horse feed suppliers,

123
00:07:16,400 --> 00:07:20,640
Speaker 2: horse manure haulers, horse carriage manufacturers. Not to mention the

124
00:07:20,680 --> 00:07:24,680
Speaker 2: teamsters Teamsters today the word makes me think of the

125
00:07:24,680 --> 00:07:28,160
Speaker 2: Teamsters union, But originally the teamsters were the workers who

126
00:07:28,240 --> 00:07:31,800
Speaker 2: drove teams of horses. Teamsters were like truckers before we

127
00:07:31,800 --> 00:07:36,920
Speaker 2: had trucks. Cars seemed to imperil all these horse related jobs.

128
00:07:37,360 --> 00:07:40,040
Speaker 2: And even if you weren't worried about these workers, the

129
00:07:40,080 --> 00:07:43,760
Speaker 2: cars were also less safe. Some anti car activists battled

130
00:07:43,760 --> 00:07:47,920
Speaker 2: to stop or slow the new technology, mainly with regulations.

131
00:07:48,800 --> 00:07:51,120
Speaker 2: There were red flag laws, which said if you had

132
00:07:51,120 --> 00:07:53,760
Speaker 2: an automobile, you had to hire a person to walk

133
00:07:53,800 --> 00:07:56,400
Speaker 2: in front of it, waving a giant red flag to

134
00:07:56,480 --> 00:08:01,720
Speaker 2: warn people. In Pennsylvania, a law was prepared requiring horseless

135
00:08:01,760 --> 00:08:06,000
Speaker 2: carriage drivers who encountered livestock to stop, disassemble their car,

136
00:08:06,520 --> 00:08:10,160
Speaker 2: and hide the parts behind the bushes. The governor vetoed it.

137
00:08:12,920 --> 00:08:16,600
Speaker 2: But to think about these crazy anti car activists is

138
00:08:16,720 --> 00:08:20,320
Speaker 2: that directionally they were right. Those cars did initially wipe

139
00:08:20,320 --> 00:08:22,480
Speaker 2: out a lot of jobs, even if they created more,

140
00:08:23,040 --> 00:08:26,040
Speaker 2: and cars were very unsafe. The cities that threw their

141
00:08:26,080 --> 00:08:30,240
Speaker 2: doors open to cars without regulation were rewarded with astonishing

142
00:08:30,280 --> 00:08:34,160
Speaker 2: death rates. Detroit let drivers pretty much run wild. In

143
00:08:34,200 --> 00:08:38,000
Speaker 2: the early nineteen hundreds, deaths accumulated in a Detroit without

144
00:08:38,080 --> 00:08:41,760
Speaker 2: drivers licenses, stop lights, or turn signals. Many of those

145
00:08:41,800 --> 00:08:46,040
Speaker 2: deaths were children. It took decades for society to mostly

146
00:08:46,120 --> 00:08:49,000
Speaker 2: learn to live with cars. The rest of the story

147
00:08:49,120 --> 00:08:52,520
Speaker 2: is just the world you grew up in. We invented laws, licenses,

148
00:08:52,600 --> 00:08:55,760
Speaker 2: drivers ed, We learned to better design roads. We invented

149
00:08:55,760 --> 00:08:59,280
Speaker 2: the highway, the seatbelt, the airbag. All those things made

150
00:08:59,320 --> 00:09:04,800
Speaker 2: driving less debts, although the smartphone reverse some of that progress. Nationally, Today,

151
00:09:05,000 --> 00:09:07,559
Speaker 2: deaths from cars are about as common in America as

152
00:09:07,600 --> 00:09:10,720
Speaker 2: deaths from guns or opioids, about one in one hundred.

153
00:09:11,840 --> 00:09:14,120
Speaker 2: It'll probably happen to someone you know in your life,

154
00:09:14,720 --> 00:09:18,280
Speaker 2: maybe several someone's. Whether or not you see that as

155
00:09:18,320 --> 00:09:22,240
Speaker 2: an urgent problem to solve depends on you. But as

156
00:09:22,360 --> 00:09:24,440
Speaker 2: long as there have been cars, there have been people

157
00:09:24,480 --> 00:09:27,040
Speaker 2: who wanted to truly solve what's left of the safety

158
00:09:27,040 --> 00:09:30,160
Speaker 2: problem the best way we knew how. They wanted to

159
00:09:30,160 --> 00:09:33,439
Speaker 2: make the car more like the horse it replaced, make

160
00:09:33,480 --> 00:09:34,520
Speaker 2: the car more sentient.

161
00:09:36,000 --> 00:09:39,800
Speaker 4: So that thought is there early and like early visions

162
00:09:39,840 --> 00:09:44,480
Speaker 4: have hit include, oh well, we'll have radio controlled cars,

163
00:09:44,920 --> 00:09:49,000
Speaker 4: because they had radios at the time. There's a real

164
00:09:49,120 --> 00:09:52,800
Speaker 4: effort at one point to build magnets under the road,

165
00:09:53,840 --> 00:09:57,440
Speaker 4: and at each stage what a self driving car can

166
00:09:57,520 --> 00:10:01,400
Speaker 4: be is dictated by the technology that's available at the time.

167
00:10:01,880 --> 00:10:05,439
Speaker 4: For the most part, no one's thinking that much about

168
00:10:05,880 --> 00:10:09,680
Speaker 4: a vehicle that thinks for itself. They're just thinking about

169
00:10:09,720 --> 00:10:12,640
Speaker 4: a vehicle that the person in it doesn't have to drive.

170
00:10:13,880 --> 00:10:18,079
Speaker 2: Many different attempts, many different failures, as many wonders as

171
00:10:18,080 --> 00:10:22,120
Speaker 2: we invented, we could not approach nature's most majestic creation,

172
00:10:23,200 --> 00:10:26,280
Speaker 2: a horse's brain, at least not until the turn of

173
00:10:26,280 --> 00:10:29,720
Speaker 2: the millennium.

174
00:10:30,880 --> 00:10:33,880
Speaker 5: Are to.

175
00:10:38,760 --> 00:10:41,360
Speaker 6: Deep within the Department of Defense, there's a little known

176
00:10:41,400 --> 00:10:44,400
Speaker 6: military agency that has created some of the most innovative

177
00:10:44,440 --> 00:10:49,000
Speaker 6: technology of the twentieth century. This is the story of Dark.

178
00:10:49,040 --> 00:10:52,040
Speaker 2: Chapter two, DARPA's Million Dollar Prize.

179
00:10:52,440 --> 00:10:56,880
Speaker 6: DARPA's current goal is to develop autonomous military vehicles machines

180
00:10:56,920 --> 00:10:59,559
Speaker 6: that can operate on their own without drivers.

181
00:11:00,040 --> 00:11:01,600
Speaker 4: Darpe has always been intrigued with him.

182
00:11:01,720 --> 00:11:04,600
Speaker 2: This is from a documentary called The Million Dollar Challenge.

183
00:11:04,720 --> 00:11:07,440
Speaker 2: Honestly less a doc more an ad for DARPA, the

184
00:11:07,480 --> 00:11:11,160
Speaker 2: Pentagon's research arm. DARPA's mission is to try to keep

185
00:11:11,160 --> 00:11:14,920
Speaker 2: American technology one generation ahead of everybody else. It doesn't

186
00:11:14,920 --> 00:11:18,440
Speaker 2: always work, but DARPA has invented or funded a lot

187
00:11:18,840 --> 00:11:22,280
Speaker 2: GPS and the M sixteen, the early Internet, and the

188
00:11:22,280 --> 00:11:25,640
Speaker 2: Predator drone. In two thousand and two, DARPA decided to

189
00:11:25,640 --> 00:11:28,720
Speaker 2: pursue the driver's car in a very unusual way.

190
00:11:29,559 --> 00:11:31,680
Speaker 4: The director of DARPA at the time, a guy named

191
00:11:31,679 --> 00:11:35,400
Speaker 4: Tony Tether, who had been a door to door salesman

192
00:11:35,679 --> 00:11:39,120
Speaker 4: in his use definitely has that flare in that way

193
00:11:39,160 --> 00:11:44,800
Speaker 4: of thinking, says, let's have a contest. Let's see who

194
00:11:44,800 --> 00:11:49,000
Speaker 4: can put all of these ingredients that we've developed together

195
00:11:49,320 --> 00:11:53,320
Speaker 4: into a proper self driving car. His original idea is

196
00:11:53,440 --> 00:11:56,760
Speaker 4: we'll drive him down the Las Vegas Strip that's almost

197
00:11:56,800 --> 00:11:58,880
Speaker 4: immediately next because it's insane.

198
00:11:59,080 --> 00:12:03,000
Speaker 7: Oh right, you would have to like literally gridlock a

199
00:12:03,160 --> 00:12:07,480
Speaker 7: huge American city so people could put robot cars on

200
00:12:07,520 --> 00:12:08,679
Speaker 7: it exactly.

201
00:12:09,000 --> 00:12:10,640
Speaker 4: So he says, Okay, do you know what, We'll do

202
00:12:10,679 --> 00:12:12,720
Speaker 4: it in the desert. We'll do it in the desert

203
00:12:12,800 --> 00:12:17,520
Speaker 4: outside Las Vegas, and anyone who wants to can make

204
00:12:17,559 --> 00:12:20,320
Speaker 4: a team build a self driving car, bring it to

205
00:12:20,400 --> 00:12:22,079
Speaker 4: the desert, and we'll race them.

206
00:12:22,600 --> 00:12:25,800
Speaker 2: The driver that DARPA wanted to replace was the American soldier.

207
00:12:26,400 --> 00:12:29,240
Speaker 2: DARPA wanted a vehicle that could drive itself down roads

208
00:12:29,280 --> 00:12:32,680
Speaker 2: that might be filled with hidden explosive devices. So, in

209
00:12:32,720 --> 00:12:35,200
Speaker 2: this moment at the tail end of the dot com boom,

210
00:12:35,520 --> 00:12:38,559
Speaker 2: darpest trying to inspire tech to build something besides another

211
00:12:38,600 --> 00:12:42,600
Speaker 2: website darpest. Tony Tether announces that the prize for whoever

212
00:12:42,640 --> 00:12:45,640
Speaker 2: can win its Grand Challenge will be one million dollars.

213
00:12:46,520 --> 00:12:50,320
Speaker 4: The rules for very open there were little rules like

214
00:12:50,360 --> 00:12:53,600
Speaker 4: you couldn't have two vehicles communicating with one another, but

215
00:12:53,720 --> 00:12:56,320
Speaker 4: you could build any kind of vehicle you wanted, could

216
00:12:56,320 --> 00:12:58,040
Speaker 4: have six wheels. It could be a truck, it could

217
00:12:58,080 --> 00:13:01,640
Speaker 4: be a motorcycle, could be a trice. It just couldn't

218
00:13:01,679 --> 00:13:04,640
Speaker 4: attack other vehicles. That was rolled out early on.

219
00:13:04,920 --> 00:13:06,800
Speaker 2: Oh, was that a concern that people would just like

220
00:13:07,360 --> 00:13:09,880
Speaker 2: sort of battlebot the thing you're auto's vehicle would have

221
00:13:09,960 --> 00:13:11,880
Speaker 2: like a little shredder that would take out somebody else's.

222
00:13:12,800 --> 00:13:15,520
Speaker 4: Someone asked in the first Q and A at this

223
00:13:15,720 --> 00:13:18,520
Speaker 4: like they said, can we attack other vehicles? And they

224
00:13:18,520 --> 00:13:21,760
Speaker 4: said no. And it's funny you bring up BattleBots because

225
00:13:21,880 --> 00:13:26,319
Speaker 4: a lot of teams who entered this had BattleBots history interesting.

226
00:13:26,480 --> 00:13:30,920
Speaker 4: They were used to building robots for interesting purposes, and

227
00:13:30,960 --> 00:13:33,280
Speaker 4: when they caught wind of this, they said, we can

228
00:13:33,360 --> 00:13:36,360
Speaker 4: do this. We can scrap together some money and this

229
00:13:36,400 --> 00:13:37,320
Speaker 4: will just be fun.

230
00:13:41,120 --> 00:13:43,920
Speaker 2: I'm going to tell you what happened in this robot

231
00:13:44,000 --> 00:13:46,760
Speaker 2: race in the desert, not because I care so much

232
00:13:46,760 --> 00:13:49,880
Speaker 2: about these early robot vehicles, but because I care a

233
00:13:49,920 --> 00:13:53,240
Speaker 2: lot about the engineers who were making them. These would

234
00:13:53,280 --> 00:13:55,000
Speaker 2: be the people who would later go on to lead

235
00:13:55,000 --> 00:13:58,640
Speaker 2: development for the billion dollar companies creating today's drive those cars.

236
00:13:59,840 --> 00:14:02,840
Speaker 2: The people had very different views about how to get

237
00:14:02,840 --> 00:14:06,080
Speaker 2: that technology ready, different values when it came to things

238
00:14:06,160 --> 00:14:10,920
Speaker 2: like the acceptability of risking human life, abstract differences that

239
00:14:10,920 --> 00:14:14,000
Speaker 2: would become very concrete later on, to the point where

240
00:14:14,000 --> 00:14:19,440
Speaker 2: people would be charged with federal crimes. That's the future.

241
00:14:20,040 --> 00:14:22,080
Speaker 2: But listening to this part of the story, what I

242
00:14:22,240 --> 00:14:25,080
Speaker 2: listen for is how much of it can you detect already?

243
00:14:25,320 --> 00:14:28,960
Speaker 2: How much of the differences already present. The first engineer

244
00:14:28,960 --> 00:14:30,480
Speaker 2: I want you to pay attention to is a man

245
00:14:30,560 --> 00:14:35,360
Speaker 2: named Chris Armson, and way back in two thousand and two,

246
00:14:35,680 --> 00:14:37,680
Speaker 2: how did you end up being part of the Dark

247
00:14:37,680 --> 00:14:38,600
Speaker 2: Program challenge?

248
00:14:39,960 --> 00:14:40,880
Speaker 8: It sounded like fun.

249
00:14:43,080 --> 00:14:45,800
Speaker 2: Chris these days, the CEO of a large tech company

250
00:14:46,200 --> 00:14:50,200
Speaker 2: back then, a PhD student at Carnegie Mellon University. When

251
00:14:50,200 --> 00:14:52,200
Speaker 2: he first got recruited for the race, he was out

252
00:14:52,200 --> 00:14:55,560
Speaker 2: in the field observing a robot as it crept across

253
00:14:55,560 --> 00:14:58,720
Speaker 2: the Autacama Desert training for its future deployment on the

254
00:14:58,720 --> 00:14:59,560
Speaker 2: surface of Mars.

255
00:15:00,040 --> 00:15:03,200
Speaker 8: H advisor came down and was really excited about this

256
00:15:03,360 --> 00:15:06,240
Speaker 8: Darker Grind challenge thing, and the idea that you'd have

257
00:15:06,320 --> 00:15:09,880
Speaker 8: a robot run across the desert at fifty miles an

258
00:15:09,880 --> 00:15:14,560
Speaker 8: hour just sounded exciting having spent the last couple of

259
00:15:14,560 --> 00:15:17,960
Speaker 8: weeks walking behind a robot at very low speed.

260
00:15:20,080 --> 00:15:22,920
Speaker 2: So Chris would join Carnegie Mellon's Red Team and help

261
00:15:22,960 --> 00:15:25,920
Speaker 2: build a car called Sandstorm, a bright red humvey with

262
00:15:25,960 --> 00:15:29,640
Speaker 2: the top lopped off, a plethora of futuristic sensors mounted

263
00:15:29,640 --> 00:15:32,560
Speaker 2: to it like scanners a crackpot would use to search

264
00:15:32,560 --> 00:15:35,960
Speaker 2: for aliens. You can see Chris back in that documentary.

265
00:15:36,280 --> 00:15:38,200
Speaker 2: He explains to the filmmaker at the time that the

266
00:15:38,200 --> 00:15:40,560
Speaker 2: hard part, of course, isn't the vehicle, it's the driver.

267
00:15:41,280 --> 00:15:43,480
Speaker 2: How do you even begin to teach a computer to

268
00:15:43,480 --> 00:15:45,320
Speaker 2: operate a hum vy at all? How does a.

269
00:15:45,320 --> 00:15:46,720
Speaker 4: Computer make the steering wheel turn?

270
00:15:46,760 --> 00:15:48,920
Speaker 5: How does a computer change the.

271
00:15:48,640 --> 00:15:50,720
Speaker 4: Pressure on the break and the throttle? Those are the

272
00:15:50,760 --> 00:15:52,600
Speaker 4: issues that we're fighting through right now.

273
00:15:53,560 --> 00:15:56,880
Speaker 2: The answer Sandstorm represented the best entry from the contest's

274
00:15:56,920 --> 00:16:00,480
Speaker 2: traditional academic crowd, but there's a different crowd there too.

275
00:16:00,960 --> 00:16:04,600
Speaker 2: Represented best by a man named Anthony Lewandowski. Can you

276
00:16:04,600 --> 00:16:06,120
Speaker 2: tell me about Anthony Lewandowski?

277
00:16:07,000 --> 00:16:18,600
Speaker 4: Anthony Lewandowski. Where to begin? So Anthony is like an entrepreneur.

278
00:16:19,040 --> 00:16:24,080
Speaker 4: He's a really charming guy. He's six foot six, He's

279
00:16:24,280 --> 00:16:28,480
Speaker 4: gangly as all get down. He grew up mostly in

280
00:16:28,560 --> 00:16:32,360
Speaker 4: Belgium because his mom was working for the EU. For

281
00:16:32,520 --> 00:16:36,240
Speaker 4: high school, he moved to Marin to live with his dad.

282
00:16:37,280 --> 00:16:38,760
Speaker 4: And he's a hustler.

283
00:16:39,400 --> 00:16:41,040
Speaker 9: My name is Anthony Lewandowski.

284
00:16:42,400 --> 00:16:44,320
Speaker 4: I was a grad student at Berkeley.

285
00:16:44,480 --> 00:16:47,720
Speaker 9: Instead of continuing on to finish my PhD, I decided

286
00:16:47,760 --> 00:16:49,960
Speaker 9: it was much better to do the Grand Challenge.

287
00:16:50,440 --> 00:16:52,960
Speaker 2: We asked Anthony for an interview. He didn't respond, but

288
00:16:53,040 --> 00:16:55,600
Speaker 2: here he is in the footage from back then. Anthony

289
00:16:55,640 --> 00:16:58,640
Speaker 2: did not have the engineering experience or resources of a

290
00:16:58,640 --> 00:17:01,880
Speaker 2: team like Carnegie Mellons Red Team. So you tried something

291
00:17:01,960 --> 00:17:04,879
Speaker 2: very different, a vehicle that had almost no chance of

292
00:17:04,920 --> 00:17:07,840
Speaker 2: winning the race, but which was also perfectly designed to

293
00:17:07,880 --> 00:17:10,359
Speaker 2: stand out to get him a lot of attention, maybe

294
00:17:10,359 --> 00:17:14,840
Speaker 2: a job. The race's only self driving motorcycle, it was

295
00:17:14,920 --> 00:17:18,240
Speaker 2: named ghost Rider, a stubby little thing covered in stickers

296
00:17:18,440 --> 00:17:20,639
Speaker 2: with then inten on the back and cameras on the front.

297
00:17:21,840 --> 00:17:24,639
Speaker 9: There's a steering actuator on the top here, which allows

298
00:17:24,680 --> 00:17:29,040
Speaker 9: us to modify the steering angle. So basically, if you're driving,

299
00:17:29,080 --> 00:17:31,480
Speaker 9: you start to follow the left, you steer left. That

300
00:17:31,560 --> 00:17:33,159
Speaker 9: makes you turn the left, and then you get the

301
00:17:33,160 --> 00:17:35,120
Speaker 9: tripleal acceleration to put you back up to the right.

302
00:17:35,720 --> 00:17:38,520
Speaker 9: And you're monitoring that in real time and making small adjustments,

303
00:17:38,520 --> 00:17:39,720
Speaker 9: and you stay bounced.

304
00:17:42,320 --> 00:17:45,919
Speaker 10: Stroll Blight is on. The command from the tower is

305
00:17:45,960 --> 00:17:49,399
Speaker 10: to move, ladies and gentlemen Sandstorm.

306
00:17:50,240 --> 00:17:52,440
Speaker 2: The race happens on a Saturday in March of two

307
00:17:52,440 --> 00:17:52,840
Speaker 2: thousand and.

308
00:17:52,840 --> 00:17:59,879
Speaker 10: Four, autonomous vehicle traversing the desert with the goal of

309
00:18:00,119 --> 00:18:07,680
Speaker 10: keeping our young military personnel out of harm's way.

310
00:18:08,600 --> 00:18:11,840
Speaker 2: Oh yeah, what happens the first time they try to

311
00:18:11,840 --> 00:18:12,560
Speaker 2: do this competition?

312
00:18:14,000 --> 00:18:16,920
Speaker 4: The two thousand and four Grand Challenge is an utter

313
00:18:17,760 --> 00:18:19,520
Speaker 4: hysterical disaster.

314
00:18:22,280 --> 00:18:27,320
Speaker 2: Disaster Number one ghost Rider the motorcycle Anthony Lewandowski forgot

315
00:18:27,320 --> 00:18:30,199
Speaker 2: to flip on the switch for the stabilization system. The

316
00:18:30,240 --> 00:18:34,560
Speaker 2: bike immediately topples ghost Rider down.

317
00:18:34,640 --> 00:18:43,119
Speaker 4: Anthony good effort, and then every vehicle after it fails miserably.

318
00:18:43,480 --> 00:18:46,800
Speaker 4: Like one vehicle drives up onto a burm flips off.

319
00:18:47,160 --> 00:18:50,920
Speaker 4: One vehicle, drives straight out, does an inexplicable U turn

320
00:18:51,440 --> 00:18:54,040
Speaker 4: and just drives back to the starting line. And the

321
00:18:54,119 --> 00:18:57,000
Speaker 4: rules are that once your vehicle starts, you can't do anything.

322
00:18:57,800 --> 00:19:01,080
Speaker 2: Even Sandstorm got stuck on a burm. Chris Urmson just

323
00:19:01,240 --> 00:19:03,200
Speaker 2: standing there, unable to help his robot.

324
00:19:03,720 --> 00:19:06,000
Speaker 8: Poor thing was trying to get going, but its wheels

325
00:19:06,000 --> 00:19:10,000
Speaker 8: were just spinning on the gravel and tried so hard

326
00:19:10,080 --> 00:19:12,200
Speaker 8: that it actually melted the rubber of the tires.

327
00:19:12,200 --> 00:19:13,480
Speaker 5: And so there's this plums of.

328
00:19:13,440 --> 00:19:15,800
Speaker 8: Black squoke before they killed it.

329
00:19:16,600 --> 00:19:20,320
Speaker 2: For the roboticists, this was obviously very disappointing. Chris Urmson

330
00:19:20,359 --> 00:19:22,919
Speaker 2: compared it to an Olympic marathon where the best runner

331
00:19:22,960 --> 00:19:26,200
Speaker 2: only makes it two of the twenty six miles. What

332
00:19:26,240 --> 00:19:29,240
Speaker 2: this contest had done, though, was it had flushed all

333
00:19:29,280 --> 00:19:32,040
Speaker 2: these inventors out. It had jumpstarted the scene that would

334
00:19:32,040 --> 00:19:35,280
Speaker 2: develop this technology. One of the most important people there

335
00:19:35,320 --> 00:19:39,240
Speaker 2: that day, actually just watching, was someone I haven't mentioned yet,

336
00:19:39,480 --> 00:19:42,159
Speaker 2: a legendary roboticist named Sebastian Thrun.

337
00:19:43,080 --> 00:19:46,600
Speaker 4: Sebastian Thrun, he was at the first Grand Challenge. He

338
00:19:46,640 --> 00:19:50,439
Speaker 4: didn't bring a team, he wasn't participating. DARPA wanted to

339
00:19:50,440 --> 00:19:52,959
Speaker 4: show off some other projects they'd been funding, including one

340
00:19:53,000 --> 00:19:55,479
Speaker 4: of his robots. So he brings the robot and so

341
00:19:55,560 --> 00:20:00,159
Speaker 4: he's there and he watches this disaster or anything's can

342
00:20:00,200 --> 00:20:01,040
Speaker 4: do better for mess.

343
00:20:03,440 --> 00:20:06,160
Speaker 11: I looked at the very first iteration of this quan challenge,

344
00:20:06,200 --> 00:20:08,119
Speaker 11: but it didn't participate. It was a spectator.

345
00:20:08,960 --> 00:20:11,119
Speaker 2: This, of course is Sebastian Thrun. He grew up in

346
00:20:11,160 --> 00:20:14,160
Speaker 2: West Germany, moved to the US Toddic Carnegie Mellon before

347
00:20:14,160 --> 00:20:18,720
Speaker 2: moving to Stanford. Watching that day, he saw this fundamental error.

348
00:20:18,760 --> 00:20:20,560
Speaker 2: He believed all the entrance had made.

349
00:20:21,200 --> 00:20:24,560
Speaker 11: I saw that all the teams treated this like a

350
00:20:24,560 --> 00:20:26,919
Speaker 11: hardware problem. They looked at this and say, we have

351
00:20:27,000 --> 00:20:31,360
Speaker 11: to build a bigger wheels and bigger chassis and so on.

352
00:20:32,480 --> 00:20:34,680
Speaker 11: And I looked at this and said, about wait a minute.

353
00:20:34,960 --> 00:20:37,400
Speaker 11: The challenge really is to build a self driving car.

354
00:20:37,640 --> 00:20:40,440
Speaker 11: They can drive for the desert. I can get a

355
00:20:40,480 --> 00:20:43,080
Speaker 11: rental car. They can do it just fine, provided as

356
00:20:43,080 --> 00:20:45,719
Speaker 11: a person insight and the challenges we need to take

357
00:20:45,760 --> 00:20:47,560
Speaker 11: the person out of the driver's seat and replace it

358
00:20:47,640 --> 00:20:51,000
Speaker 11: by computer. That is not a problem with bigger tires.

359
00:20:51,080 --> 00:20:55,480
Speaker 11: That's actually be a software problem.

360
00:20:55,640 --> 00:21:00,000
Speaker 2: Sebastian Thrun had a dual background robotics and artificial intelligence,

361
00:21:00,480 --> 00:21:03,800
Speaker 2: which probably explains his focus here on the robot driver's mind.

362
00:21:04,240 --> 00:21:07,600
Speaker 2: He was thinking about something else too. The military wanted

363
00:21:07,600 --> 00:21:10,400
Speaker 2: this tech to replace a relatively small number of drivers

364
00:21:10,400 --> 00:21:14,400
Speaker 2: in its war zones, but Sebastian was already imagining something bigger.

365
00:21:15,119 --> 00:21:18,440
Speaker 2: What would happen to traffic deaths worldwide? If one day

366
00:21:18,680 --> 00:21:20,600
Speaker 2: everyone had access to a driverless car.

367
00:21:21,040 --> 00:21:23,399
Speaker 11: I had experiences of losing people in my life to

368
00:21:23,480 --> 00:21:26,359
Speaker 11: traffic accidents, and I felt we lost over the million

369
00:21:26,359 --> 00:21:29,040
Speaker 11: people in the world to traffic accidents. Wouldn't it be

370
00:21:29,080 --> 00:21:32,480
Speaker 11: amazing if Dabok invented something that would save a million

371
00:21:32,520 --> 00:21:33,120
Speaker 11: lives a year.

372
00:21:33,920 --> 00:21:37,280
Speaker 12: In October of two thousand and five, forty three teams

373
00:21:37,320 --> 00:21:40,439
Speaker 12: have brought their vehicles to compete in a unique event,

374
00:21:41,200 --> 00:21:44,840
Speaker 12: a race driven not by testosterone but computer.

375
00:21:44,520 --> 00:21:50,480
Speaker 2: Coke Chapter three Machine Learning.

376
00:21:54,320 --> 00:21:57,640
Speaker 12: The race course is a circular maze that zigzags for one

377
00:21:57,680 --> 00:21:58,359
Speaker 12: hundred and thirty two.

378
00:21:58,440 --> 00:22:02,080
Speaker 2: Eighteen months later this second Grand Challenge, DARPA doubled the

379
00:22:02,080 --> 00:22:05,280
Speaker 2: bounty two million dollars. This footage is from a PBS

380
00:22:05,320 --> 00:22:08,720
Speaker 2: documentary called The Great Robot Race, narrated to My Mild

381
00:22:08,840 --> 00:22:13,160
Speaker 2: Joy by John Lithgow. Familiar faces have returned. Chris Earmsen

382
00:22:13,240 --> 00:22:16,119
Speaker 2: back with the Carnegie Mellon team, the Sign with two vehicles,

383
00:22:16,400 --> 00:22:21,200
Speaker 2: Highlander and Sandstorm. Anthony Lewandowski back with his motorcycle, which

384
00:22:21,320 --> 00:22:24,280
Speaker 2: still doesn't work. He's knocked out in the qualifiers. And

385
00:22:24,320 --> 00:22:28,640
Speaker 2: now there's also Stanford's entrant compared to Sandstorm, the bulked

386
00:22:28,680 --> 00:22:32,760
Speaker 2: up hummer. The car looks easily a blue suv donated

387
00:22:32,760 --> 00:22:36,399
Speaker 2: by Volkswagen. A baby face, Run smiles next to his

388
00:22:36,440 --> 00:22:37,760
Speaker 2: soccer mom looking vehicle.

389
00:22:38,359 --> 00:22:41,280
Speaker 13: The Vika's name is Stanley, so Stanley is nothing else

390
00:22:41,320 --> 00:22:44,360
Speaker 13: but Stanford, but it also gives the vehicle a personality.

391
00:22:45,440 --> 00:22:47,080
Speaker 13: If you think of the Vega more and more as

392
00:22:47,119 --> 00:22:48,600
Speaker 13: an intelligent decision maker.

393
00:22:50,440 --> 00:22:54,600
Speaker 4: Run is a computer scientist, and Thrun really broad more

394
00:22:54,720 --> 00:22:57,879
Speaker 4: artificial intelligence, which at the time we're talking two thousand

395
00:22:57,920 --> 00:23:02,520
Speaker 4: and five was still rather primitive, especially compared to what

396
00:23:02,560 --> 00:23:06,000
Speaker 4: we have today. But he could use it to teach

397
00:23:06,480 --> 00:23:10,320
Speaker 4: his vehicle how to recognize the road and how to

398
00:23:10,359 --> 00:23:13,640
Speaker 4: do it much faster. They found a dirt road out

399
00:23:13,640 --> 00:23:15,920
Speaker 4: near Stanford, and they drive it down a dirt road

400
00:23:16,960 --> 00:23:20,000
Speaker 4: and have the car's cameras record what they were seeing.

401
00:23:20,760 --> 00:23:24,040
Speaker 11: The robot Standy was able to train itself as it

402
00:23:24,320 --> 00:23:27,159
Speaker 11: and the way it worked. Its eyes looked way ahead

403
00:23:27,400 --> 00:23:31,080
Speaker 11: and it could see stuff way at distance. When it

404
00:23:31,160 --> 00:23:32,760
Speaker 11: drives over the stuff, you could tell it wasn't a

405
00:23:32,760 --> 00:23:34,840
Speaker 11: good place to drive or not, because it could measure

406
00:23:34,960 --> 00:23:38,360
Speaker 11: how slippery or how bumpy the vote was. And they

407
00:23:38,400 --> 00:23:41,639
Speaker 11: could then retroactively train and say, say, this green stuff

408
00:23:41,680 --> 00:23:44,200
Speaker 11: over there, it's something good to drive on aka grass,

409
00:23:44,680 --> 00:23:48,000
Speaker 11: and this browner stuff aka mutt is not so good

410
00:23:48,000 --> 00:23:48,359
Speaker 11: to drive.

411
00:23:49,160 --> 00:23:53,720
Speaker 2: And so it was able to detect patterns and generalize

412
00:23:53,880 --> 00:23:54,800
Speaker 2: from what it had learned.

413
00:23:55,160 --> 00:23:55,440
Speaker 4: Yeah.

414
00:23:55,480 --> 00:23:58,880
Speaker 11: Absolutely, and this is like thirty times a second, I mean,

415
00:23:58,960 --> 00:23:59,679
Speaker 11: just like a person.

416
00:24:00,840 --> 00:24:04,160
Speaker 2: The race kicks off with Stanley Sandwich between Carnegie Mountains

417
00:24:04,160 --> 00:24:04,959
Speaker 2: tow behemoths.

418
00:24:05,640 --> 00:24:09,760
Speaker 12: Highlander leads the path, followed by Stanley and Sandstorm.

419
00:24:10,359 --> 00:24:11,840
Speaker 2: What happens in the second race?

420
00:24:12,119 --> 00:24:15,920
Speaker 4: The second race is as successful as the first race

421
00:24:16,200 --> 00:24:20,000
Speaker 4: is disastrous.

422
00:24:20,040 --> 00:24:22,879
Speaker 2: Nearly every entrance in the second race would go further

423
00:24:22,960 --> 00:24:26,280
Speaker 2: than Sandstorm had in the first. Multiple vehicles would finish

424
00:24:26,280 --> 00:24:29,600
Speaker 2: the course. The real question was who would do it fastest?

425
00:24:30,320 --> 00:24:32,159
Speaker 2: And so at what point was it clear to you

426
00:24:32,200 --> 00:24:34,600
Speaker 2: that you were going to win well.

427
00:24:34,640 --> 00:24:38,080
Speaker 11: Once we passed the front running team, we kind of

428
00:24:38,160 --> 00:24:41,480
Speaker 11: saw the vehicle descend into what was the hardest part

429
00:24:41,520 --> 00:24:45,560
Speaker 11: of the race course, a very treachery mountain pass, and

430
00:24:45,920 --> 00:24:49,200
Speaker 11: we saw at a distance a dust cloud. We saw

431
00:24:49,200 --> 00:24:52,240
Speaker 11: a helicopter. We so a little features that must believe

432
00:24:52,280 --> 00:24:55,800
Speaker 11: ow there's something happening that's magical, and this dust cloud

433
00:24:55,880 --> 00:24:58,560
Speaker 11: then all of a sudden turned bluish because the cover

434
00:24:58,680 --> 00:25:01,800
Speaker 11: was blue, and came closer, and then it came first

435
00:25:01,840 --> 00:25:03,760
Speaker 11: to the finish line and was unbelievably magical.

436
00:25:04,480 --> 00:25:07,119
Speaker 2: At the end of the dock over some criminally corny

437
00:25:07,119 --> 00:25:10,680
Speaker 2: piano music. Sebastian Thron gives his post race interview. He's

438
00:25:10,760 --> 00:25:13,159
Speaker 2: dressed a lot like a race car driver. Watching you

439
00:25:13,200 --> 00:25:14,680
Speaker 2: could forget he wasn't in the car.

440
00:25:14,800 --> 00:25:17,760
Speaker 11: It was just amazing to see this community of people,

441
00:25:18,320 --> 00:25:20,159
Speaker 11: that community succeeded.

442
00:25:20,200 --> 00:25:20,520
Speaker 4: Today.

443
00:25:21,200 --> 00:25:23,439
Speaker 11: Behind me, there are three vobos that made it all

444
00:25:23,480 --> 00:25:25,920
Speaker 11: the way through the desert, and all three of them

445
00:25:25,920 --> 00:25:29,639
Speaker 11: did be unthinkable. It's such a fantastic successful this community.

446
00:25:30,160 --> 00:25:35,920
Speaker 2: I think we all win a made for TV Kumbaya moment.

447
00:25:36,240 --> 00:25:38,720
Speaker 2: Still years before the race to build driverless cars would

448
00:25:38,800 --> 00:25:46,439
Speaker 2: enter its cutthroat phase. What would happen next is that

449
00:25:46,520 --> 00:25:49,720
Speaker 2: a small band of lunatics would take driverless cars out

450
00:25:49,720 --> 00:25:53,280
Speaker 2: of the desert start secretly driving them on public roads

451
00:25:53,320 --> 00:25:57,479
Speaker 2: in the state of California. They would do this at

452
00:25:57,480 --> 00:25:59,440
Speaker 2: the behest of a man who had been observing from

453
00:25:59,440 --> 00:26:03,320
Speaker 2: the stands that day, disguised and hat and sunglasses, who

454
00:26:03,440 --> 00:26:09,280
Speaker 2: watched the challenge while his mind spun this after a

455
00:26:09,280 --> 00:26:47,080
Speaker 2: short break, Welcome back to the show. Chapter four, something

456
00:26:47,200 --> 00:26:51,000
Speaker 2: actually useful for the world. The race in the Desert

457
00:26:51,040 --> 00:26:54,320
Speaker 2: had been designed as a spectacle, something flashy to dry

458
00:26:54,320 --> 00:26:58,280
Speaker 2: out America's smartest roboticists, but it had drawn another person

459
00:26:58,480 --> 00:27:02,760
Speaker 2: who come for his own reasons. Google's Larry Page arrived

460
00:27:02,760 --> 00:27:05,119
Speaker 2: at the Darker Grand Challenge in a baseball hat and

461
00:27:05,200 --> 00:27:10,000
Speaker 2: sunglasses disguise. He found Sebastian Throne and buttonhold him, asking

462
00:27:10,040 --> 00:27:13,359
Speaker 2: him a million highly specific questions about things like the

463
00:27:13,400 --> 00:27:17,159
Speaker 2: wavelength his light our system used. But this meeting in

464
00:27:17,200 --> 00:27:19,680
Speaker 2: the desert, this was not actually their first introduction.

465
00:27:20,920 --> 00:27:23,000
Speaker 11: Well, the first time I met Larry was a bit earlier.

466
00:27:23,160 --> 00:27:26,159
Speaker 11: He had built a small little robot that acted as

467
00:27:26,160 --> 00:27:28,320
Speaker 11: a tailor presence for meetings, and he was trying to

468
00:27:28,400 --> 00:27:31,280
Speaker 11: drive it around the Google officers instead of himself going

469
00:27:31,320 --> 00:27:33,879
Speaker 11: to meeting with a robot. And he sent me a

470
00:27:33,880 --> 00:27:35,680
Speaker 11: message and said, I'm going to show you the vote

471
00:27:35,680 --> 00:27:39,880
Speaker 11: I've built. And I, in a spur of like craziness,

472
00:27:39,920 --> 00:27:43,320
Speaker 11: I sent the message Breck saying, Larry, I'm so glad

473
00:27:43,760 --> 00:27:45,720
Speaker 11: that Google it he used twenty percent of time. It

474
00:27:45,840 --> 00:27:51,800
Speaker 11: was something useful for the world. I couldn't. I either

475
00:27:51,880 --> 00:27:56,159
Speaker 11: expected a rapid response or never hear from him again.

476
00:27:57,280 --> 00:27:59,600
Speaker 11: It turns out I was lucky. He responded immediately. I

477
00:27:59,640 --> 00:28:02,120
Speaker 11: took his role, would fix it next fenty four hours.

478
00:28:01,880 --> 00:28:06,040
Speaker 2: And he was very heavy. Larry Page, it turned out,

479
00:28:06,119 --> 00:28:09,359
Speaker 2: had actually been interested in autonomous vehicles since at least

480
00:28:09,359 --> 00:28:12,080
Speaker 2: grad school. That's what he'd wanted to do his thesis

481
00:28:12,119 --> 00:28:16,000
Speaker 2: on before being guided by some wise PhD advisor towards

482
00:28:16,000 --> 00:28:20,679
Speaker 2: search engines instead. Now as a spectator at DARPA's second

483
00:28:20,680 --> 00:28:24,159
Speaker 2: Grand Challenge, he could see real world evidence that autonomous

484
00:28:24,240 --> 00:28:28,399
Speaker 2: vehicles might actually be a thing. At first, Larry Page

485
00:28:28,440 --> 00:28:32,320
Speaker 2: hires Sebastian's run along with fellow Darbik contestant Anthony Lewandowski,

486
00:28:32,640 --> 00:28:35,720
Speaker 2: just to build what will become Google street View. They'll

487
00:28:35,720 --> 00:28:39,400
Speaker 2: actually modify the system that Stanley the car's roof mounted

488
00:28:39,440 --> 00:28:44,360
Speaker 2: cameras had used to begin photographing American streets. But before long,

489
00:28:44,800 --> 00:28:48,000
Speaker 2: Larry Page returns to Sebastian with his dream of a

490
00:28:48,080 --> 00:28:53,480
Speaker 2: driverless car, and so how soon after arriving at Google

491
00:28:53,800 --> 00:28:56,880
Speaker 2: this project chauffeur again, like Larry Page says to you,

492
00:28:58,080 --> 00:28:59,760
Speaker 2: I have a mission, like how does this happen?

493
00:29:00,120 --> 00:29:02,440
Speaker 11: And this is an embarrassing moment for me. It's about

494
00:29:02,440 --> 00:29:04,600
Speaker 11: two years later, two thousand and nine, where I sit

495
00:29:04,680 --> 00:29:08,800
Speaker 11: in a cubicle and like Page comes by and says, Sebastian,

496
00:29:09,520 --> 00:29:11,400
Speaker 11: I think you should build a self diving car that

497
00:29:11,480 --> 00:29:16,520
Speaker 11: can drive anywhere in the world. And my immediate reaction was, no,

498
00:29:17,920 --> 00:29:20,280
Speaker 11: taking the technology we build for this empty desert and

499
00:29:20,320 --> 00:29:21,719
Speaker 11: put it in the middle of Market Street in San

500
00:29:21,720 --> 00:29:26,680
Speaker 11: Francisco is going to kill somebody. And Larry would come

501
00:29:26,720 --> 00:29:28,960
Speaker 11: back the next day with the same idea, and I

502
00:29:29,000 --> 00:29:31,760
Speaker 11: would give them the same answer, and both of us

503
00:29:31,880 --> 00:29:35,640
Speaker 11: got increasingly more frustrated. God damn it, it can't be done,

504
00:29:36,040 --> 00:29:37,800
Speaker 11: and eventually came and said, look, Sebastian, OK, care, I

505
00:29:37,800 --> 00:29:40,200
Speaker 11: get it. You can't do it. I want to explain

506
00:29:40,240 --> 00:29:42,640
Speaker 11: to Erk Schmidt the CEO at the time and Sergey

507
00:29:42,640 --> 00:29:45,320
Speaker 11: Britt my cofounder, why it can't be done? Can you

508
00:29:45,360 --> 00:29:47,440
Speaker 11: give me the technical reason why it can't be done?

509
00:29:48,000 --> 00:29:50,760
Speaker 11: And that's the moment of incredible pain, because I go

510
00:29:50,800 --> 00:29:52,800
Speaker 11: home and I can't think of a technical reason why not.

511
00:29:53,520 --> 00:29:55,080
Speaker 11: It was this kind of moment where I felt, look,

512
00:29:55,120 --> 00:29:57,320
Speaker 11: I'm the world expert on self diving cars, and I'm

513
00:29:57,360 --> 00:30:01,360
Speaker 11: the person who denies that it can be done. Like

514
00:30:01,880 --> 00:30:06,160
Speaker 11: that taught me an incredibly important lesson about experts that

515
00:30:06,480 --> 00:30:10,040
Speaker 11: for the rest of my life, I decided experts I

516
00:30:10,120 --> 00:30:11,000
Speaker 11: usually explore.

517
00:30:10,720 --> 00:30:11,840
Speaker 4: The past and not the future.

518
00:30:12,600 --> 00:30:15,720
Speaker 11: And if you ask an expert about innovation, something crazy new,

519
00:30:16,440 --> 00:30:18,680
Speaker 11: they're the least likely person to say, yes, it can

520
00:30:18,760 --> 00:30:19,080
Speaker 11: be done.

521
00:30:20,120 --> 00:30:22,560
Speaker 2: So this is where the Google's self driving car project

522
00:30:22,640 --> 00:30:25,520
Speaker 2: begins in two thousand and nine. It's led by Sebastian,

523
00:30:25,640 --> 00:30:28,600
Speaker 2: joined by others from the Darker Challenges. The methodical Chris

524
00:30:28,720 --> 00:30:32,080
Speaker 2: Armsen was running most things day to day. Anthony Lewandowski,

525
00:30:32,160 --> 00:30:35,760
Speaker 2: the flashy motorcycle guy, would work on hardware. Dmitriy Dolgov,

526
00:30:35,800 --> 00:30:39,240
Speaker 2: another darker veteran, would be responsible for planning and optimization.

527
00:30:39,920 --> 00:30:43,240
Speaker 2: It was a secret project did report directly to Larry Page,

528
00:30:43,600 --> 00:30:46,640
Speaker 2: a small enough team that there'd be no bureaucracy, few emails,

529
00:30:46,720 --> 00:30:51,240
Speaker 2: fewer meetings, just eleven engineers, who writer Alex Davies says,

530
00:30:51,360 --> 00:30:53,760
Speaker 2: represented some of the best young talent in the country.

531
00:30:54,360 --> 00:30:59,080
Speaker 4: And so Google builds this very quiet team and it

532
00:30:59,120 --> 00:31:03,600
Speaker 4: says to them, buildness a self driving car. And because

533
00:31:03,760 --> 00:31:07,240
Speaker 4: that goal is super nebulous, they give them two challenges.

534
00:31:07,840 --> 00:31:13,640
Speaker 4: They say, safely log one hundred thousand miles on public roads.

535
00:31:14,480 --> 00:31:17,600
Speaker 4: But they also give them a challenge called the Larry

536
00:31:17,600 --> 00:31:18,000
Speaker 4: one K.

537
00:31:19,040 --> 00:31:21,760
Speaker 11: So Larry and Serge and I said together and the

538
00:31:21,800 --> 00:31:25,680
Speaker 11: two of them carved out one thousand total miles of

539
00:31:25,800 --> 00:31:27,280
Speaker 11: road surface in California.

540
00:31:27,520 --> 00:31:30,440
Speaker 4: They open up Google Maps and they just click around

541
00:31:30,520 --> 00:31:35,040
Speaker 4: and they look for ten separate one hundred mile routes

542
00:31:36,000 --> 00:31:37,600
Speaker 4: that are really tricky.

543
00:31:37,880 --> 00:31:41,320
Speaker 11: Absolutely everything like the Bay Bridge and Lake Tao and

544
00:31:41,520 --> 00:31:44,840
Speaker 11: Highway one to Los Angeles and Market Street and even

545
00:31:44,840 --> 00:31:45,960
Speaker 11: crooked Lambas Street.

546
00:31:46,240 --> 00:31:48,400
Speaker 4: And they say to the team, you have to drive

547
00:31:48,560 --> 00:31:52,880
Speaker 4: each of these one hundred mile routes without one human

548
00:31:52,960 --> 00:31:56,840
Speaker 4: takeover of the system, without one failure of the car to.

549
00:31:56,800 --> 00:31:59,120
Speaker 2: Get off to your running start. The team licenses the

550
00:31:59,160 --> 00:32:03,600
Speaker 2: code from Sanford darpa Urban Challenge vehicle. Anthony Lewandowski goes

551
00:32:03,600 --> 00:32:06,880
Speaker 2: to a local Toyota dealership and buys eight priuses, takes

552
00:32:06,880 --> 00:32:09,480
Speaker 2: them back to Google and retrofits them to accept a

553
00:32:09,520 --> 00:32:14,360
Speaker 2: computer as a driver. He hooks that computer driver electronically

554
00:32:14,720 --> 00:32:18,600
Speaker 2: into the brakes, the gas, the steering. These Priuses get

555
00:32:18,600 --> 00:32:22,120
Speaker 2: a radar system behind the bumper cameras alied Our system

556
00:32:22,160 --> 00:32:25,840
Speaker 2: spenning three hundred and sixty degrees on topli like radar,

557
00:32:25,960 --> 00:32:29,400
Speaker 2: but it shoots lasers instead of sound waves. At first,

558
00:32:29,440 --> 00:32:32,840
Speaker 2: the team gives each Prius a cool name, like night Rider.

559
00:32:33,280 --> 00:32:35,360
Speaker 14: But I think we quickly realized that we're not going

560
00:32:35,440 --> 00:32:37,120
Speaker 14: to be able to name all these vehicles as we

561
00:32:37,160 --> 00:32:38,760
Speaker 14: scale up our fleet, and so we just started to

562
00:32:38,880 --> 00:32:41,320
Speaker 14: number them, like you know, Prius twenty seven.

563
00:32:42,040 --> 00:32:45,160
Speaker 2: This is Don Burnett. He'd been a researcher working on

564
00:32:45,240 --> 00:32:48,320
Speaker 2: autonomous submarines. He lost a friend in a car accident,

565
00:32:48,440 --> 00:32:51,720
Speaker 2: separately gotten a bad accident himself, and decided he wanted

566
00:32:51,720 --> 00:32:54,320
Speaker 2: to do work on self driving cars. That's how he

567
00:32:54,480 --> 00:32:55,760
Speaker 2: eventually ended up on the team.

568
00:32:55,800 --> 00:32:58,560
Speaker 14: In its early days, I was on the motion planning

569
00:32:58,560 --> 00:33:02,320
Speaker 14: and behavior decision making team, and my responsibility was to

570
00:33:02,440 --> 00:33:04,400
Speaker 14: work on the nudging behavior.

571
00:33:05,120 --> 00:33:08,000
Speaker 2: Nudging what a big truck passes a human driver on

572
00:33:08,000 --> 00:33:10,640
Speaker 2: the right, The driver will nudge a little to the left.

573
00:33:10,960 --> 00:33:13,840
Speaker 2: For us, it's an instinct. Don's job was to teach

574
00:33:13,840 --> 00:33:15,080
Speaker 2: a computer to nudge.

575
00:33:15,520 --> 00:33:18,560
Speaker 14: They're trying to encode the behavior that you would use

576
00:33:18,600 --> 00:33:22,240
Speaker 14: as a driver under kind of partially good perception.

577
00:33:22,800 --> 00:33:24,360
Speaker 15: And it's a really tricky problem.

578
00:33:24,920 --> 00:33:28,160
Speaker 2: A team of academic roboticists, some of whom had had

579
00:33:28,200 --> 00:33:31,360
Speaker 2: friends die in cars, spending Google's money to see if

580
00:33:31,360 --> 00:33:34,240
Speaker 2: they could make driving safer. It was a weird era.

581
00:33:38,640 --> 00:33:41,600
Speaker 2: There's this big concert venue near Google's offices called the

582
00:33:41,640 --> 00:33:44,720
Speaker 2: Shoreline Amphitheater. In two thousand and nine, you could have

583
00:33:44,720 --> 00:33:49,040
Speaker 2: seen Cheryl Crow there the Killers Fish. But the most

584
00:33:49,080 --> 00:33:51,800
Speaker 2: interesting show that year was one almost nobody knew about.

585
00:33:52,480 --> 00:33:54,560
Speaker 2: In the venue parking lot. On days when there was

586
00:33:54,560 --> 00:33:57,400
Speaker 2: no concert, no tour buses around to see them, the

587
00:33:57,480 --> 00:33:59,719
Speaker 2: Google team would run its first test runs of their

588
00:33:59,760 --> 00:34:05,400
Speaker 2: driverless cars, essentially hiding in plane sight a prius driving

589
00:34:05,440 --> 00:34:08,880
Speaker 2: itself around the Amphitheater parking lot with an attentive safety

590
00:34:08,920 --> 00:34:11,959
Speaker 2: driver sitting behind the wheel just in case. The team

591
00:34:12,040 --> 00:34:14,920
Speaker 2: was making sure the basics functioned that the censors could

592
00:34:14,960 --> 00:34:17,919
Speaker 2: really recognize another car that the computer in the car

593
00:34:18,000 --> 00:34:21,040
Speaker 2: was abiding by their orders. These were the baby steps

594
00:34:21,560 --> 00:34:24,200
Speaker 2: that happened in this parking lot and at an empty

595
00:34:24,239 --> 00:34:28,200
Speaker 2: airplane runway that was close to their offices. Spring two

596
00:34:28,200 --> 00:34:31,560
Speaker 2: thousand and nine, the team tries actual real road driving

597
00:34:31,600 --> 00:34:34,400
Speaker 2: for the first time. Chris Armson takes one of the

598
00:34:34,400 --> 00:34:37,799
Speaker 2: priuses out on the Central Expressway, speed limit forty five

599
00:34:37,800 --> 00:34:41,920
Speaker 2: miles per hour. There are humans driving here and immediately

600
00:34:42,040 --> 00:34:44,560
Speaker 2: outside the confines of the empty parking lot and empty

601
00:34:44,560 --> 00:34:48,960
Speaker 2: airplane runway. Here's what's clear. They had a real problem.

602
00:34:49,200 --> 00:34:50,840
Speaker 2: The car was swerving wildly.

603
00:34:51,760 --> 00:34:54,879
Speaker 8: It was weaving around like a drunken sailor. And we

604
00:34:55,160 --> 00:34:58,640
Speaker 8: realized that the scale of the runway was such that

605
00:34:58,680 --> 00:35:01,600
Speaker 8: you didn't notice the one or two foot kind of

606
00:35:02,120 --> 00:35:06,480
Speaker 8: oscillation it had in lateral control, and you put it

607
00:35:06,480 --> 00:35:10,840
Speaker 8: on Central Expressway and suddenly, you know, yep. Turns out, actually,

608
00:35:10,840 --> 00:35:14,400
Speaker 8: that's a problem.

609
00:35:13,000 --> 00:35:20,239
Speaker 2: One more problem to fix. Listening to the story, it's

610
00:35:20,239 --> 00:35:22,600
Speaker 2: funny because I can imagine it giving me a totally

611
00:35:22,600 --> 00:35:26,400
Speaker 2: different feeling than it does. A tech company with nobody's

612
00:35:26,400 --> 00:35:30,759
Speaker 2: permission was testing driverless cars on public roads in California.

613
00:35:31,680 --> 00:35:33,800
Speaker 2: I don't know why that strikes me as being about

614
00:35:33,800 --> 00:35:38,280
Speaker 2: invention instead of just hubris and impunity. Maybe it's because

615
00:35:38,320 --> 00:35:40,120
Speaker 2: I know that Google would be one of the few

616
00:35:40,160 --> 00:35:43,440
Speaker 2: tech companies whose driverless cars would not cause any fatal

617
00:35:43,480 --> 00:35:47,200
Speaker 2: accidents in testing, and that the team would just take

618
00:35:47,320 --> 00:35:50,319
Speaker 2: more safety precautions than the other companies who'd rush in

619
00:35:50,400 --> 00:35:52,480
Speaker 2: later to catch up with them once. This was an

620
00:35:52,600 --> 00:35:56,560
Speaker 2: arms race. The way these cars were designed, the safety

621
00:35:56,640 --> 00:35:59,160
Speaker 2: driver set behind the steering wheel, ready to take over

622
00:36:00,000 --> 00:36:02,800
Speaker 2: when the other seat was their partner watching the monitor

623
00:36:02,840 --> 00:36:07,040
Speaker 2: displaying a graphical interface designed by Dmitri Dolgov. The people

624
00:36:07,080 --> 00:36:10,319
Speaker 2: watching the screen would call out problems ahead, some discrepancy

625
00:36:10,360 --> 00:36:12,480
Speaker 2: between what the sensors were seeing and what was actually

626
00:36:12,480 --> 00:36:15,400
Speaker 2: in the road. This is what teaching a car to

627
00:36:15,520 --> 00:36:19,000
Speaker 2: drive actually looked like. Two person teams spanning the cars,

628
00:36:19,239 --> 00:36:22,360
Speaker 2: logging errors, going back to the office to troubleshoot, and

629
00:36:22,400 --> 00:36:26,080
Speaker 2: then updating the code. I asked Don Burnette about this era,

630
00:36:26,880 --> 00:36:28,919
Speaker 2: and while you're doing this and then like you leave

631
00:36:28,960 --> 00:36:31,040
Speaker 2: work and you get in your car that you drive

632
00:36:31,080 --> 00:36:34,040
Speaker 2: as a human, did you find yourself thinking more carefully, like,

633
00:36:34,520 --> 00:36:36,640
Speaker 2: how do I know what I know when I'm driving,

634
00:36:36,680 --> 00:36:38,799
Speaker 2: like you're trying to teach a machine by day, did

635
00:36:38,800 --> 00:36:40,879
Speaker 2: it affect how you thought about human driving? By night?

636
00:36:41,520 --> 00:36:45,440
Speaker 14: Almost obnoxiously so to any passengers in the car with me.

637
00:36:46,000 --> 00:36:51,320
Speaker 14: I was obsessed with one big question, which is why

638
00:36:51,360 --> 00:36:55,160
Speaker 14: do humans drive the way they drive? And it turns

639
00:36:55,160 --> 00:36:57,399
Speaker 14: out there were no good answers, and I still think

640
00:36:57,400 --> 00:37:00,960
Speaker 14: they're not great answers. And instead of actually answering that question,

641
00:37:01,120 --> 00:37:04,239
Speaker 14: we've just turned to machine learning to infer the deep

642
00:37:04,320 --> 00:37:07,799
Speaker 14: truths behind why humans do what they do. Then, so

643
00:37:07,840 --> 00:37:11,160
Speaker 14: there's some basic principles that you can understand, Like we

644
00:37:11,239 --> 00:37:14,440
Speaker 14: try to minimize lateral acceleration, meaning you don't want to

645
00:37:14,480 --> 00:37:16,080
Speaker 14: be thrown to the outside of your car when you're

646
00:37:16,080 --> 00:37:16,560
Speaker 14: making a turn.

647
00:37:16,600 --> 00:37:18,360
Speaker 15: So you're going to slow down, but how much do

648
00:37:18,440 --> 00:37:19,040
Speaker 15: you slow down?

649
00:37:19,200 --> 00:37:19,399
Speaker 4: Right?

650
00:37:19,800 --> 00:37:21,200
Speaker 15: And it turns out that's contextual.

651
00:37:23,480 --> 00:37:26,239
Speaker 2: Don gave me an example. So you're trying to figure

652
00:37:26,239 --> 00:37:28,600
Speaker 2: out the right speed and angle for the car on

653
00:37:28,600 --> 00:37:31,080
Speaker 2: one of those tight curvy on ramps onto the highway.

654
00:37:31,560 --> 00:37:34,560
Speaker 2: You want it to feel comfortable for a passenger. Don says,

655
00:37:34,640 --> 00:37:37,080
Speaker 2: you can work out the math. The lateral acceleration is

656
00:37:37,080 --> 00:37:40,239
Speaker 2: two meters per second squared but the surprising thing is

657
00:37:40,640 --> 00:37:43,200
Speaker 2: that number only applies on the on ramp.

658
00:37:45,200 --> 00:37:48,520
Speaker 14: If I put you at a col de sac in

659
00:37:48,600 --> 00:37:51,160
Speaker 14: a neighborhood and you were going to do a U

660
00:37:51,200 --> 00:37:54,080
Speaker 14: turn at the end of the cold de sac, even

661
00:37:54,160 --> 00:37:58,239
Speaker 14: though the speed is significantly slower, if you did two

662
00:37:58,239 --> 00:38:01,879
Speaker 14: meters per second squared of lateral acceleration around a cul

663
00:38:01,880 --> 00:38:05,240
Speaker 14: de sac, you would tell your driver they were crazy.

664
00:38:05,920 --> 00:38:10,440
Speaker 14: It would be incredibly uncomfortable, like incredibly uncomfortable.

665
00:38:10,600 --> 00:38:12,320
Speaker 2: You would feel like you're in Mario Kart.

666
00:38:12,600 --> 00:38:14,360
Speaker 15: Yes, it would feel Mario Kart.

667
00:38:14,520 --> 00:38:17,480
Speaker 14: And remember this is a force, so it's a physical

668
00:38:17,560 --> 00:38:20,440
Speaker 14: feeling on your body is exactly the same. But the

669
00:38:20,440 --> 00:38:24,600
Speaker 14: contextual awareness of the situation of speeding up to get

670
00:38:24,640 --> 00:38:27,239
Speaker 14: on the highway versus making a U turn in a

671
00:38:27,280 --> 00:38:33,280
Speaker 14: residential street tricks your brain into feeling opposite about the situation.

672
00:38:33,800 --> 00:38:35,439
Speaker 14: And so it turns out the limit for a cul

673
00:38:35,480 --> 00:38:38,799
Speaker 14: de sac is around point seventy five. It's almost three

674
00:38:38,840 --> 00:38:42,200
Speaker 14: times less than you would be willing to tolerate as

675
00:38:42,239 --> 00:38:43,760
Speaker 14: you accelerate onto a highway.

676
00:38:44,280 --> 00:38:45,560
Speaker 15: And so there were.

677
00:38:45,360 --> 00:38:49,160
Speaker 14: Things like that where you couldn't just say humans have

678
00:38:49,320 --> 00:38:56,560
Speaker 14: specific physical restrictions right from a force's perspective, the context matters,

679
00:38:56,920 --> 00:38:58,960
Speaker 14: and when the context matters, now all of a sudden,

680
00:38:59,000 --> 00:38:59,760
Speaker 14: anything is game.

681
00:39:00,160 --> 00:39:02,920
Speaker 15: So things like that is where.

682
00:39:02,719 --> 00:39:05,880
Speaker 14: I spent my time as a researcher trying to figure out, Okay,

683
00:39:06,000 --> 00:39:07,960
Speaker 14: how are we going to make this comfortable for passengers?

684
00:39:08,680 --> 00:39:11,880
Speaker 2: All these little problems to solve. But there's one gift,

685
00:39:12,120 --> 00:39:14,239
Speaker 2: which is that the team at this point had an

686
00:39:14,280 --> 00:39:18,799
Speaker 2: overarching goal uniting them. The Darba Challenge told them drive

687
00:39:18,840 --> 00:39:21,759
Speaker 2: across this patch of desert Valaria. One k Challenge told

688
00:39:21,800 --> 00:39:26,600
Speaker 2: them drive these ten roots without human intervention. The specificity

689
00:39:26,600 --> 00:39:29,040
Speaker 2: of the mission meant they never had to squabble about

690
00:39:29,040 --> 00:39:32,239
Speaker 2: why they were there. By twenty ten, just a year in,

691
00:39:32,560 --> 00:39:33,839
Speaker 2: the team was really on a roll.

692
00:39:35,000 --> 00:39:37,200
Speaker 4: They start knocking out roots.

693
00:39:37,840 --> 00:39:40,600
Speaker 14: Each one of the routes was unique and distinct and

694
00:39:40,640 --> 00:39:42,680
Speaker 14: different and had its own challenges.

695
00:39:42,920 --> 00:39:46,600
Speaker 4: Down Route one Silicon Valley to Car Mount.

696
00:39:46,680 --> 00:39:49,600
Speaker 14: The Bridges run where we had to go across all

697
00:39:49,640 --> 00:39:52,080
Speaker 14: of the bridges in the Bay area, starting in Mountain View,

698
00:39:52,400 --> 00:39:54,600
Speaker 14: finishing crossing the Golden Gate Bridge.

699
00:39:54,680 --> 00:39:57,520
Speaker 4: It's Chris Hermsen in the car. It's Anthony Lewandowski in

700
00:39:57,560 --> 00:39:58,040
Speaker 4: the car.

701
00:39:58,320 --> 00:40:01,760
Speaker 14: I was in the car with Dimitri, Chris and Anthony.

702
00:40:01,800 --> 00:40:03,560
Speaker 14: It was the four of us in the prius.

703
00:40:03,800 --> 00:40:06,640
Speaker 4: They're figuring out the technology much faster than they thought

704
00:40:06,640 --> 00:40:07,040
Speaker 4: they could.

705
00:40:07,239 --> 00:40:09,440
Speaker 2: The Larry one K was set up like a video game,

706
00:40:09,640 --> 00:40:12,120
Speaker 2: meaning they'd get to try the route over and over

707
00:40:12,239 --> 00:40:14,960
Speaker 2: until they could complete it without a single human takeover.

708
00:40:15,719 --> 00:40:17,439
Speaker 2: Then they'd move on to the next one.

709
00:40:17,719 --> 00:40:21,680
Speaker 8: It was really a proof of concept exercise. Can you

710
00:40:21,800 --> 00:40:23,760
Speaker 8: even make this happen?

711
00:40:24,400 --> 00:40:26,840
Speaker 4: Once? When they fail a route, they know what the

712
00:40:26,880 --> 00:40:29,759
Speaker 4: car can't handle, so they go back and say they

713
00:40:29,840 --> 00:40:31,760
Speaker 4: have to be better at doing XYZ.

714
00:40:32,000 --> 00:40:36,600
Speaker 14: And then we got back to the office, we regrouped,

715
00:40:36,760 --> 00:40:38,920
Speaker 14: we went back out I think at like eleven PM,

716
00:40:39,320 --> 00:40:41,839
Speaker 14: and by one am we had completed the route.

717
00:40:41,920 --> 00:40:45,439
Speaker 4: They buy a bottle of Corbel champagne. They all write

718
00:40:45,440 --> 00:40:46,239
Speaker 4: their names on it.

719
00:40:46,560 --> 00:40:49,520
Speaker 2: Corbell thirteen ninety nine, a bottle the champagne they have

720
00:40:49,560 --> 00:40:52,920
Speaker 2: at Trader Joe's. They had won for every route they completed.

721
00:40:52,880 --> 00:40:55,160
Speaker 4: And one by one they pick off the Larry one

722
00:40:55,239 --> 00:40:58,080
Speaker 4: K routes. And they think this is going to take

723
00:40:58,120 --> 00:41:00,839
Speaker 4: them about two years when they start out, and they

724
00:41:00,920 --> 00:41:03,240
Speaker 4: do it in a little bit more than a year,

725
00:41:03,800 --> 00:41:06,200
Speaker 4: nearly twice as fast as they had expected.

726
00:41:07,000 --> 00:41:10,400
Speaker 2: By fall of twenty ten. They're done. Here's Chris Armsen.

727
00:41:10,360 --> 00:41:13,120
Speaker 8: And I think we had a big party up at

728
00:41:13,280 --> 00:41:16,440
Speaker 8: Sebastian's house and Los Athos Hills. So you know, it

729
00:41:16,480 --> 00:41:17,760
Speaker 8: was pretty spectacular, right.

730
00:41:17,800 --> 00:41:21,680
Speaker 4: They throw each other in the pool, they celebrate, and

731
00:41:21,719 --> 00:41:25,040
Speaker 4: then they're not entirely sure what to do next.

732
00:41:25,440 --> 00:41:27,680
Speaker 8: It was kind of a okay, And now.

733
00:41:27,560 --> 00:41:31,000
Speaker 2: What the team had pulled off a kind of miracle

734
00:41:31,120 --> 00:41:36,200
Speaker 2: in a year, a driverless car with human supervision, with

735
00:41:36,360 --> 00:41:40,560
Speaker 2: lots of human coding, but still a driverless car successfully

736
00:41:40,600 --> 00:41:44,920
Speaker 2: navigating some very tricky roads in California. They've done this safely,

737
00:41:45,120 --> 00:41:49,200
Speaker 2: they've done it quickly, and now things would begin to wobble.

738
00:41:50,360 --> 00:41:53,760
Speaker 2: Competition would arrive, the team itself would begin to schism,

739
00:41:54,400 --> 00:41:56,960
Speaker 2: and one member, a person who believed the team was

740
00:41:57,000 --> 00:42:00,000
Speaker 2: moving too slowly, would actually take matters into his own

741
00:42:00,080 --> 00:42:25,200
Speaker 2: hands in a particularly extreme way after the break mutiny.

742
00:42:27,080 --> 00:42:33,439
Speaker 2: Welcome back to the show. As early as twenty ten,

743
00:42:33,640 --> 00:42:37,160
Speaker 2: Google's Driver's car project had developed some very impressive self

744
00:42:37,200 --> 00:42:40,880
Speaker 2: driving technology, but what they were struggling to decide was this,

745
00:42:41,719 --> 00:42:44,759
Speaker 2: what was the actual product they were developing. Here, here's

746
00:42:44,760 --> 00:42:45,720
Speaker 2: a Bastian throne.

747
00:42:46,040 --> 00:42:47,840
Speaker 11: We had a lot of debates inside Google what the

748
00:42:47,920 --> 00:42:51,919
Speaker 11: right business model was. At some point, v actually had

749
00:42:51,920 --> 00:42:54,799
Speaker 11: a big debate Beato just by Tesla, and Tesla was

750
00:42:54,840 --> 00:42:56,759
Speaker 11: worth two billion dollars at the time. I remember this,

751
00:42:58,920 --> 00:43:03,120
Speaker 11: maybe you should have been hindsight, but joking is idea.

752
00:43:03,239 --> 00:43:05,759
Speaker 11: There was a debate whether this is more of an

753
00:43:05,760 --> 00:43:09,880
Speaker 11: assistive technology or a disruptive replacements anology.

754
00:43:11,680 --> 00:43:14,800
Speaker 2: Basically, should they follow the route that Tesla ultimately d

755
00:43:15,200 --> 00:43:17,759
Speaker 2: design self driving as a feature in your car, something

756
00:43:17,800 --> 00:43:20,600
Speaker 2: that could take over sometimes but still need human monitoring,

757
00:43:21,719 --> 00:43:23,799
Speaker 2: or was it better to wait until the car could

758
00:43:23,880 --> 00:43:27,239
Speaker 2: fully drive itself. Thron would eventually come around to this

759
00:43:27,360 --> 00:43:30,160
Speaker 2: version of self driving. Specifically, he'd come around to the

760
00:43:30,160 --> 00:43:32,640
Speaker 2: idea of self driving robotaxis.

761
00:43:33,400 --> 00:43:37,000
Speaker 11: A taxi service type system is way more capital efficient

762
00:43:37,440 --> 00:43:39,960
Speaker 11: than ownership. An owned car is being used for four

763
00:43:39,960 --> 00:43:42,160
Speaker 11: percent of the time, and it's parked ninety six plent

764
00:43:42,200 --> 00:43:45,239
Speaker 11: a time. Imagine a city without parked cars, where every

765
00:43:45,239 --> 00:43:47,480
Speaker 11: car is being utilized called it fifty percent of the time,

766
00:43:48,000 --> 00:43:50,640
Speaker 11: which means we have like only ten percent number of

767
00:43:50,640 --> 00:43:53,920
Speaker 11: cars needed that we need today when we own own cars,

768
00:43:54,320 --> 00:43:56,000
Speaker 11: that's going to happen. There's no absolut question.

769
00:43:57,480 --> 00:44:00,000
Speaker 2: What Sebastian is describing here, as a matter of fact,

770
00:44:01,040 --> 00:44:05,440
Speaker 2: is a fairly radical reimagination of American cities. The idea

771
00:44:05,480 --> 00:44:09,239
Speaker 2: that robotaxis would be so cheap and widely available that

772
00:44:09,280 --> 00:44:12,319
Speaker 2: most people just wouldn't own cars, that we could put

773
00:44:12,400 --> 00:44:15,040
Speaker 2: something else, anything else, in the places where we put

774
00:44:15,120 --> 00:44:18,560
Speaker 2: most of our parking lots and parking spaces. That is

775
00:44:18,560 --> 00:44:21,680
Speaker 2: a far fetched idea, just given how much of American

776
00:44:21,719 --> 00:44:26,480
Speaker 2: identity is tied into personal car ownership. A farvetched idea,

777
00:44:26,560 --> 00:44:29,279
Speaker 2: and for it to begin to happen, Google would have

778
00:44:29,360 --> 00:44:32,280
Speaker 2: to bring a product to market. But the years passed

779
00:44:32,440 --> 00:44:35,840
Speaker 2: and they didn't, and some people who were there felt stuck.

780
00:44:36,680 --> 00:44:40,000
Speaker 2: Don Burnette says he believes life at Google got dangerously cushy.

781
00:44:40,719 --> 00:44:43,720
Speaker 2: The food was great, the money was too, these former

782
00:44:43,760 --> 00:44:46,160
Speaker 2: academics making much more than they'd ever expected.

783
00:44:49,400 --> 00:44:52,520
Speaker 14: There was a lack of urgency on the team to

784
00:44:52,640 --> 00:44:57,319
Speaker 14: actually make something viable. We had a funding supply that

785
00:44:57,360 --> 00:45:01,080
Speaker 14: effectively felt infinite, and maybe it was, maybe it wasn't,

786
00:45:01,360 --> 00:45:04,280
Speaker 14: but it certainly felt infinite. And when you have infinite funding,

787
00:45:04,600 --> 00:45:07,280
Speaker 14: you're not forced to make hard decisions, You're not forced

788
00:45:07,280 --> 00:45:11,040
Speaker 14: to focus, you're not forced to look at the opportunity,

789
00:45:11,200 --> 00:45:14,319
Speaker 14: the market, the customer and be the best. It was

790
00:45:14,360 --> 00:45:16,520
Speaker 14: more like, hey, let's take our time, let's make sure

791
00:45:16,520 --> 00:45:20,400
Speaker 14: we do it right, which is on its face a

792
00:45:20,440 --> 00:45:23,799
Speaker 14: good principle, but at the end of the day, I

793
00:45:23,840 --> 00:45:27,240
Speaker 14: think the lack of urgency wasn't for everyone.

794
00:45:27,840 --> 00:45:31,040
Speaker 4: And within the team you got team Chris and team Anthony,

795
00:45:31,719 --> 00:45:34,640
Speaker 4: and they start butting heads all the time.

796
00:45:35,080 --> 00:45:38,120
Speaker 2: Chris and Anthony meaning Chris Armsen, official head of the project,

797
00:45:38,360 --> 00:45:41,000
Speaker 2: versus Anthony Lewandowski, who I still think of as the

798
00:45:41,000 --> 00:45:41,799
Speaker 2: motorcycle guy.

799
00:45:42,600 --> 00:45:45,279
Speaker 4: The main difference in their approach is how quickly they

800
00:45:45,280 --> 00:45:49,800
Speaker 4: want to move. Anthony is very okay with risk.

801
00:45:50,520 --> 00:45:52,240
Speaker 2: We'll say.

802
00:45:53,120 --> 00:45:55,680
Speaker 4: He gets one of these cars and he's driving it back,

803
00:45:55,680 --> 00:45:58,520
Speaker 4: and he lives in Berkeley, works in Palauto. He's just

804
00:45:58,840 --> 00:46:02,960
Speaker 4: using this car like the Bay Bridge every day, probably

805
00:46:03,040 --> 00:46:06,200
Speaker 4: outside the bounds of what the team actually wanted, and

806
00:46:06,239 --> 00:46:09,439
Speaker 4: he's not necessarily logging data. He's just enjoying his self

807
00:46:09,520 --> 00:46:12,760
Speaker 4: driving car and taking it all over the place. Chris

808
00:46:12,800 --> 00:46:17,240
Speaker 4: comes from an academic background. He's that Canadian, very nice,

809
00:46:17,480 --> 00:46:20,120
Speaker 4: very careful, very risk averse.

810
00:46:21,480 --> 00:46:24,000
Speaker 2: When I asked Chris Armson about all this, his memory

811
00:46:24,040 --> 00:46:27,920
Speaker 2: was slightly different. In his memory, Team Anthony was pretty

812
00:46:28,000 --> 00:46:31,439
Speaker 2: much just Anthony and Anthony, he said, was a move

813
00:46:31,560 --> 00:46:35,160
Speaker 2: fast and break things kind of guy. Move fast and

814
00:46:35,200 --> 00:46:38,600
Speaker 2: break things a motto famously coined by Mark Zuckerberg. It

815
00:46:38,680 --> 00:46:41,440
Speaker 2: defines a way of developing technology which once might have

816
00:46:41,480 --> 00:46:45,040
Speaker 2: felt cute and revolutionary, but which today, at least to me,

817
00:46:45,160 --> 00:46:49,759
Speaker 2: feels pretty irresponsible. Chris didn't think that philosophy was an

818
00:46:49,800 --> 00:46:53,719
Speaker 2: option for their team, even if their cars were statistically

819
00:46:53,760 --> 00:46:56,920
Speaker 2: safer than human drivers. He knew that the first news

820
00:46:56,920 --> 00:46:59,760
Speaker 2: story about a self driving car in a fatal accident,

821
00:47:00,000 --> 00:47:03,040
Speaker 2: it was going to be a huge deal. Anecdote was

822
00:47:03,080 --> 00:47:07,680
Speaker 2: going to demolish data if they weren't extremely careful. By

823
00:47:07,760 --> 00:47:11,960
Speaker 2: all accounts, Anthony Lewandowski felt differently, but he actually wasn't

824
00:47:12,000 --> 00:47:14,040
Speaker 2: the only one. Here's Don Burnett.

825
00:47:14,880 --> 00:47:18,720
Speaker 14: There were some people on the team, very famously including myself,

826
00:47:18,760 --> 00:47:21,520
Speaker 14: that started to get the itch kind of towards the

827
00:47:21,680 --> 00:47:25,040
Speaker 14: three to four year mark, the itch of like, Okay,

828
00:47:25,520 --> 00:47:27,640
Speaker 14: where is this going, who is it for?

829
00:47:27,960 --> 00:47:29,400
Speaker 15: How are they going to use it? Where are they

830
00:47:29,400 --> 00:47:30,000
Speaker 15: going to use it?

831
00:47:30,280 --> 00:47:33,000
Speaker 14: And I felt like the leadership didn't have great answers

832
00:47:33,040 --> 00:47:35,160
Speaker 14: to that. There was no commercial race, right. We had

833
00:47:35,160 --> 00:47:37,280
Speaker 14: no competition and there was no market for the product.

834
00:47:38,040 --> 00:47:41,800
Speaker 2: But competition would soon arrive in the form of Uber.

835
00:47:45,880 --> 00:47:48,120
Speaker 2: This was the oh shit moment for me.

836
00:47:48,600 --> 00:47:53,040
Speaker 14: Uber announced their self driving program, and I remember like

837
00:47:53,080 --> 00:47:56,799
Speaker 14: it was yesterday, waking up, reading the news, going to

838
00:47:56,840 --> 00:48:00,160
Speaker 14: my desk in the morning, and thinking, Oh crap, these

839
00:48:00,200 --> 00:48:01,360
Speaker 14: guys are going to eat our lunch.

840
00:48:02,880 --> 00:48:06,560
Speaker 2: In twenty thirteen, then CEO of Uber, Travis Kalanek, had

841
00:48:06,560 --> 00:48:09,360
Speaker 2: gotten a ride in one of Google's prototype driverless cars,

842
00:48:10,080 --> 00:48:13,160
Speaker 2: sitting in a taxi without a human driver. He'd understood

843
00:48:13,160 --> 00:48:15,600
Speaker 2: that this could mean the end of his company, and

844
00:48:15,640 --> 00:48:18,680
Speaker 2: so Uber had plunged headlong into the driverless car race.

845
00:48:19,280 --> 00:48:23,000
Speaker 2: The company hired nearly half of Carnegie Mailn's top Robotics lab,

846
00:48:23,680 --> 00:48:26,760
Speaker 2: and not long after we also know through court records

847
00:48:26,760 --> 00:48:30,600
Speaker 2: and emails that Uber also began communicating with Anthony Lewandowski,

848
00:48:31,080 --> 00:48:35,360
Speaker 2: who in twenty sixteen would leave Google, quitting just before

849
00:48:35,440 --> 00:48:37,680
Speaker 2: he could be fired for a recruiting team members away,

850
00:48:37,960 --> 00:48:43,680
Speaker 2: including Don Burnett. Anthony would then start his own autonomous

851
00:48:43,760 --> 00:48:47,320
Speaker 2: vehicle company. Uber would soon buy that company for almost

852
00:48:47,360 --> 00:48:50,799
Speaker 2: seven hundred million dollars, even though the company had no

853
00:48:50,920 --> 00:48:54,480
Speaker 2: product and was only months old, which raised a mystery.

854
00:48:55,200 --> 00:48:57,600
Speaker 2: Why would Uber pay so much for a company whose

855
00:48:57,719 --> 00:48:59,360
Speaker 2: only assets seemed to be its people.

856
00:49:00,160 --> 00:49:01,160
Speaker 4: This is where.

857
00:49:00,920 --> 00:49:04,879
Speaker 16: Google goes into its computer security logs and realizes that

858
00:49:05,000 --> 00:49:08,799
Speaker 16: not long before he left, Anthony Lewandowski downloaded something like

859
00:49:08,960 --> 00:49:12,520
Speaker 16: fourteen thousand technical files onto his.

860
00:49:12,520 --> 00:49:16,279
Speaker 4: Computer and moved them onto an external disc.

861
00:49:16,520 --> 00:49:19,040
Speaker 2: Obviously you can't do that. I mean, I'm assuming obviously

862
00:49:19,040 --> 00:49:19,680
Speaker 2: you can't do that.

863
00:49:19,840 --> 00:49:27,520
Speaker 4: No, you definitely cannot see And this is the kind

864
00:49:27,560 --> 00:49:29,759
Speaker 4: of thing that maybe if he had stayed there, this

865
00:49:29,800 --> 00:49:31,400
Speaker 4: is the kind of thing Anthea would have done, and

866
00:49:31,440 --> 00:49:32,799
Speaker 4: he would have been like, oh, it's just so I

867
00:49:32,800 --> 00:49:34,560
Speaker 4: could have access to to it somewhere else. Then he

868
00:49:34,560 --> 00:49:37,319
Speaker 4: probably would have gotten away with it. But when you

869
00:49:37,680 --> 00:49:41,279
Speaker 4: then go and work for Uber and start running their

870
00:49:41,400 --> 00:49:47,160
Speaker 4: direct competitor self driving car program, that's when you get

871
00:49:47,160 --> 00:49:51,759
Speaker 4: in trouble. And that's when what's technically called WEIMO. At

872
00:49:51,760 --> 00:49:57,600
Speaker 4: this point, Google's program sues UK and puts Anthony at

873
00:49:57,600 --> 00:50:01,480
Speaker 4: the center of an enormous legal battle between these.

874
00:50:01,360 --> 00:50:05,840
Speaker 15: Tech giants, secrets and subterfugia.

875
00:50:05,840 --> 00:50:09,680
Speaker 4: In Silicon Valley, a former Google engineer has been charged

876
00:50:09,719 --> 00:50:13,319
Speaker 4: with stealing files from Alphabet's self driving car project and

877
00:50:13,360 --> 00:50:14,239
Speaker 4: taking them to Uber.

878
00:50:14,600 --> 00:50:18,799
Speaker 11: Specifically, it involves a former lead engineer of Google's self

879
00:50:18,880 --> 00:50:21,480
Speaker 11: driving car unit, Anthony Lewandowski.

880
00:50:21,560 --> 00:50:24,160
Speaker 4: Now he's accused of using.

881
00:50:23,920 --> 00:50:27,120
Speaker 11: His personal laptop and downloading more than fourteen.

882
00:50:26,760 --> 00:50:29,799
Speaker 2: In twenty sixteen, Google had just spun its driverless car

883
00:50:29,880 --> 00:50:34,000
Speaker 2: unit into a new entity, Weimo. Weimo sued Uber. Uber

884
00:50:34,040 --> 00:50:36,120
Speaker 2: had to settled to the tune of two hundred and

885
00:50:36,160 --> 00:50:39,760
Speaker 2: forty five million dollars, and in a separate criminal trial,

886
00:50:40,120 --> 00:50:45,320
Speaker 2: Anthony Lewandowski pled guilty to stealing trade secrets. Afterwards, Uber

887
00:50:45,400 --> 00:50:48,920
Speaker 2: continues their driverless car program without him, continuing to pursue

888
00:50:48,920 --> 00:50:52,680
Speaker 2: its move fast, break things strategy, which in twenty eighteen

889
00:50:53,000 --> 00:50:55,400
Speaker 2: leads to the death of a woman named Elaine Herzberg.

890
00:50:55,640 --> 00:50:58,200
Speaker 8: Uber is sitting the brakes on its self driving cars

891
00:50:58,560 --> 00:51:01,160
Speaker 8: after one of them hit and kill the woman in Arizona.

892
00:51:01,560 --> 00:51:04,640
Speaker 1: The vehicle was in autonomous mode, but it did have

893
00:51:04,680 --> 00:51:06,759
Speaker 1: a safety driver on board, but.

894
00:51:06,800 --> 00:51:10,640
Speaker 17: A police report later indicating the safety driver was streaming

895
00:51:10,719 --> 00:51:13,759
Speaker 17: TV shows on her phone for three hours that night,

896
00:51:14,000 --> 00:51:15,840
Speaker 17: including at the time of the crash.

897
00:51:16,480 --> 00:51:19,319
Speaker 2: The way this story was reported, nearly everyone blamed the

898
00:51:19,320 --> 00:51:22,839
Speaker 2: safety driver. She was on her phone. She's streaming an episode.

899
00:51:22,480 --> 00:51:26,359
Speaker 17: Of the Voice Tempe investigator saying, had Vasquez been paying

900
00:51:26,440 --> 00:51:29,040
Speaker 17: attention to the road, she could have stopped the car

901
00:51:29,320 --> 00:51:33,440
Speaker 17: forty two feet before impact the NTSB slamming.

902
00:51:33,440 --> 00:51:37,240
Speaker 2: There were some important additional context, which is that Uber's

903
00:51:37,320 --> 00:51:40,560
Speaker 2: robot driver was also just much worse than way Moo's,

904
00:51:41,400 --> 00:51:44,440
Speaker 2: a statistic I found jaw dropping. At this point, Waymos's

905
00:51:44,440 --> 00:51:46,359
Speaker 2: safety drivers were having to take over from the car

906
00:51:46,600 --> 00:51:51,040
Speaker 2: once every five six hundred miles. Uber's safety drivers that

907
00:51:51,120 --> 00:51:54,320
Speaker 2: year had to intervene more than once every thirteen miles.

908
00:51:55,880 --> 00:51:59,759
Speaker 2: Despite that, five months before the crash, over employee objections,

909
00:52:00,120 --> 00:52:03,360
Speaker 2: Uber had cut its safety crews. Instead of two humans,

910
00:52:03,480 --> 00:52:08,040
Speaker 2: they just used one. One safety driver overseeing a robot

911
00:52:08,080 --> 00:52:10,880
Speaker 2: driver that was arguably not ready to be on public roads.

912
00:52:12,120 --> 00:52:15,000
Speaker 2: In the last moments of Alane Herzburg's life, the robot

913
00:52:15,040 --> 00:52:19,120
Speaker 2: spent an indefensible five point six seconds trying and failing

914
00:52:19,239 --> 00:52:20,960
Speaker 2: to guess the shape in the road there was a

915
00:52:21,040 --> 00:52:24,920
Speaker 2: human body pushing a bike. Over those five point six seconds,

916
00:52:25,040 --> 00:52:29,000
Speaker 2: the robot kept reclassifying our whishing an unknown object a

917
00:52:29,080 --> 00:52:33,600
Speaker 2: vehicle a bicycle. During that time, spent wondering the car

918
00:52:33,760 --> 00:52:37,760
Speaker 2: did not slow down. Soon after Elaine Hertzberg's death, Uber

919
00:52:37,800 --> 00:52:39,440
Speaker 2: halted its testing program.

920
00:52:39,760 --> 00:52:43,400
Speaker 17: Uber has temporarily suspended its driverless fleet nationwide, as the

921
00:52:43,480 --> 00:52:49,000
Speaker 17: NTSB police, Uber and the National Highway Traffic Safety Administration investigate.

922
00:52:49,560 --> 00:52:52,239
Speaker 2: We reached out to Uber for comment. A spokesperson said

923
00:52:52,280 --> 00:52:55,040
Speaker 2: that the fatal collision was indeed a tragedy which had

924
00:52:55,040 --> 00:52:59,360
Speaker 2: a significant impact on Uber and the entire industry. There'd

925
00:52:59,360 --> 00:53:02,520
Speaker 2: be other competitors who would shut down after similar accidents.

926
00:53:02,840 --> 00:53:05,319
Speaker 2: There would also be Tesla, which by twenty twenty was

927
00:53:05,440 --> 00:53:08,640
Speaker 2: publicly marketing a product of the company called full self driving,

928
00:53:09,080 --> 00:53:13,560
Speaker 2: but which absolutely was not. Meanwhile, Wimo had slowly continued

929
00:53:13,600 --> 00:53:16,640
Speaker 2: develop its tech. Their robotaxis would be ready for riders

930
00:53:16,640 --> 00:53:19,560
Speaker 2: by twenty twenty. The team had gotten an unexpected boost

931
00:53:19,600 --> 00:53:23,200
Speaker 2: from a technology that was at the time very little understood.

932
00:53:27,320 --> 00:53:30,839
Speaker 2: In twenty twenty six, when most people talk about artificial intelligence,

933
00:53:31,160 --> 00:53:34,440
Speaker 2: the conversation defaults to products like chat, GPT, and Claude,

934
00:53:35,200 --> 00:53:37,719
Speaker 2: But artificial intelligence has been a core part of driver

935
00:53:37,800 --> 00:53:41,400
Speaker 2: lest cars going back two decades. In the twenty ten's,

936
00:53:41,640 --> 00:53:43,840
Speaker 2: neural net advances meant that you can now begin to

937
00:53:43,920 --> 00:53:47,280
Speaker 2: feed a computer system large amounts of data and watch

938
00:53:47,280 --> 00:53:51,839
Speaker 2: as its perception, prediction, and decision making abilities improved. Here's

939
00:53:51,840 --> 00:53:52,720
Speaker 2: Sebastian Thront.

940
00:53:53,520 --> 00:53:56,480
Speaker 11: Their technology of massive data training was with us from

941
00:53:56,520 --> 00:53:58,680
Speaker 11: the get go, but has become more and more and

942
00:53:58,719 --> 00:54:02,319
Speaker 11: more and more important. The surprise for all of us

943
00:54:02,360 --> 00:54:06,240
Speaker 11: has been that size matters. When you put a million

944
00:54:06,239 --> 00:54:09,560
Speaker 11: documents into an AI, it's fine, one hundred million is fine,

945
00:54:10,120 --> 00:54:12,399
Speaker 11: And when you put one hundred billion documents into ANI,

946
00:54:12,800 --> 00:54:16,440
Speaker 11: it is umbiliately smart. And then a thing shocked everybody,

947
00:54:16,480 --> 00:54:17,160
Speaker 11: myself into.

948
00:54:20,000 --> 00:54:22,759
Speaker 2: The Google brand team. The deep learning people started working

949
00:54:22,840 --> 00:54:25,040
Speaker 2: with the driverless car team to use training data to

950
00:54:25,040 --> 00:54:27,839
Speaker 2: help the computer driver learn things like how to better

951
00:54:27,880 --> 00:54:30,240
Speaker 2: predict when another car was about to suddenly switch lanes,

952
00:54:30,400 --> 00:54:34,120
Speaker 2: how to more reliably spot pedestrians. Over the years, as

953
00:54:34,120 --> 00:54:36,839
Speaker 2: a car drove more miles, as the team gathered more data,

954
00:54:37,080 --> 00:54:39,920
Speaker 2: plugged that data into their AI systems, and tweaked those systems.

955
00:54:40,280 --> 00:54:43,839
Speaker 2: The engineers say the robot driver kept improving as they

956
00:54:43,880 --> 00:54:46,600
Speaker 2: tested the car in new weather conditions, they discovered problems

957
00:54:46,600 --> 00:54:50,080
Speaker 2: that required hardware fixes. For instance, in Phoenix, Weimo had

958
00:54:50,120 --> 00:54:52,960
Speaker 2: to design miniature wipers for their cars. Led our sensors

959
00:54:53,040 --> 00:54:55,880
Speaker 2: to deal with the dust storms and heavy rains. In

960
00:54:55,920 --> 00:54:58,800
Speaker 2: twenty twenty, Weaimo finally debuts to the public in Arizona.

961
00:54:59,440 --> 00:55:01,719
Speaker 2: In the years after, it'll roll out to ten more

962
00:55:01,760 --> 00:55:06,440
Speaker 2: American cities. A funny consequence of Weymo's long development cycle

963
00:55:06,880 --> 00:55:09,360
Speaker 2: is that the public's attitude towards Silicon Valley has just

964
00:55:09,400 --> 00:55:12,920
Speaker 2: really changed in that time. There's more suspicion towards Google

965
00:55:12,960 --> 00:55:14,880
Speaker 2: than there was back in two thousand and nine when

966
00:55:14,920 --> 00:55:18,600
Speaker 2: the project first started, And so now many people look

967
00:55:18,640 --> 00:55:21,440
Speaker 2: at the Waimo driver with a raised eyebrow with a

968
00:55:21,520 --> 00:55:28,040
Speaker 2: question immediately on their lips. Chapter five, Are you a

969
00:55:28,080 --> 00:55:28,560
Speaker 2: good driver?

970
00:55:29,000 --> 00:55:32,480
Speaker 4: All right? Autonomous vehicles can now get you around Atlanta yesterday.

971
00:55:33,040 --> 00:55:36,400
Speaker 4: Driving through Austin is here, except it comes without.

972
00:55:36,320 --> 00:55:40,680
Speaker 10: Drive Light hailing app is now taking passengers in Miami.

973
00:55:40,600 --> 00:55:46,280
Speaker 2: A fleet of white electric Jaguars covered in forty different sensors, cameras, radar, lidar.

974
00:55:46,800 --> 00:55:48,960
Speaker 2: It's an expensive car, as much as one hundred and

975
00:55:48,960 --> 00:55:52,520
Speaker 2: fifty thousand dollars by some estimates. In the news stories,

976
00:55:52,520 --> 00:55:55,319
Speaker 2: you see the inside where the human driver would normally sit.

977
00:55:55,480 --> 00:55:58,160
Speaker 2: There's an empty seat you're not allowed in with a

978
00:55:58,200 --> 00:56:01,200
Speaker 2: steering wheel in front of it. It turns itself.

979
00:56:01,600 --> 00:56:03,520
Speaker 10: Cars without drivers are here.

980
00:56:03,680 --> 00:56:05,759
Speaker 3: Yeah, it sounds like something out of the Jetsons.

981
00:56:05,800 --> 00:56:08,360
Speaker 4: But get ready because you may look over at the

982
00:56:08,360 --> 00:56:11,240
Speaker 4: car next to you and see it rolling down the street.

983
00:56:11,320 --> 00:56:14,200
Speaker 2: The TV newscasters always use the same g whiz tone.

984
00:56:14,480 --> 00:56:17,480
Speaker 2: They can never resist the Jetson's reference. In every city,

985
00:56:17,680 --> 00:56:21,200
Speaker 2: the influencers hop into record testimonials for their daily serving

986
00:56:21,239 --> 00:56:21,720
Speaker 2: of clout.

987
00:56:21,800 --> 00:56:23,800
Speaker 14: So in today's video, I'm about to take my first

988
00:56:23,840 --> 00:56:25,120
Speaker 14: ever driverless car.

989
00:56:25,280 --> 00:56:26,600
Speaker 8: It's with an app called Weimo.

990
00:56:26,920 --> 00:56:32,480
Speaker 3: Weimo is basically driverless car uber where it's like ride service.

991
00:56:32,600 --> 00:56:34,799
Speaker 3: You call it going wherever you need it to go,

992
00:56:35,120 --> 00:56:36,240
Speaker 3: but there's no driver.

993
00:56:36,480 --> 00:56:37,640
Speaker 2: You guys, this is creepy.

994
00:56:37,920 --> 00:56:40,320
Speaker 4: It's like I'm being driven around by a ghost person.

995
00:56:40,560 --> 00:56:41,600
Speaker 4: It's a little terrifying.

996
00:56:41,800 --> 00:56:46,320
Speaker 2: It is definitely Romo taxis pull hilariously badly. According to

997
00:56:46,480 --> 00:56:49,480
Speaker 2: JD Power, a data analytics firm, among people who've not

998
00:56:49,600 --> 00:56:53,719
Speaker 2: ridden in one consumer confidence is at twenty percent, but

999
00:56:54,239 --> 00:56:57,319
Speaker 2: among people who have taken a ride Denver shoots up

1000
00:56:57,360 --> 00:57:01,120
Speaker 2: to seventy six percent. It's the thing that capture this story.

1001
00:57:01,440 --> 00:57:03,279
Speaker 2: But when I sad and won a couple of years ago.

1002
00:57:03,719 --> 00:57:06,279
Speaker 2: I just found it persuasive as an experience.

1003
00:57:06,560 --> 00:57:08,719
Speaker 1: You know what, I'm not as nervous as I thought

1004
00:57:08,760 --> 00:57:09,359
Speaker 1: I was gonna be.

1005
00:57:09,600 --> 00:57:11,600
Speaker 9: This is actually quite relaxing.

1006
00:57:11,360 --> 00:57:13,240
Speaker 2: Nice gradual turn, felt very safe.

1007
00:57:13,440 --> 00:57:15,520
Speaker 18: You know, it was kind of freaky at first, but

1008
00:57:15,640 --> 00:57:17,520
Speaker 18: now it's pretty chill smooth.

1009
00:57:17,320 --> 00:57:19,600
Speaker 4: Right though it wasn't driving fast, it wasn't jerking.

1010
00:57:20,000 --> 00:57:22,760
Speaker 18: It's driving like you always hope your Uber driver would.

1011
00:57:22,880 --> 00:57:24,280
Speaker 4: So I guess that's one of the big sells.

1012
00:57:24,360 --> 00:57:27,320
Speaker 2: Chris Arms and that methodical team leader had left Google

1013
00:57:27,400 --> 00:57:29,800
Speaker 2: years ago, but he told me about his experience as

1014
00:57:29,800 --> 00:57:32,720
Speaker 2: a civilian consumer trying away mom out in the world.

1015
00:57:33,440 --> 00:57:37,080
Speaker 8: My universal experience has been and you can tell me

1016
00:57:37,120 --> 00:57:39,880
Speaker 8: if this was your experience. The first couple of minutes

1017
00:57:39,880 --> 00:57:44,560
Speaker 8: in the vehicle, it's huh, that's crazy. I dished nobody

1018
00:57:44,600 --> 00:57:48,680
Speaker 8: behind the wheel swinging with sharks. And then a few

1019
00:57:48,680 --> 00:57:52,520
Speaker 8: minutes in and it's like, okay, you know, it's just

1020
00:57:52,520 --> 00:57:53,200
Speaker 8: just gonna drive.

1021
00:57:53,360 --> 00:57:54,160
Speaker 4: Is that all it does?

1022
00:57:54,320 --> 00:57:56,720
Speaker 8: And then you know, ten minutes and people are looking

1023
00:57:56,720 --> 00:57:57,240
Speaker 8: at their phone.

1024
00:57:58,360 --> 00:58:01,520
Speaker 2: People tend to feel safe in these but are they

1025
00:58:01,920 --> 00:58:05,200
Speaker 2: actually so we know that the Weimo driver has now

1026
00:58:05,280 --> 00:58:08,760
Speaker 2: driven over two hundred million real world miles, and they

1027
00:58:08,840 --> 00:58:11,600
Speaker 2: release safety data so far for the first one hundred

1028
00:58:11,640 --> 00:58:16,040
Speaker 2: and twenty seven million miles. Weymo's fairly transparent. They release

1029
00:58:16,080 --> 00:58:20,000
Speaker 2: their crash and safety data unredacted to the public. By contrast,

1030
00:58:20,160 --> 00:58:23,240
Speaker 2: Tesla redacts the details of its crashes. The company says

1031
00:58:23,240 --> 00:58:27,560
Speaker 2: they are confidential business information. In Weymo's case, I've looked

1032
00:58:27,600 --> 00:58:29,920
Speaker 2: at the data, I've looked at how the company interprets it,

1033
00:58:30,120 --> 00:58:33,880
Speaker 2: how skeptical independent researchers interpret it. I wanted to walk

1034
00:58:33,920 --> 00:58:37,400
Speaker 2: through it with an autonomous vehicle reporter I trust. His

1035
00:58:37,520 --> 00:58:40,840
Speaker 2: name is Timothy Beeley, author of the newsletter Understanding AI.

1036
00:58:41,440 --> 00:58:43,560
Speaker 2: I asked him how much our picture of the Weymos

1037
00:58:43,680 --> 00:58:45,080
Speaker 2: safety data has been evolving.

1038
00:58:45,680 --> 00:58:47,880
Speaker 19: So it's been pretty consistent the last couple of years.

1039
00:58:48,000 --> 00:58:50,960
Speaker 19: They are scaling up, and so all the numbers get bigger,

1040
00:58:50,960 --> 00:58:52,600
Speaker 19: like the total number of miles get bigger, the number

1041
00:58:52,600 --> 00:58:55,000
Speaker 19: of crashes get bigger, but the light crashes per mile

1042
00:58:55,240 --> 00:58:58,160
Speaker 19: have not changed a ton, Weimos says, and I think

1043
00:58:58,160 --> 00:59:01,160
Speaker 19: this is correct, that it's roughly eighty brass safer in

1044
00:59:01,240 --> 00:59:04,480
Speaker 19: terms of crashes are severe enough to turn down an airbag.

1045
00:59:04,840 --> 00:59:08,520
Speaker 19: Crashes severe enough to cause an injury, and also crashes

1046
00:59:08,560 --> 00:59:13,720
Speaker 19: involving vulnerable road users like pedestrians or bicyclists.

1047
00:59:16,640 --> 00:59:19,360
Speaker 2: So eighty percent fewer air bag crashes than human drivers,

1048
00:59:19,360 --> 00:59:22,520
Speaker 2: and actually ninety percent fewer crashes that cause a serious injury.

1049
00:59:23,200 --> 00:59:26,600
Speaker 2: Some independent experts have small quibbles with the methodology, but

1050
00:59:26,680 --> 00:59:32,080
Speaker 2: broadly they find Waymos's data credible. Timothy pointed out, there's

1051
00:59:32,080 --> 00:59:36,040
Speaker 2: one very important thing we don't know, the fatal crash comparison.

1052
00:59:37,080 --> 00:59:39,640
Speaker 2: For every one hundred million miles humans drive, we cause

1053
00:59:39,680 --> 00:59:43,240
Speaker 2: a little over one fatal crash. The Waimo driver has

1054
00:59:43,320 --> 00:59:45,960
Speaker 2: driven two hundred million miles without causing a fatal crash,

1055
00:59:46,440 --> 00:59:50,240
Speaker 2: but statistically speaking, that could still be a fluke. Some

1056
00:59:50,400 --> 00:59:54,080
Speaker 2: academics have suggested we need about three hundred million miles

1057
00:59:54,120 --> 00:59:58,320
Speaker 2: to have statistical confidence in the hundreds of millions of

1058
00:59:58,320 --> 01:00:01,240
Speaker 2: miles the Waymo driver has traveled. It was involved in

1059
01:00:01,320 --> 01:00:03,960
Speaker 2: two fatal crashes which it did not appear to cause.

1060
01:00:04,520 --> 01:00:07,600
Speaker 2: Here are the details of those crashes. In one, a

1061
01:00:07,640 --> 01:00:10,560
Speaker 2: speeding human driver rear ended a line of vehicles at

1062
01:00:10,560 --> 01:00:12,960
Speaker 2: a stoplight. There's an empty Weimo in the line of

1063
01:00:12,960 --> 01:00:16,840
Speaker 2: struck cars. In another crash, a Weimo is yielding for

1064
01:00:16,880 --> 01:00:20,120
Speaker 2: a pedestrian. It was rear ended by a motorcycle. The

1065
01:00:20,200 --> 01:00:23,840
Speaker 2: motorcycle driver was then struck by a second car. That's

1066
01:00:23,880 --> 01:00:28,400
Speaker 2: everything when Timothy bee Lee looks at the entire safety picture,

1067
01:00:28,800 --> 01:00:31,720
Speaker 2: the results we have so far from this big experiment

1068
01:00:31,800 --> 01:00:35,480
Speaker 2: Weimo is conducting on American roads, what he sees is

1069
01:00:35,520 --> 01:00:36,400
Speaker 2: mainly promising.

1070
01:00:37,120 --> 01:00:39,720
Speaker 19: So far it's been better than human drivers, and so far,

1071
01:00:39,760 --> 01:00:41,720
Speaker 19: I think the case for allowing them they continue.

1072
01:00:41,720 --> 01:00:43,560
Speaker 4: The experiment is very strong.

1073
01:00:44,320 --> 01:00:47,280
Speaker 2: Which doesn't mean we shouldn't scrutinize this Weimo experiment as

1074
01:00:47,280 --> 01:00:50,480
Speaker 2: it continues. I find myself paying a lot of attention

1075
01:00:50,560 --> 01:00:54,560
Speaker 2: to Weimo crashes, which isn't hard. They make headlines. The

1076
01:00:54,600 --> 01:00:56,800
Speaker 2: most harrowing one recently was this January.

1077
01:00:57,200 --> 01:00:59,600
Speaker 6: A child at near to Elementary school in Santa Monica

1078
01:00:59,680 --> 01:01:00,800
Speaker 6: is Weymo.

1079
01:01:01,000 --> 01:01:03,320
Speaker 18: A child ran across the street from behind a double

1080
01:01:03,360 --> 01:01:05,200
Speaker 18: part car and a Weimo hit the kid.

1081
01:01:05,760 --> 01:01:08,200
Speaker 13: Santa Monica police say the child, a ten year old girl,

1082
01:01:08,280 --> 01:01:08,840
Speaker 13: was not hurt.

1083
01:01:09,520 --> 01:01:12,480
Speaker 2: The company issued a statement. Weimo said its driver had

1084
01:01:12,520 --> 01:01:16,160
Speaker 2: breaked hard, reducing speed from seventeen to under six miles

1085
01:01:16,200 --> 01:01:19,200
Speaker 2: per hour, a faster reaction, they claimed than a human

1086
01:01:19,320 --> 01:01:22,640
Speaker 2: driver would have been capable of what happened next at

1087
01:01:22,640 --> 01:01:25,600
Speaker 2: the accident scene. Actually answers a question i'd had, what

1088
01:01:25,640 --> 01:01:28,160
Speaker 2: does a WEIMO do after a car crash. Since there's

1089
01:01:28,160 --> 01:01:31,720
Speaker 2: no human driver to help, WEIMO employs what they call

1090
01:01:31,920 --> 01:01:35,520
Speaker 2: human fleet response agents, human beings who can't remotely drive

1091
01:01:35,520 --> 01:01:37,720
Speaker 2: the cars, but who the car can ask questions to

1092
01:01:37,800 --> 01:01:41,160
Speaker 2: if it gets confused. In Santa Monica, the WEIMO called

1093
01:01:41,200 --> 01:01:43,440
Speaker 2: one of those humans, the human called nine one one.

1094
01:01:44,040 --> 01:01:47,280
Speaker 2: And this is the strangest part of Weymo's statement. Apparently

1095
01:01:47,280 --> 01:01:49,560
Speaker 2: the car then waited at the scene of the accident

1096
01:01:49,640 --> 01:01:52,800
Speaker 2: until the police dismissed it. That's what we know so far.

1097
01:01:52,920 --> 01:01:56,000
Speaker 2: But there's two federal agencies investigating this crash, and so

1098
01:01:56,040 --> 01:01:59,360
Speaker 2: we'll have a full report in the future. One problem

1099
01:01:59,440 --> 01:02:01,680
Speaker 2: that's not really captured in the safety data that I've

1100
01:02:01,680 --> 01:02:04,880
Speaker 2: seen is what i'd call troubling edge cases. You see

1101
01:02:04,880 --> 01:02:07,800
Speaker 2: them in videos on social media. A WAIMO gets stuck

1102
01:02:07,840 --> 01:02:10,480
Speaker 2: at a dead stop light or blocks an emergency vehicle,

1103
01:02:11,040 --> 01:02:14,680
Speaker 2: or an example, Timothy gave waymo's were driving past stopped

1104
01:02:14,680 --> 01:02:16,000
Speaker 2: school buses in Austin.

1105
01:02:16,280 --> 01:02:18,080
Speaker 19: I think it's reasonable to say this is like a

1106
01:02:18,080 --> 01:02:20,880
Speaker 19: clear cut rule that the vehicle should follow this role.

1107
01:02:21,040 --> 01:02:23,080
Speaker 19: These educads are still very rare, and so if it's

1108
01:02:23,080 --> 01:02:25,080
Speaker 19: a one to ten million thing, I think it's not

1109
01:02:25,200 --> 01:02:27,040
Speaker 19: that big a deal as long as they are making progress,

1110
01:02:27,040 --> 01:02:28,520
Speaker 19: which for most of these I think they are.

1111
01:02:29,040 --> 01:02:31,520
Speaker 2: Timothy pointed to one area where Waymo's not been as

1112
01:02:31,520 --> 01:02:34,760
Speaker 2: transparent as he'd like, those human response agents, some of

1113
01:02:34,760 --> 01:02:37,760
Speaker 2: which are based here some of the Philippines. There's questions

1114
01:02:37,800 --> 01:02:40,840
Speaker 2: about what specifically they do and about how this will

1115
01:02:40,840 --> 01:02:43,680
Speaker 2: all work as way most scales up. We asked Waymo

1116
01:02:43,720 --> 01:02:46,160
Speaker 2: for comment on everything you heard in this episode, especially

1117
01:02:46,280 --> 01:02:49,439
Speaker 2: the recent safety incidents. A spokesperson said that the data

1118
01:02:49,560 --> 01:02:52,200
Speaker 2: to date indicates that the Weimo driver is already making

1119
01:02:52,320 --> 01:02:54,640
Speaker 2: roads safer in the places where they operate, and says

1120
01:02:54,640 --> 01:02:57,200
Speaker 2: that Weymo can used to work with policymakers and regulators

1121
01:02:57,200 --> 01:03:02,640
Speaker 2: to improve its technology. That's the safety picture so far,

1122
01:03:02,760 --> 01:03:05,280
Speaker 2: which to me, after many months of looking at this

1123
01:03:05,480 --> 01:03:09,680
Speaker 2: and talking to experts, looks pretty good. As Weimo continues

1124
01:03:09,680 --> 01:03:12,360
Speaker 2: its rollout, other companies are quickly following behind.

1125
01:03:12,840 --> 01:03:16,960
Speaker 3: Amazon's new driverless taxi is launching in Las Vegas this summer,

1126
01:03:17,080 --> 01:03:18,680
Speaker 3: and it's expected to arrive.

1127
01:03:18,440 --> 01:03:21,400
Speaker 2: And now there's other robo taxi companies like Amazon, Zookes.

1128
01:03:21,800 --> 01:03:24,360
Speaker 2: Uber is back in the mix, not making technology, but

1129
01:03:24,520 --> 01:03:27,760
Speaker 2: partnering with these robo taxi companies. We Ride recently struck

1130
01:03:27,840 --> 01:03:30,960
Speaker 2: partnership with Uber to bring its avs to Abu Dhabi,

1131
01:03:31,320 --> 01:03:33,520
Speaker 2: another sign of it. And many of those early WEIMO

1132
01:03:33,600 --> 01:03:38,000
Speaker 2: engineers are now CEOs of autonomous companies themselves. Dmitri Dolgov

1133
01:03:38,160 --> 01:03:41,520
Speaker 2: is actually co CEO Weimo, but other team members run

1134
01:03:41,600 --> 01:03:43,120
Speaker 2: driverless trucking companies.

1135
01:03:43,240 --> 01:03:46,240
Speaker 18: Got Don Burnette, founder and CEO of kodiak Ai. Don,

1136
01:03:46,320 --> 01:03:47,480
Speaker 18: thank you so much for joining us.

1137
01:03:47,480 --> 01:03:48,520
Speaker 12: It's good to see you again.

1138
01:03:48,880 --> 01:03:51,440
Speaker 2: Don Burnett is head of kodiak Ai, which has its

1139
01:03:51,480 --> 01:03:54,400
Speaker 2: technology deployed in driverless trucks in the premium basin.

1140
01:03:54,600 --> 01:03:59,280
Speaker 18: Please welcome CEO of Aurora, Chris Ermthin.

1141
01:03:59,040 --> 01:04:00,000
Speaker 4: A big round of a plot.

1142
01:04:00,600 --> 01:04:04,000
Speaker 2: Chris Armsen now heads Aurora, which currently has semi trucks

1143
01:04:04,000 --> 01:04:07,760
Speaker 2: on Texas highways. And my personal favorite plot development which

1144
01:04:07,800 --> 01:04:08,720
Speaker 2: just emerged this week.

1145
01:04:08,880 --> 01:04:12,320
Speaker 18: I just broke on the information that Uber founder Travis

1146
01:04:12,400 --> 01:04:15,720
Speaker 18: Kalanik is starting a new self driving car company with

1147
01:04:15,880 --> 01:04:20,880
Speaker 18: financial backing from Uber and in partnership with Anthony Lewandowski.

1148
01:04:21,440 --> 01:04:22,680
Speaker 17: Now, for those who've been they.

1149
01:04:22,560 --> 01:04:25,640
Speaker 2: Say there's no second acts in American lives. Somehow, both

1150
01:04:25,640 --> 01:04:28,160
Speaker 2: of these men seem to be on their fourth. The

1151
01:04:28,200 --> 01:04:31,280
Speaker 2: big picture, though, is that everywhere in America today that

1152
01:04:31,320 --> 01:04:35,680
Speaker 2: you see a driver, taxi, truck, food delivery, there are

1153
01:04:35,760 --> 01:04:39,400
Speaker 2: several companies working on the robot version trying their best

1154
01:04:39,400 --> 01:04:41,880
Speaker 2: to make driver as a job start to go the

1155
01:04:41,880 --> 01:04:46,600
Speaker 2: way of the knocker Upper of the Lamplighter. Those knocker Ruppers,

1156
01:04:46,680 --> 01:04:50,760
Speaker 2: by the way, they disappeared quietly. The Lamplighters did not.

1157
01:04:52,920 --> 01:04:56,080
Speaker 2: Writer Carl Benedict Frey tells the story of the Lamplighters Union,

1158
01:04:56,480 --> 01:04:59,320
Speaker 2: how their strikes plunged New York City briefly into darkness

1159
01:04:59,600 --> 01:05:03,520
Speaker 2: to the light of lovers and thieves. In Vervier, Belgium,

1160
01:05:03,720 --> 01:05:07,120
Speaker 2: the Lamplighters strikes turned violent, ending in an attack on

1161
01:05:07,160 --> 01:05:10,760
Speaker 2: the local police headquarters. The army was brought in. The

1162
01:05:10,840 --> 01:05:13,400
Speaker 2: lamp Layers lost their fight, in part just because they

1163
01:05:13,400 --> 01:05:16,720
Speaker 2: were so outnumbered. But the drivers today fighting to save

1164
01:05:16,760 --> 01:05:19,360
Speaker 2: their livelihoods are a significantly bigger force.

1165
01:05:19,920 --> 01:05:24,600
Speaker 8: Please stand up, everybody that's ride share union members are

1166
01:05:24,640 --> 01:05:26,120
Speaker 8: someone who drives the vehicle.

1167
01:05:27,480 --> 01:05:28,040
Speaker 4: Stand up.

1168
01:05:29,200 --> 01:05:31,920
Speaker 2: Four point eight million Americans drive for a living. It's

1169
01:05:31,920 --> 01:05:34,400
Speaker 2: one of the most common jobs we have, and these

1170
01:05:34,440 --> 01:05:38,040
Speaker 2: workers do not plan to surrender to the California tech companies.

1171
01:05:38,240 --> 01:05:42,000
Speaker 2: They're doing this because they stand to make an unfathomable

1172
01:05:42,320 --> 01:05:45,640
Speaker 2: amount of money if they eliminate driving jobs for working

1173
01:05:45,680 --> 01:05:46,440
Speaker 2: class of people.

1174
01:05:46,800 --> 01:05:51,240
Speaker 4: I understand they this a business, they this capitalism, but

1175
01:05:51,480 --> 01:05:55,040
Speaker 4: not in my city at the expense of our jobs.

1176
01:05:55,320 --> 01:05:59,120
Speaker 2: These drivers are represented by unions backed by politicians and

1177
01:05:59,120 --> 01:06:03,080
Speaker 2: in cities across America blue cities. They're organizing. So far

1178
01:06:03,760 --> 01:06:07,880
Speaker 2: they're winning. Humans drive the city, lot machines, labor drives

1179
01:06:07,880 --> 01:06:10,040
Speaker 2: this city, keep the workers in the workforce.

1180
01:06:10,560 --> 01:06:13,439
Speaker 4: If it works in another city, great, have fun, not here,

1181
01:06:13,760 --> 01:06:14,360
Speaker 4: not Boston.

1182
01:06:14,880 --> 01:06:15,200
Speaker 9: Thank you.

1183
01:06:20,640 --> 01:06:24,160
Speaker 2: Next week the Fight to save a Job, to save

1184
01:06:24,200 --> 01:06:24,960
Speaker 2: the human Driver.

1185
01:06:25,960 --> 01:06:40,400
Speaker 5: Don't miss this one.

1186
01:06:41,480 --> 01:06:43,400
Speaker 2: Thank you for listening to our episode. I just want

1187
01:06:43,440 --> 01:06:46,240
Speaker 2: to say, making deeply reported stories like this one is

1188
01:06:46,360 --> 01:06:49,960
Speaker 2: only possible because for our listeners, particularly our premium subscribers

1189
01:06:49,960 --> 01:06:52,800
Speaker 2: who pay to support the show. We are releasing our

1190
01:06:52,880 --> 01:06:55,440
Speaker 2: full interview with Sebastian Throne, who used to lead Google

1191
01:06:55,600 --> 01:06:59,320
Speaker 2: X their secret Special Projects Lab. Totally fascinating conversation with

1192
01:06:59,360 --> 01:07:01,520
Speaker 2: the kind of person who just sort of lives in

1193
01:07:01,560 --> 01:07:03,720
Speaker 2: the future and has a million strange ideas about it.

1194
01:07:04,200 --> 01:07:07,240
Speaker 2: We are releasing that for our incognito mode members only.

1195
01:07:07,360 --> 01:07:09,360
Speaker 2: It'll be in your feed. If you would like to

1196
01:07:09,400 --> 01:07:11,400
Speaker 2: know the future, sign up at search Engine Dot Show

1197
01:07:11,680 --> 01:07:15,640
Speaker 2: and again. Your membership specifically enables projects like this one,

1198
01:07:15,840 --> 01:07:20,440
Speaker 2: so thank you. Search Engine is a presentation of Odyssey.

1199
01:07:20,680 --> 01:07:23,160
Speaker 2: It is created by me PJ Vote and Truthy Pinaminini.

1200
01:07:23,640 --> 01:07:26,280
Speaker 2: Garrett Graham is our senior producer. Emily Malterre is our

1201
01:07:26,320 --> 01:07:30,880
Speaker 2: associate producer. Theme, original composition and mixing by armand Bazarian.

1202
01:07:31,240 --> 01:07:34,360
Speaker 2: Our production intern is Piper Dumont. This episode was fact

1203
01:07:34,400 --> 01:07:38,120
Speaker 2: checked by Mary Mathis. Our executive producer is Lea Reese Dennis.

1204
01:07:38,240 --> 01:07:40,240
Speaker 2: Thanks to the rest of the team at Odyssey, Rob

1205
01:07:40,280 --> 01:07:43,520
Speaker 2: Mirandy Craig Cox, Eric Donnelly, Colin Gaynor, Mark Curran, just

1206
01:07:43,520 --> 01:07:47,280
Speaker 2: Fina Francis, Kurt Courtney, and Hillary Scheff. Thanks for listening.

1207
01:07:47,680 --> 01:07:49,280
Speaker 2: We'll see you next week with the second part of

1208
01:07:49,280 --> 01:07:50,960
Speaker 2: this story.

1209
01:07:54,240 --> 01:07:56,600
Speaker 1: That was part one of this two part story on

1210
01:07:56,600 --> 01:08:00,120
Speaker 1: one of the most transformative technologies of today, driverless car.

1211
01:08:00,400 --> 01:08:02,280
Speaker 1: If you want to hear how much more complicated the

1212
01:08:02,320 --> 01:08:05,680
Speaker 1: story gets as the technology rolls into American cities you

1213
01:08:05,720 --> 01:08:08,520
Speaker 1: can find search Engine wherever you get your podcasts.