1
00:00:15,316 --> 00:00:22,316
Speaker 1: Pushkin. It seems like we've been hearing for I don't

2
00:00:22,316 --> 00:00:25,956
Speaker 1: know ten years now that self driving cars are two

3
00:00:26,036 --> 00:00:29,756
Speaker 1: years away. And yes, I know Tesla's are amazing because

4
00:00:29,796 --> 00:00:33,076
Speaker 1: Tesla owners keep telling me so. But we are clearly

5
00:00:33,396 --> 00:00:37,516
Speaker 1: not yet at that magic transformative moment when cars can

6
00:00:37,636 --> 00:00:41,556
Speaker 1: truly drive themselves. We're not yet in a world where

7
00:00:41,716 --> 00:00:44,076
Speaker 1: a human driver seems as out of place as a

8
00:00:44,476 --> 00:00:48,796
Speaker 1: human elevator operator. So the question for today's show is this,

9
00:00:49,716 --> 00:00:52,596
Speaker 1: what is the most important problem that we have to

10
00:00:52,676 --> 00:00:55,556
Speaker 1: solve for that to happen. What's it going to take

11
00:00:55,556 --> 00:00:57,796
Speaker 1: to get to a world where I'm heading north on

12
00:00:57,916 --> 00:01:00,356
Speaker 1: I five and I just climb in the back of

13
00:01:00,356 --> 00:01:05,276
Speaker 1: my car and take a nap. I'm Jacob Goldstein, and

14
00:01:05,356 --> 00:01:07,476
Speaker 1: this is What's Your Problem, the show where we talk

15
00:01:07,556 --> 00:01:10,556
Speaker 1: to entrepreneurs and engineers about the future they're going to

16
00:01:10,636 --> 00:01:14,036
Speaker 1: build once they solve a few problems. My guest today

17
00:01:14,196 --> 00:01:17,836
Speaker 1: is Aisha Evans, CEO of the autonomous car company Zooks

18
00:01:18,236 --> 00:01:23,036
Speaker 1: zoo X. Zooks was acquired by Amazon back in twenty twenty,

19
00:01:23,196 --> 00:01:25,436
Speaker 1: and I wanted to talk to Aisha because Zooks is

20
00:01:25,516 --> 00:01:29,956
Speaker 1: really all in on this self driving car dream. They're

21
00:01:29,956 --> 00:01:32,636
Speaker 1: not going for half measures. They're not going for baby steps.

22
00:01:32,956 --> 00:01:36,996
Speaker 1: They're trying to build a truly fully AI take the

23
00:01:36,996 --> 00:01:40,116
Speaker 1: wheel self driving car. So I should seem like the

24
00:01:40,156 --> 00:01:43,356
Speaker 1: perfect person to talk about the problems engineers are going

25
00:01:43,396 --> 00:01:45,916
Speaker 1: to have to solve to build a car that can

26
00:01:45,956 --> 00:01:50,996
Speaker 1: truly drive itself. We started our conversation talking about the

27
00:01:51,036 --> 00:01:54,316
Speaker 1: self driving car that Zoos is building now, which really

28
00:01:54,636 --> 00:01:58,796
Speaker 1: isn't exactly a car. It looks like a toaster on wheels.

29
00:01:58,796 --> 00:02:02,516
Speaker 1: It's very boxy. There are sliding doors on either side,

30
00:02:02,836 --> 00:02:06,276
Speaker 1: and inside there are two bench seats facing each other,

31
00:02:06,396 --> 00:02:09,996
Speaker 1: and that's kind of it. There's no driver seat, there's

32
00:02:10,036 --> 00:02:15,196
Speaker 1: no dashboard, there's no steering wheel. If you step into

33
00:02:15,196 --> 00:02:19,116
Speaker 1: our vehicle and you think about driving, we consider that

34
00:02:19,156 --> 00:02:23,556
Speaker 1: we failed what transporting you no pedals, no steering wheel,

35
00:02:23,596 --> 00:02:25,956
Speaker 1: You are not involved in the driving. You're a rider,

36
00:02:25,996 --> 00:02:29,236
Speaker 1: and we conceived it for you. There's also no front

37
00:02:29,276 --> 00:02:32,956
Speaker 1: or back right, no fully symmetrical. When you think again

38
00:02:33,436 --> 00:02:37,276
Speaker 1: about the effectiveness and the efficiency of transportation, especially in

39
00:02:37,396 --> 00:02:41,276
Speaker 1: dense urban environment, imagine pulling into a narrow alley and

40
00:02:41,476 --> 00:02:43,516
Speaker 1: not having to make a U turn and just flipping

41
00:02:43,556 --> 00:02:46,476
Speaker 1: the lights and the vehicle goes into the other direction.

42
00:02:46,676 --> 00:02:48,676
Speaker 1: So the thing I like about it, not having a

43
00:02:48,716 --> 00:02:51,796
Speaker 1: steering wheel, not having a front or back is just

44
00:02:51,876 --> 00:02:56,556
Speaker 1: the way the thing looks. The physical thing itself is

45
00:02:56,636 --> 00:03:00,116
Speaker 1: this manifestation of how all in you are. Right. It's

46
00:03:00,156 --> 00:03:02,476
Speaker 1: not like, let's take a car and make a robot

47
00:03:02,556 --> 00:03:05,116
Speaker 1: drive it. It's what does the world look like? What

48
00:03:05,156 --> 00:03:08,356
Speaker 1: does a thing look like if there is no driver? Yes,

49
00:03:08,876 --> 00:03:12,836
Speaker 1: if you you know, you become dispassionate no matter what

50
00:03:12,916 --> 00:03:14,996
Speaker 1: car you have today, a personal car, and you look

51
00:03:15,036 --> 00:03:19,676
Speaker 1: at it purely from an engineering standpoint. It's architected and

52
00:03:19,916 --> 00:03:24,316
Speaker 1: designed with the concept of a human driver. Yea. What

53
00:03:24,316 --> 00:03:28,556
Speaker 1: we're doing, by the way, is really okay. If AI

54
00:03:28,756 --> 00:03:32,556
Speaker 1: is going to be responsible for the driving amongst other

55
00:03:32,716 --> 00:03:37,716
Speaker 1: human drivers, how do you architect and design the vehicle

56
00:03:38,476 --> 00:03:42,756
Speaker 1: so that AI is the best and safest drival possible.

57
00:03:43,076 --> 00:03:45,476
Speaker 1: That's our point of view, and then you work backward

58
00:03:45,516 --> 00:03:51,316
Speaker 1: from there. So so how's it going to work? Basically,

59
00:03:51,996 --> 00:03:53,716
Speaker 1: you have the app, you say I want to go

60
00:03:53,916 --> 00:03:55,716
Speaker 1: from here to there. We come, we pick you up,

61
00:03:55,756 --> 00:03:58,516
Speaker 1: you sit down, you buckle up, push start. That's a

62
00:03:58,556 --> 00:04:01,356
Speaker 1: safety thing, and then we take you to point B

63
00:04:02,116 --> 00:04:05,156
Speaker 1: and your unbuckle, step off, and by that time we

64
00:04:05,276 --> 00:04:09,396
Speaker 1: probably already know the other passenger. When can I call

65
00:04:09,436 --> 00:04:13,636
Speaker 1: one depends on where you are. So if you're in

66
00:04:13,716 --> 00:04:17,516
Speaker 1: Las Vegas fairly soon, much sooner than you think on

67
00:04:17,596 --> 00:04:19,796
Speaker 1: the strip, you'll be able to call one. Does that

68
00:04:19,876 --> 00:04:23,276
Speaker 1: mean this year? Next year? It won't be this year.

69
00:04:24,716 --> 00:04:28,956
Speaker 1: I can tell you what happened by twenty twenty five. Yes,

70
00:04:29,836 --> 00:04:32,716
Speaker 1: And once we do that, then we'll go city by city.

71
00:04:32,756 --> 00:04:36,396
Speaker 1: We've already been public saying that we'll go Las Vegas

72
00:04:36,556 --> 00:04:40,436
Speaker 1: and then San Francisco. Then as we want to move east,

73
00:04:40,956 --> 00:04:43,396
Speaker 1: but then as you're moving east you have to handle snow.

74
00:04:43,716 --> 00:04:46,036
Speaker 1: And then we want to be global, and as you

75
00:04:46,156 --> 00:04:48,956
Speaker 1: go global, you get a whole new set of parameters.

76
00:04:49,076 --> 00:04:53,156
Speaker 1: The roundabouts in London come to mind, and so is

77
00:04:53,196 --> 00:04:57,876
Speaker 1: the idea that zoos vehicles will be like cabs are. Now,

78
00:04:58,036 --> 00:05:00,916
Speaker 1: like I have a car, and then if I'm in

79
00:05:00,956 --> 00:05:02,676
Speaker 1: the city and I need to get from one place

80
00:05:02,716 --> 00:05:05,676
Speaker 1: to another, maybe I'll take a cab. Is it like that?

81
00:05:05,956 --> 00:05:09,476
Speaker 1: It's like that with a little bit deeper philosophy, which is, look,

82
00:05:10,356 --> 00:05:13,636
Speaker 1: we know that let's not with the United States, right,

83
00:05:13,996 --> 00:05:16,876
Speaker 1: it's around two and a half cars roughly per family.

84
00:05:17,356 --> 00:05:19,916
Speaker 1: We're not saying it'll be zero overnight, obviously you have

85
00:05:19,956 --> 00:05:22,636
Speaker 1: to be rational, but we can do better than two

86
00:05:22,676 --> 00:05:25,436
Speaker 1: and a half cars per family. So let's say we

87
00:05:25,516 --> 00:05:27,676
Speaker 1: get to one or one and a half. We feel

88
00:05:27,716 --> 00:05:31,476
Speaker 1: that we will have done something after the break. The

89
00:05:31,516 --> 00:05:34,476
Speaker 1: technical problems that Zoos and for that matter, all the

90
00:05:34,516 --> 00:05:37,636
Speaker 1: other self driving car companies are still trying to fix.

91
00:05:38,276 --> 00:05:41,356
Speaker 1: One surprisingly hard one, how to figure out who goes

92
00:05:41,396 --> 00:05:44,156
Speaker 1: first when two cars pull up to an intersection at

93
00:05:44,156 --> 00:05:47,196
Speaker 1: the same time. I should calls it the UGO I

94
00:05:47,356 --> 00:05:54,876
Speaker 1: goo problem. Now let's get back to what's your problem. So, Okay,

95
00:05:54,876 --> 00:05:57,156
Speaker 1: the world has been talking about self driving cars for

96
00:05:57,596 --> 00:06:00,356
Speaker 1: well over ten years now, and we know that to

97
00:06:00,476 --> 00:06:02,516
Speaker 1: get to this self driving car world, there's going to

98
00:06:02,556 --> 00:06:05,876
Speaker 1: be regulatory hurdles. There's going to be people worried about safety.

99
00:06:05,956 --> 00:06:08,156
Speaker 1: I feel like all of that is pretty familiar by now.

100
00:06:08,716 --> 00:06:11,196
Speaker 1: The thing I really wanted to talk with Aisha about

101
00:06:11,396 --> 00:06:15,076
Speaker 1: is the technical side. You know, like, what problem exactly

102
00:06:15,156 --> 00:06:18,036
Speaker 1: do engineers have to solve to build cars that can

103
00:06:18,156 --> 00:06:23,156
Speaker 1: really drive themselves. All of the companies can drive, and

104
00:06:23,236 --> 00:06:27,676
Speaker 1: the normal very kind of constrained circumstances, The thing is

105
00:06:28,236 --> 00:06:30,836
Speaker 1: all of these scenarios that can pop up, how do

106
00:06:30,876 --> 00:06:32,476
Speaker 1: you deal with them? And by the way, how do

107
00:06:32,516 --> 00:06:36,196
Speaker 1: you deal with them knowing you have human drivers around you,

108
00:06:36,276 --> 00:06:41,156
Speaker 1: they have their own learned behaviors and learned expectations as

109
00:06:41,156 --> 00:06:43,556
Speaker 1: to what's going to come from a human driver. And

110
00:06:43,596 --> 00:06:47,196
Speaker 1: the etiquette, the etiquette of it too that will not

111
00:06:47,236 --> 00:06:50,836
Speaker 1: fully there yet. Tell me about the etiquette, Well, depending

112
00:06:50,836 --> 00:06:53,036
Speaker 1: on which parts of the city, like if you're in

113
00:06:53,116 --> 00:06:56,796
Speaker 1: more of a neighborhood area versus more of a business area,

114
00:06:57,156 --> 00:07:00,516
Speaker 1: the behaviors are slightly different in how you approach things.

115
00:07:00,516 --> 00:07:04,156
Speaker 1: So for example, we call it Ugo I goo. The

116
00:07:04,356 --> 00:07:06,676
Speaker 1: ugo igo is a little bit more assertive on the

117
00:07:06,716 --> 00:07:09,996
Speaker 1: business side of things versus in a place where that's

118
00:07:09,996 --> 00:07:13,276
Speaker 1: a little bit more residential. So all that is again

119
00:07:13,636 --> 00:07:16,476
Speaker 1: the long tale of scenarios that we have to deal

120
00:07:16,516 --> 00:07:18,876
Speaker 1: with and be ready for. The you go, I go

121
00:07:19,156 --> 00:07:23,836
Speaker 1: is a really interesting one because that's not about the

122
00:07:23,876 --> 00:07:27,596
Speaker 1: formal like rules of the road, right, that is very

123
00:07:27,716 --> 00:07:31,876
Speaker 1: much a cultural thing that's gonna even vary from town

124
00:07:31,916 --> 00:07:34,396
Speaker 1: to town. Right. I've lived different places and you go,

125
00:07:34,596 --> 00:07:37,956
Speaker 1: I go. It's totally different in you know, Brooklyn than

126
00:07:37,996 --> 00:07:41,556
Speaker 1: it is in Bozeman, Montana, So like, how's an AI

127
00:07:41,676 --> 00:07:44,436
Speaker 1: going to figure that out? This is where training is

128
00:07:44,476 --> 00:07:48,796
Speaker 1: important because before it'll be a long, long, long, long

129
00:07:48,876 --> 00:07:52,476
Speaker 1: time where you just deploy what we call generic AI,

130
00:07:52,556 --> 00:07:54,316
Speaker 1: which is you come in for the first time, you

131
00:07:54,436 --> 00:07:56,396
Speaker 1: drop in and boom, you know how to do it.

132
00:07:56,756 --> 00:07:59,396
Speaker 1: So a lot of a lot of what we do

133
00:07:59,556 --> 00:08:01,316
Speaker 1: is learn and train, and that's why you know it's

134
00:08:01,316 --> 00:08:04,676
Speaker 1: called deep learning for example. And so we do it

135
00:08:04,996 --> 00:08:09,396
Speaker 1: enough times enough forms that the stack knows what to

136
00:08:09,476 --> 00:08:11,436
Speaker 1: look for to make the call. And we have those

137
00:08:11,476 --> 00:08:15,156
Speaker 1: examples today. For example, how we drive in San Francisco

138
00:08:15,316 --> 00:08:18,476
Speaker 1: versus how we drive in the campus were on today,

139
00:08:18,956 --> 00:08:22,516
Speaker 1: it's totally different. We're a lot more assertive in San

140
00:08:22,556 --> 00:08:27,676
Speaker 1: Francisco because that's what's expected. The buffers are smaller because

141
00:08:27,676 --> 00:08:31,196
Speaker 1: that's also what's expected than we do on the campus.

142
00:08:31,236 --> 00:08:33,596
Speaker 1: So all that is built into the stack, both with

143
00:08:33,716 --> 00:08:38,076
Speaker 1: control logic as well as with algorithms logic based on

144
00:08:38,316 --> 00:08:42,316
Speaker 1: our training models. So I'm trying to sort of boil

145
00:08:42,396 --> 00:08:44,436
Speaker 1: down what you're saying. I mean, it's interesting, like on

146
00:08:44,476 --> 00:08:47,596
Speaker 1: a certain level, you're saying you've solved the kind of

147
00:08:47,636 --> 00:08:53,196
Speaker 1: big macro technical problems for computers to drive cars basically,

148
00:08:53,636 --> 00:08:57,556
Speaker 1: But you're also saying there are a million edge cases

149
00:08:57,636 --> 00:09:01,956
Speaker 1: double parked cars and bikes doing weird things, and who knows,

150
00:09:02,036 --> 00:09:06,596
Speaker 1: people are weird, The world is weird, and those you

151
00:09:06,676 --> 00:09:09,516
Speaker 1: haven't solved exactly. You're just saying you just have to

152
00:09:09,556 --> 00:09:11,956
Speaker 1: practice more to get it. You just have to do

153
00:09:12,076 --> 00:09:14,796
Speaker 1: it like that. Why am I unsatisfied by that answer?

154
00:09:15,876 --> 00:09:19,076
Speaker 1: Because it's to be fair, you have to practice. But

155
00:09:19,276 --> 00:09:24,316
Speaker 1: practice is a feedback loop of doing yeah, finding errors

156
00:09:25,476 --> 00:09:28,196
Speaker 1: and figuring out how to deal with the errors, doing

157
00:09:28,236 --> 00:09:31,276
Speaker 1: it again. So it's a continuous feedback loop of that,

158
00:09:31,476 --> 00:09:34,276
Speaker 1: and that is what is left to solve. Can you

159
00:09:34,276 --> 00:09:36,716
Speaker 1: give me an example of a version of that loop

160
00:09:36,716 --> 00:09:39,596
Speaker 1: you've just completed where you found an error and then

161
00:09:39,676 --> 00:09:42,236
Speaker 1: fixed it. What's an example of that? Yeah, this one

162
00:09:42,316 --> 00:09:48,556
Speaker 1: is very personal. My first Zooks ride ever, So that's

163
00:09:48,596 --> 00:09:51,556
Speaker 1: three and a half years ago. When the vehicle saw

164
00:09:51,596 --> 00:09:55,956
Speaker 1: a double parked vehicle in front of it, but there

165
00:09:56,116 --> 00:09:59,156
Speaker 1: was it's a single lane and so you have basically

166
00:09:59,236 --> 00:10:03,556
Speaker 1: a double yellow line. The vehicle says, oh, the rule

167
00:10:03,676 --> 00:10:07,036
Speaker 1: says through the AI stack that I cannot cross a

168
00:10:07,076 --> 00:10:12,196
Speaker 1: double yellow line. Right, We would actually stop, disengage and

169
00:10:12,276 --> 00:10:16,196
Speaker 1: do it manually and go around it. But over time

170
00:10:16,276 --> 00:10:19,516
Speaker 1: we've now learned through all of the dpvs we've seen,

171
00:10:19,956 --> 00:10:23,236
Speaker 1: and so there are many times now when the app

172
00:10:23,436 --> 00:10:27,996
Speaker 1: is a vehicle, ye sorry, when when there's a double

173
00:10:28,036 --> 00:10:32,036
Speaker 1: park vehicle in front of the vehicle. Now the vehicle

174
00:10:32,116 --> 00:10:34,836
Speaker 1: is able to look through the sense of pod, look

175
00:10:34,876 --> 00:10:39,076
Speaker 1: at oncoming traffic, look at everything else, speeds and feeds

176
00:10:39,076 --> 00:10:42,636
Speaker 1: and everything else around it, and many times now it

177
00:10:42,796 --> 00:10:45,356
Speaker 1: is able to just smoothly like a human would do,

178
00:10:46,076 --> 00:10:48,836
Speaker 1: make the decision itself. That's a good one because there's

179
00:10:48,836 --> 00:10:51,396
Speaker 1: a clear rule you can't cross a double yellow line.

180
00:10:51,436 --> 00:10:53,596
Speaker 1: We all learn that. But yeah, we also all know

181
00:10:53,636 --> 00:10:55,916
Speaker 1: if you're driving along and somebody's double parked in front

182
00:10:55,916 --> 00:10:58,036
Speaker 1: of you, got to swerve over and nobody's coming the

183
00:10:58,076 --> 00:11:00,556
Speaker 1: other way, like you're allowed to do that exactly. So

184
00:11:00,996 --> 00:11:03,716
Speaker 1: how does an AI figure out when it's okay to

185
00:11:03,836 --> 00:11:07,956
Speaker 1: do that? So we through again the algorithms, through a

186
00:11:07,996 --> 00:11:10,956
Speaker 1: lot of training, meaning all the scenarios. If it's seen

187
00:11:11,076 --> 00:11:14,876
Speaker 1: over and over again, it's able to say, okay, I looked.

188
00:11:15,476 --> 00:11:17,516
Speaker 1: And this is where again the placement of the center

189
00:11:17,596 --> 00:11:20,876
Speaker 1: architecture is very important because in our vehicle. Since it's

190
00:11:20,916 --> 00:11:23,036
Speaker 1: every top four corner and you have a two hundred

191
00:11:23,036 --> 00:11:26,156
Speaker 1: and seventy degree view, it's able to basically say, hey,

192
00:11:26,236 --> 00:11:28,636
Speaker 1: I'm looking in front of me. I have plenty of space.

193
00:11:28,836 --> 00:11:31,636
Speaker 1: There's no car coming on the oncoming lane, and I

194
00:11:31,676 --> 00:11:34,236
Speaker 1: see that I have space. Also in front of the

195
00:11:34,716 --> 00:11:38,276
Speaker 1: double park vehicle, there are no pedestrian, there's no bike

196
00:11:38,316 --> 00:11:40,876
Speaker 1: coming up. Oh yeah, I can do this. It goes ahead.

197
00:11:41,116 --> 00:11:44,316
Speaker 1: So you're telling me, basically, like the AI, the machine,

198
00:11:44,316 --> 00:11:46,316
Speaker 1: the vehicle already knows how to drive. It just needs

199
00:11:46,316 --> 00:11:49,596
Speaker 1: to practice, just needs to practice for millions of hours.

200
00:11:50,036 --> 00:11:52,116
Speaker 1: It's funny you should say this. I have a sixteen

201
00:11:52,196 --> 00:11:54,836
Speaker 1: year old daughter, so she started the journey of driving,

202
00:11:55,716 --> 00:11:58,956
Speaker 1: and you know, at first we were just within a

203
00:11:58,996 --> 00:12:02,756
Speaker 1: mile of the house in the neighborhood, then maybe some

204
00:12:02,836 --> 00:12:05,036
Speaker 1: areas where they are stop signs, and now we're up

205
00:12:05,076 --> 00:12:08,476
Speaker 1: to a sort of the supermarket, and very soon we'll

206
00:12:08,476 --> 00:12:10,916
Speaker 1: be up to going to school. It is the case

207
00:12:11,036 --> 00:12:16,476
Speaker 1: that human beings can learn to drive pretty well in

208
00:12:16,556 --> 00:12:23,116
Speaker 1: like what forty hours or so, and computers clearly cannot. Right.

209
00:12:24,276 --> 00:12:29,116
Speaker 1: And I used to think people were bad drivers, right,

210
00:12:29,196 --> 00:12:32,836
Speaker 1: Like it seems obvious to me frankly, that people are

211
00:12:32,836 --> 00:12:35,036
Speaker 1: bad drivers, right. You know, we look at our phones

212
00:12:35,076 --> 00:12:38,116
Speaker 1: while we're driving. We overrate our driving abilities. We have

213
00:12:38,196 --> 00:12:43,436
Speaker 1: literal blind spots. On the other hand, zooks, vehicles can

214
00:12:43,556 --> 00:12:45,876
Speaker 1: see way better than people can see all the way

215
00:12:45,876 --> 00:12:48,076
Speaker 1: around in three hundred and sixty degrees. They have like

216
00:12:48,396 --> 00:12:51,116
Speaker 1: the smartest people in the world trying to teach them

217
00:12:51,116 --> 00:12:53,996
Speaker 1: how to drive. They have millions of hours of practicing,

218
00:12:54,276 --> 00:12:58,636
Speaker 1: and still they're worse than human drivers. So, like, I

219
00:12:58,676 --> 00:13:01,196
Speaker 1: don't know how to parse that, right, Like, are people

220
00:13:01,276 --> 00:13:05,876
Speaker 1: actually way better drivers than I thought? Well, so I

221
00:13:05,996 --> 00:13:08,276
Speaker 1: think that we conflate a lot of things when we

222
00:13:08,316 --> 00:13:10,996
Speaker 1: talk about driving. So let's go back to my daughter.

223
00:13:11,076 --> 00:13:13,876
Speaker 1: I don't think I dispute a little bit that my

224
00:13:14,356 --> 00:13:19,196
Speaker 1: child started learning to drive at sixteen. She started learning

225
00:13:19,196 --> 00:13:22,036
Speaker 1: to drive the first day she was in a car, Yeah,

226
00:13:22,156 --> 00:13:25,196
Speaker 1: or maybe even the first day she was alive. That's

227
00:13:25,276 --> 00:13:29,956
Speaker 1: my seeing physics and seeing the world and understanding people.

228
00:13:30,396 --> 00:13:33,316
Speaker 1: I mean, that's maybe the most interesting thing for me

229
00:13:33,356 --> 00:13:36,716
Speaker 1: in this conversation, right, is like the hard part in

230
00:13:36,796 --> 00:13:39,436
Speaker 1: teaching computers to drive is teaching them to figure out

231
00:13:39,476 --> 00:13:44,996
Speaker 1: people exactly. And ecosystems that are built around and four people.

232
00:13:45,556 --> 00:13:47,756
Speaker 1: That in a way is the ultimate problem you're trying

233
00:13:47,756 --> 00:13:50,636
Speaker 1: to solve, right, Like try and teach a computer to

234
00:13:50,716 --> 00:13:54,196
Speaker 1: think like a person. That's exactly right. And I think

235
00:13:54,196 --> 00:13:58,036
Speaker 1: when I put my firstborn in the car seat coming

236
00:13:58,076 --> 00:14:01,876
Speaker 1: home from the hospital, she was already starting to learn huh,

237
00:14:02,556 --> 00:14:05,676
Speaker 1: say more. You know you're in a car, you know

238
00:14:05,756 --> 00:14:09,316
Speaker 1: it's moving, you know there's a driver, your parents, you're

239
00:14:09,356 --> 00:14:13,556
Speaker 1: looking around you. You're already starting to your internal algorithms

240
00:14:13,556 --> 00:14:16,796
Speaker 1: are already starting to learn and take inputs. You know

241
00:14:16,916 --> 00:14:18,916
Speaker 1: that you need to stop at a light before you're

242
00:14:18,956 --> 00:14:21,996
Speaker 1: an actual driver, right, and then all of the weirdness

243
00:14:22,036 --> 00:14:24,716
Speaker 1: of like if somebody is nodding, or if somebody's pointing,

244
00:14:24,796 --> 00:14:27,196
Speaker 1: or if somebody's waving, and the different things a wave

245
00:14:27,276 --> 00:14:29,636
Speaker 1: can mean, Like there can be like the nice guy point,

246
00:14:29,676 --> 00:14:32,276
Speaker 1: like hey, nice job, or like the angry point, like

247
00:14:32,356 --> 00:14:35,436
Speaker 1: what are you doing when you're driving? Certainly in the city,

248
00:14:35,476 --> 00:14:38,236
Speaker 1: you need to know what those different things mean. That

249
00:14:38,396 --> 00:14:42,036
Speaker 1: is exactly right, and that is the essence of the

250
00:14:42,116 --> 00:14:45,156
Speaker 1: problem that is left to solve. And how do you

251
00:14:45,196 --> 00:14:52,396
Speaker 1: solve it? Practice train in code, figuring out to figure

252
00:14:52,436 --> 00:14:57,276
Speaker 1: out a way to give the computers as many inputs

253
00:14:57,316 --> 00:15:01,116
Speaker 1: as possible, teaching it how to make decisions, and very important,

254
00:15:02,236 --> 00:15:05,396
Speaker 1: making sure that you teach the computer to know what

255
00:15:05,436 --> 00:15:08,876
Speaker 1: it is it doesn't know, so that when it doesn't

256
00:15:08,876 --> 00:15:12,436
Speaker 1: know something, it tells us or it shows us. So

257
00:15:12,476 --> 00:15:14,836
Speaker 1: then we can sit down and say, Okay, that's a problem.

258
00:15:14,916 --> 00:15:17,076
Speaker 1: How do we get around that? How do we solve

259
00:15:17,116 --> 00:15:22,116
Speaker 1: for that? That's really interesting. Fundamentally, this is a discussion

260
00:15:22,156 --> 00:15:25,316
Speaker 1: about risk, right. I mean, what you're saying is Zoo's

261
00:15:25,396 --> 00:15:30,276
Speaker 1: vehicles can basically drive themselves now, but not safely enough.

262
00:15:30,636 --> 00:15:34,116
Speaker 1: That's exactly right, And so I mean one question is

263
00:15:36,756 --> 00:15:38,876
Speaker 1: how safe are they going to have to be? Right?

264
00:15:38,956 --> 00:15:42,516
Speaker 1: I imagine as good as human drivers is not good enough.

265
00:15:43,276 --> 00:15:46,316
Speaker 1: This is clearly a super high stakes, literally a life

266
00:15:46,316 --> 00:15:52,596
Speaker 1: and death question. No system is infinitely safe. How good

267
00:15:52,676 --> 00:15:56,156
Speaker 1: is good enough? We have to do way, way, way

268
00:15:56,236 --> 00:16:01,556
Speaker 1: better than humans today, both on the crashes and on

269
00:16:01,956 --> 00:16:08,596
Speaker 1: the number of miles driven per incident, basically and twice

270
00:16:08,716 --> 00:16:11,836
Speaker 1: twice as good. That's like, how much how goodes it

271
00:16:11,876 --> 00:16:14,596
Speaker 1: have to be? Like in a purely rational world, a

272
00:16:14,596 --> 00:16:16,836
Speaker 1: little bit better would be good enough. Right. Human beings

273
00:16:16,876 --> 00:16:19,316
Speaker 1: are not just purely irrational. They are very rational, but

274
00:16:19,356 --> 00:16:22,956
Speaker 1: they are not purely irrational, and so again I don't.

275
00:16:22,996 --> 00:16:25,356
Speaker 1: None of us in the industry have given our metrics,

276
00:16:25,356 --> 00:16:26,716
Speaker 1: so I'm not going to be the first one to

277
00:16:26,716 --> 00:16:29,516
Speaker 1: do that. There's a reason we haven't. But we have

278
00:16:29,556 --> 00:16:32,036
Speaker 1: to be much safer than humans. This is not a

279
00:16:32,236 --> 00:16:34,876
Speaker 1: be as safe as the type situation a little bit better.

280
00:16:34,956 --> 00:16:37,396
Speaker 1: I would not consider that to be a responsible thing

281
00:16:37,436 --> 00:16:40,076
Speaker 1: to do. In a minute, in the Lightning round, I

282
00:16:40,116 --> 00:16:42,836
Speaker 1: should tells us where of all the places she's lived,

283
00:16:43,036 --> 00:16:47,076
Speaker 1: human drivers are the worst. Also one domain where AI

284
00:16:47,196 --> 00:16:55,836
Speaker 1: will never be better than humans. Okay, let's get back

285
00:16:55,836 --> 00:16:58,156
Speaker 1: to the show. We're going to close with the Lightning round.

286
00:16:58,556 --> 00:17:01,316
Speaker 1: So I love this. I love the conversation. I love

287
00:17:01,356 --> 00:17:04,916
Speaker 1: talking about big, complicated technical things. But I also love

288
00:17:05,116 --> 00:17:09,876
Speaker 1: asking lots of questions really fast at the end of

289
00:17:09,916 --> 00:17:14,356
Speaker 1: an interview. Are you ready? I am. What is the

290
00:17:14,356 --> 00:17:16,676
Speaker 1: one piece of advice you'd give to somebody who is

291
00:17:16,716 --> 00:17:20,636
Speaker 1: trying to solve a hard problem? Break it down. What's

292
00:17:20,676 --> 00:17:24,836
Speaker 1: the biggest misconception people have about self driving cars? That

293
00:17:24,996 --> 00:17:31,676
Speaker 1: cameras only, that cameras are enough? That's the tesla you Yes,

294
00:17:32,036 --> 00:17:34,916
Speaker 1: what driving advice have you given your sixteen year old

295
00:17:34,956 --> 00:17:38,796
Speaker 1: daughter who is learning to drive. Relax and pay attention.

296
00:17:39,636 --> 00:17:43,676
Speaker 1: What is one domain where AI will never be better

297
00:17:43,756 --> 00:17:48,796
Speaker 1: than humans? And don't say love understanding the soul of

298
00:17:48,836 --> 00:17:54,396
Speaker 1: a human empathy, empathy. Of all the places you've lived,

299
00:17:54,516 --> 00:17:57,436
Speaker 1: where were the drivers the worst? You're gonna get me

300
00:17:57,476 --> 00:18:02,516
Speaker 1: in trouble? Israel was pretty tough driving environment standpoint. I

301
00:18:02,596 --> 00:18:05,636
Speaker 1: love the country. I would live there again. Driving there

302
00:18:05,716 --> 00:18:09,276
Speaker 1: is pretty tough. Interesting in terms of like norms and I, Oh,

303
00:18:09,276 --> 00:18:12,156
Speaker 1: you go, right, it's basically I go, I go pretty much.

304
00:18:13,036 --> 00:18:19,636
Speaker 1: So where have AI engineers underrated humans self driving? And

305
00:18:19,676 --> 00:18:23,596
Speaker 1: where have AI engineers overrated humans? What are humans worse at? Oh?

306
00:18:23,636 --> 00:18:26,996
Speaker 1: We cannot process what I call linear known processing where

307
00:18:26,996 --> 00:18:31,396
Speaker 1: the rules are clear, the algorithm is clear, we will

308
00:18:31,436 --> 00:18:36,236
Speaker 1: never beat the machine. Yeah, so we're terrible at chess,

309
00:18:36,276 --> 00:18:39,996
Speaker 1: but actually pretty good at driving. That's exactly right. Well,

310
00:18:40,076 --> 00:18:43,516
Speaker 1: this was delightful. Thank you for your time. I really

311
00:18:43,636 --> 00:18:45,756
Speaker 1: enjoyed it too. Thank you. I have to thank you

312
00:18:45,836 --> 00:18:48,316
Speaker 1: for something that nobody has done, and I do a

313
00:18:48,356 --> 00:18:52,236
Speaker 1: lot of these over the years. You pushed me to roots,

314
00:18:52,516 --> 00:18:55,596
Speaker 1: to the root of and to finding a way to

315
00:18:56,596 --> 00:18:59,956
Speaker 1: decipher the essence of self driving and why it's hard,

316
00:18:59,956 --> 00:19:04,676
Speaker 1: and I really appreciate it. I learned something today. Aisha

317
00:19:04,756 --> 00:19:08,796
Speaker 1: Evans is the CEO of Zoos. Today's show was produced

318
00:19:08,796 --> 00:19:11,996
Speaker 1: by Edith Russelo. It was edited by Kate Parkinson Morgan

319
00:19:12,036 --> 00:19:14,756
Speaker 1: and Robert Smith, and it was engineered by Amanda kay Wong.

320
00:19:15,116 --> 00:19:17,836
Speaker 1: Theme music by Luis Gara. Our development team is Lee,

321
00:19:17,916 --> 00:19:20,876
Speaker 1: Tom Mulad and Justine Lang. A huge team of people

322
00:19:20,956 --> 00:19:24,036
Speaker 1: makes What's Your Problem possible. That team includes, but is

323
00:19:24,076 --> 00:19:28,116
Speaker 1: not limited to, Jacob Weisberg, Na Lobel, Heather Fame, John Schnars,

324
00:19:28,196 --> 00:19:31,836
Speaker 1: Kerry Brodie, Carli mcgleory, Christina Sullivan, Jason Gambrel, Grant Hayes,

325
00:19:32,596 --> 00:19:35,956
Speaker 1: Eric Sandler, Maggie Taylor, Morgan Ratner, Nicole Morano, Mary Beth Smith,

326
00:19:36,076 --> 00:19:39,916
Speaker 1: Royston Breserve, Maya Kanig, Daniella Lakhan, Kazia Tan and David Blefer.

327
00:19:40,316 --> 00:19:43,036
Speaker 1: What's Your Problem is a co production of Pushkin Industries

328
00:19:43,076 --> 00:19:46,436
Speaker 1: and iHeartMedia. To find more Pushkin podcast listen on the

329
00:19:46,516 --> 00:19:51,276
Speaker 1: iHeartRadio app, Apple Podcasts, or where Effort. I'm Jacob Boldstein

330
00:19:51,316 --> 00:19:53,796
Speaker 1: and I'll be back next week with another episode of

331
00:19:53,916 --> 00:19:54,636
Speaker 1: What's Your Problem