1
00:00:01,159 --> 00:00:03,879
Speaker 1: Now now you know, we made it a point not

2
00:00:04,000 --> 00:00:07,080
to not to rehearse anything, and so as I just

3
00:00:07,120 --> 00:00:09,919
want to just as a as a as a just reminder,

4
00:00:10,560 --> 00:00:13,560
you're you're my last thing. Okay, okay, could you not

5
00:00:13,640 --> 00:00:14,439
ruin the whole thing?

6
00:00:14,720 --> 00:00:15,080
Speaker 2: All right?

7
00:00:15,279 --> 00:00:19,559
Speaker 1: All right? So now speaking of that, speaking of that,

8
00:00:19,640 --> 00:00:21,719
I think everybody would like to before we get into

9
00:00:21,760 --> 00:00:24,480
all of the good stuff, okay, and they want to

10
00:00:24,480 --> 00:00:26,199
go directly to the juicy stuff.

11
00:00:26,600 --> 00:00:27,399
Speaker 2: Okay, okay.

12
00:00:27,440 --> 00:00:31,839
Speaker 1: And the juicy stuff is this. Look, you know, you

13
00:00:31,879 --> 00:00:36,799
were quote as a saying that that artificial intelligence is

14
00:00:37,119 --> 00:00:44,200
more dangerous than nuclear weapons, and I said potentially, and

15
00:00:44,200 --> 00:00:48,039
and well it goes on, it goes on, and you say,

16
00:00:48,200 --> 00:00:51,719
you say that it's like summoning the demon could be.

17
00:00:53,039 --> 00:00:56,679
How do you consolidate, rationalize the the the conflict between

18
00:00:56,799 --> 00:01:00,159
artificial intelligence of course deep learning that that obviously is

19
00:01:00,159 --> 00:01:01,880
going to be very important to self driving cars. How

20
00:01:01,920 --> 00:01:02,520
do you think through that?

21
00:01:03,320 --> 00:01:04,159
Speaker 2: Well, I don't think.

22
00:01:04,079 --> 00:01:07,719
Speaker 3: We're to worry about autonomous cars because that's sort of

23
00:01:07,760 --> 00:01:12,200
like a narrow form of AI, and it's not something

24
00:01:12,239 --> 00:01:14,159
that I think is very difficult. Actually, I think the

25
00:01:14,879 --> 00:01:18,640
to do autonomous driving to agree that's much safer than

26
00:01:18,680 --> 00:01:19,120
a person.

27
00:01:19,359 --> 00:01:21,200
Speaker 2: Is much easier than people think.

28
00:01:21,319 --> 00:01:26,840
Speaker 3: Yeah, right, and yeah, I think it's going to just

29
00:01:26,879 --> 00:01:29,760
become normal, Like it be like an elevator, Like nobod

30
00:01:29,760 --> 00:01:33,840
they used to have elevator operators and then we you know,

31
00:01:33,879 --> 00:01:38,159
we've developed some simple circuitry to have elevators just automatically

32
00:01:38,200 --> 00:01:40,280
come to the floor that you're at and you can

33
00:01:40,760 --> 00:01:43,640
just press the button. Nobody needs to operate the elevator.

34
00:01:44,879 --> 00:01:46,159
The car is just going to be like that.

35
00:01:46,719 --> 00:01:48,760
Speaker 1: And the elevators these days are even smart. I mean

36
00:01:48,760 --> 00:01:52,480
it knows it knows where to position an elevator, so

37
00:01:52,799 --> 00:01:55,040
that if you were to need an elevator, it's pretty

38
00:01:55,040 --> 00:01:56,840
close to you. Cars in the future will be pretty

39
00:01:56,840 --> 00:01:57,560
smart about that too.

40
00:01:57,959 --> 00:02:00,400
Speaker 3: Yeah, you'll be able to tell your car like take home,

41
00:02:01,480 --> 00:02:05,640
go here, go there, anything, and it'll just do it

42
00:02:06,280 --> 00:02:08,280
at an order of mag to safer than a person.

43
00:02:09,199 --> 00:02:11,599
Speaker 2: In fact, in the in the distant future, I.

44
00:02:11,560 --> 00:02:14,960
Speaker 3: Think it's probably going to be if people may outlaw

45
00:02:15,800 --> 00:02:19,840
driving cars because it's too dangerous, Like you can't have

46
00:02:19,919 --> 00:02:22,080
a person driving a two time death machine.

47
00:02:24,599 --> 00:02:26,639
Speaker 1: Now, if we if we have the right type of

48
00:02:26,639 --> 00:02:29,280
intelligence in a car, we we also don't have to

49
00:02:29,280 --> 00:02:32,560
make the cars that heavy. I would think, you know,

50
00:02:32,639 --> 00:02:34,800
cars are getting heavier and heavier, and it's got more

51
00:02:34,800 --> 00:02:37,159
and more stuff in it because it needs to survive

52
00:02:37,159 --> 00:02:39,800
all these incredible collisions and things like that. If I

53
00:02:39,840 --> 00:02:43,360
wonder if if we were to design cars that just

54
00:02:43,400 --> 00:02:46,360
simply don't collide as much. I wonder if we could,

55
00:02:46,479 --> 00:02:49,159
we could relax on some of those laws and and yeah,

56
00:02:49,240 --> 00:02:51,439
make cars more fuel efficient and lighter and better to.

57
00:02:51,520 --> 00:02:52,560
Speaker 2: Right, you could definitely do that.

58
00:02:52,599 --> 00:02:54,879
Speaker 3: If you could count on not having an accident, then

59
00:02:54,919 --> 00:02:57,280
you can get rid of a huge amount of the

60
00:02:57,280 --> 00:03:02,159
crash structure and the airbags. And it'll be We're a

61
00:03:02,199 --> 00:03:03,879
long way from that, because there's always gonna be some

62
00:03:04,080 --> 00:03:05,639
for a very long time, there'll be some amount of

63
00:03:05,719 --> 00:03:09,240
legacy cars on the road. And I think it is

64
00:03:09,280 --> 00:03:13,800
important to just appreciate the size of the automotive industrial base.

65
00:03:14,120 --> 00:03:18,360
Like it's not as though like when somebody makes an

66
00:03:18,360 --> 00:03:20,879
autonomous car that suddenly all the cars will be autonomous.

67
00:03:20,960 --> 00:03:22,759
Speaker 2: It's like there's two billion of them.

68
00:03:23,120 --> 00:03:26,599
Speaker 3: Okay, So the total total number of cars and trucks

69
00:03:26,599 --> 00:03:31,479
on the road is two billion. In climbing, the capacity

70
00:03:31,759 --> 00:03:34,919
of car and truck production is about one hundred million

71
00:03:34,919 --> 00:03:38,520
a year. So if tomorrow all cars were autonomous, it

72
00:03:38,520 --> 00:03:40,960
would take twenty years to replace the fleet, assuming the

73
00:03:40,960 --> 00:03:43,199
fleet stayed the same size. Arguably it could get smaller

74
00:03:43,199 --> 00:03:47,400
if things are autonomous, but still it's it's still you know,

75
00:03:47,439 --> 00:03:50,280
maybe fifteen years or something, and it's not all going

76
00:03:50,360 --> 00:03:55,319
to transition immediately. It'll take quite a while. So and

77
00:03:55,360 --> 00:04:01,080
it's the same for electrification of carsing that industrial base

78
00:04:01,199 --> 00:04:04,319
to be electric. I mean, if if all cars were

79
00:04:04,360 --> 00:04:07,120
suddenly fall cars produced were electric tomorrow, it would still

80
00:04:07,159 --> 00:04:08,800
take twenty years to replace the fleet.

81
00:04:10,400 --> 00:04:11,879
Speaker 2: And right now it's less than one percent.

82
00:04:12,759 --> 00:04:16,160
Speaker 1: So now you you you're you mentioned just now about

83
00:04:16,240 --> 00:04:19,240
about self driving cars being easier than people think. Now

84
00:04:19,480 --> 00:04:21,720
you have your vision of how to go from where

85
00:04:21,720 --> 00:04:24,759
we are today. Now, my model, my P eighty five

86
00:04:24,839 --> 00:04:28,519
D has lane detection and so it gets a little

87
00:04:28,959 --> 00:04:31,360
you know, when I get close to a lane, yep,

88
00:04:31,600 --> 00:04:34,759
it detects the uh uh the speed signs and then

89
00:04:34,839 --> 00:04:39,240
uses uses computer vision technology to do that. And but

90
00:04:39,839 --> 00:04:43,120
and that's today's adas. What is your what is your roadmap?

91
00:04:43,240 --> 00:04:45,120
You know, how is that different than other people's roadmap?

92
00:04:45,120 --> 00:04:46,560
How do you think about how to get to self

93
00:04:46,639 --> 00:04:47,439
driving cars.

94
00:04:48,040 --> 00:04:48,680
Speaker 2: Yeah, well.

95
00:04:50,120 --> 00:04:52,439
Speaker 3: You kind of need the hardware foundation, the sort of

96
00:04:52,519 --> 00:04:55,920
sensor and computing foundation, and then you can keep uploading

97
00:04:55,959 --> 00:04:57,720
new software. At least you can with the TAILS because

98
00:04:57,720 --> 00:05:01,160
it's it's always connected. So the car that you have,

99
00:05:01,319 --> 00:05:06,120
you'll notice that it's the features are steadily improving. We

100
00:05:06,240 --> 00:05:09,399
now you know, have active cruise control, so it'll it'll

101
00:05:09,480 --> 00:05:12,160
use radar and camera fusion to track the car in

102
00:05:12,199 --> 00:05:16,360
front of you. It's also looking at with some of

103
00:05:16,360 --> 00:05:18,399
the things that are coming out. It's got it looks

104
00:05:18,399 --> 00:05:20,639
at the brake lights, so it anticipates that the car's

105
00:05:20,639 --> 00:05:23,240
got the brake lights are active. It's going to get

106
00:05:23,360 --> 00:05:26,439
basically smarter and smarter even with the current hardware suite.

107
00:05:26,480 --> 00:05:28,399
So the current hardware suite is three hundred and sixty

108
00:05:28,439 --> 00:05:31,439
degree ultrasonic sensors that go off to about just over

109
00:05:31,519 --> 00:05:34,519
five meters. It's a forward camera to Ford radar. So

110
00:05:34,759 --> 00:05:38,120
we'll make even with just that sensor suite, we can

111
00:05:38,160 --> 00:05:44,480
actually make a huge progress in autonomy. We can certainly

112
00:05:44,560 --> 00:05:48,959
make the car steer itself on on a freeway, do

113
00:05:49,120 --> 00:05:54,160
lane changes. It's really autonomy is about what level of

114
00:05:54,199 --> 00:05:56,959
reliability and safety do you want?

115
00:05:58,800 --> 00:05:59,800
Speaker 2: Even with the current sensorus.

116
00:06:00,160 --> 00:06:02,759
Speaker 3: We could make the cargo fully autonomous, but only to

117
00:06:03,680 --> 00:06:05,519
but not to a level of reliabily that would be

118
00:06:05,600 --> 00:06:12,120
safe in say a complex urban environment at thirty miles

119
00:06:12,120 --> 00:06:14,920
an hour, where the lanemarking's not there and children could

120
00:06:14,920 --> 00:06:17,319
be playing and things could be coming at.

121
00:06:17,160 --> 00:06:18,040
Speaker 2: You from the side.

122
00:06:18,600 --> 00:06:23,000
Speaker 3: So in order to solve that, you need a bigger

123
00:06:23,040 --> 00:06:27,720
sensor suite, and you need more computing power. And I

124
00:06:27,720 --> 00:06:31,000
think what you're doing actually with the tigers in the

125
00:06:31,000 --> 00:06:33,279
futures is super interesting and will really be a big

126
00:06:33,399 --> 00:06:36,839
enabler for autonomous driving. So I think, you know, we're

127
00:06:37,120 --> 00:06:38,800
in video is doing really great stuff on that front.

128
00:06:38,839 --> 00:06:41,519
Speaker 1: I appreciate that. Yeah, And so some of the challenges

129
00:06:41,560 --> 00:06:43,720
that you see, what are the what are some of

130
00:06:43,720 --> 00:06:49,000
the technological hurdles that And there's all kinds of researchers

131
00:06:49,000 --> 00:06:50,560
in the room, they're all kinds of engineers in room.

132
00:06:50,720 --> 00:06:52,759
What are some what are some of the technological hurdles

133
00:06:52,759 --> 00:06:56,079
that you think are really important for us to go tackle. Surely,

134
00:06:56,240 --> 00:07:00,680
surely we're going to get to some better cruise controls

135
00:07:00,879 --> 00:07:03,399
on highways, but beyond that, what are some of the

136
00:07:03,399 --> 00:07:05,120
things that you would like is to go focus on

137
00:07:05,399 --> 00:07:06,480
the tackle for the car.

138
00:07:06,319 --> 00:07:11,639
Speaker 3: Industry, Well, it's it. You know, where it gets tricky

139
00:07:11,839 --> 00:07:16,800
is is just the is that sort of urban environment

140
00:07:16,959 --> 00:07:19,879
around thirty or forty miles an hour. So like right

141
00:07:19,920 --> 00:07:22,759
right now, it's fairly easy to deal with, say, things

142
00:07:22,759 --> 00:07:25,680
that are sub five to ten miles an hour, because

143
00:07:25,680 --> 00:07:28,160
we can do that with the ultrasonics. We just make

144
00:07:28,199 --> 00:07:30,920
sure it doesn't hit anything right, you know, because you.

145
00:07:30,839 --> 00:07:32,720
Speaker 1: Can always just the right thing to do. Largely, that's

146
00:07:32,720 --> 00:07:33,319
why would.

147
00:07:33,120 --> 00:07:36,399
Speaker 3: You want to hit anything with your exactly, So at

148
00:07:36,439 --> 00:07:38,680
five ten miles an hour, you can stop within the

149
00:07:38,759 --> 00:07:43,079
range of the ultrasonics, and that then from let's say

150
00:07:43,079 --> 00:07:44,079
ten miles an hour to.

151
00:07:46,519 --> 00:07:48,759
Speaker 2: You know, call it sort of fifty miles an hour.

152
00:07:48,800 --> 00:07:54,240
Speaker 3: That that that that area in complex suburban environments, that's

153
00:07:54,279 --> 00:07:59,240
that's where you can get a lot of unexpected things happening,

154
00:07:59,399 --> 00:08:01,439
like let's say this of like a road closure or

155
00:08:01,439 --> 00:08:05,920
a manhole cover open. Children playing is a big issue bicycles.

156
00:08:07,360 --> 00:08:09,199
Once you get about fifty miles an hour and you're

157
00:08:09,240 --> 00:08:11,279
in kind of a freeway environment, then it also gets

158
00:08:11,279 --> 00:08:15,839
easier again, like the set of possibilities is much reduced,

159
00:08:17,160 --> 00:08:22,199
so highway crews is easy. Low speed is easy, intermediate

160
00:08:22,279 --> 00:08:27,519
is hard. And so being able to recognize what you're

161
00:08:27,560 --> 00:08:31,560
seeing and make the right decision in the suburban environment

162
00:08:31,759 --> 00:08:33,759
in that ten miles an hour to fifty mile an

163
00:08:33,759 --> 00:08:39,360
hour zone is the challenging portion. But I really think

164
00:08:40,039 --> 00:08:42,919
it's I mean, I almost this may sound a little complacent,

165
00:08:42,919 --> 00:08:44,639
but I almost viewed it as like a solve problem,

166
00:08:44,799 --> 00:08:46,000
Like we know exactly what.

167
00:08:45,919 --> 00:08:47,919
Speaker 2: To do, and we'll be there in a few years.

168
00:08:48,519 --> 00:08:56,879
Speaker 1: Right, It's just like Mars. That's quite side. That's kind

169
00:08:56,879 --> 00:09:01,200
of the spirit of a I mean, in a lot

170
00:09:01,200 --> 00:09:03,360
of ways in your mind, you kind of you kind

171
00:09:03,360 --> 00:09:07,159
of see things solvable or arguably arguably solved, and and

172
00:09:07,600 --> 00:09:09,519
a lot of it is really about getting there.

173
00:09:10,200 --> 00:09:13,840
Speaker 3: Yeah, we'll take autonomous costs for granted, in quite a

174
00:09:13,919 --> 00:09:16,600
short period of time. It's amazing how comfortable you get

175
00:09:16,720 --> 00:09:18,240
at how quickly you get comfortable with it.

176
00:09:19,879 --> 00:09:21,039
Speaker 2: So, now, what.

177
00:09:21,039 --> 00:09:24,360
Speaker 1: About government government policies? Like one of the things that

178
00:09:24,360 --> 00:09:26,240
I would like to do is I would I would

179
00:09:26,240 --> 00:09:28,000
just like to keep working on my email as I'm

180
00:09:28,080 --> 00:09:30,600
driving to work. Sure, you know, there's there's a thirty

181
00:09:30,960 --> 00:09:36,200
people do that already. Like I said, I would like

182
00:09:36,240 --> 00:09:40,559
to do it without without without breaking the law. So

183
00:09:41,240 --> 00:09:44,960
where where where do you where do you think government

184
00:09:45,039 --> 00:09:48,879
intervention falls in some of this stuff? Because you know, obviously,

185
00:09:48,879 --> 00:09:50,559
if you car drivee by itself and it does it

186
00:09:50,679 --> 00:09:53,360
even better than people, you would like it to drive

187
00:09:53,399 --> 00:09:55,559
by itself, but largely the laws don't allow you to

188
00:09:55,559 --> 00:09:57,840
do that today, right, absolutely, So how do we cross

189
00:09:57,879 --> 00:10:01,879
that bridge? And how do you think about government invention regulations?

190
00:10:02,279 --> 00:10:03,360
Speaker 2: Right? So, I think.

191
00:10:05,080 --> 00:10:07,480
Speaker 3: It'll be from the point at which a car is

192
00:10:07,519 --> 00:10:12,159
definitely safer than a person, there's probably at least another

193
00:10:12,200 --> 00:10:15,480
two or three years after that before regulators will allow

194
00:10:15,559 --> 00:10:17,759
that to be the case, because they will want to

195
00:10:17,799 --> 00:10:21,480
see a large amount of statistical proof that it's not

196
00:10:21,519 --> 00:10:25,399
merely as safe as a person, but much safer. So

197
00:10:25,440 --> 00:10:27,240
I think what you can do is you can run

198
00:10:27,480 --> 00:10:30,759
run it in shadow mode and essentially say, okay, this

199
00:10:30,840 --> 00:10:32,519
is this is what the computer would have done in

200
00:10:32,519 --> 00:10:35,679
all these circumstances, and was there a crash or was

201
00:10:35,679 --> 00:10:35,960
there not?

202
00:10:36,120 --> 00:10:38,360
Speaker 2: Like what are the false parts of false negatives?

203
00:10:39,120 --> 00:10:44,320
Speaker 3: And then you know, it's achieve a large population group

204
00:10:44,759 --> 00:10:47,399
and then and then make a really clear statistical argument

205
00:10:47,799 --> 00:10:50,320
with the regulators, and then they're going to digest that,

206
00:10:50,480 --> 00:10:52,360
observe it for a while, see if they agree with it,

207
00:10:52,720 --> 00:10:54,879
and then I think they will because the evidence will

208
00:10:54,919 --> 00:10:55,519
be overwhelming.

209
00:10:55,720 --> 00:10:58,720
Speaker 1: Yeah, and the evidence is actually already quite overwhelming that

210
00:10:58,759 --> 00:11:01,879
if you, if you, if you would have would have

211
00:11:01,960 --> 00:11:03,879
noticed a break line in front of you in the

212
00:11:03,960 --> 00:11:07,000
highway and you didn't, you didn't crash into a rear

213
00:11:07,000 --> 00:11:10,559
in collision. Right, A lot of laser sate, you know, ideally, ideally,

214
00:11:10,559 --> 00:11:15,600
hopefully people don't don't overreact with this, with this unknown technology,

215
00:11:16,720 --> 00:11:22,559
and uh and prematurely regulate no premature regulations.

216
00:11:23,600 --> 00:11:25,480
Speaker 2: Well, I mean regulation that was a joke.

217
00:11:31,080 --> 00:11:33,759
Speaker 3: I think it when when it comes to public safety,

218
00:11:33,879 --> 00:11:37,879
I think there's there's an argument for being quite cautious

219
00:11:37,919 --> 00:11:41,159
and making sure that things are okay before before there's

220
00:11:41,200 --> 00:11:43,200
a change.

221
00:11:43,440 --> 00:11:45,679
Speaker 2: And I mean, and I.

222
00:11:45,679 --> 00:11:48,000
Speaker 3: Don't think it's the case that right now there's a

223
00:11:48,000 --> 00:11:51,000
fully autonomous system and regulators are not approving it, that

224
00:11:51,120 --> 00:11:53,440
that that really be a substitute for people. But they

225
00:11:53,440 --> 00:11:55,360
will be in a few years now.

226
00:11:55,200 --> 00:11:58,240
Speaker 1: As we get more computer rized technology into these cars

227
00:11:58,240 --> 00:12:01,320
and this car becomes really a software defined car. I

228
00:12:01,320 --> 00:12:03,120
mean a lot of your engineers are software engineers.

229
00:12:03,159 --> 00:12:04,120
Speaker 2: I mean that's yeah, absolutely.

230
00:12:04,200 --> 00:12:06,519
Speaker 1: One of the great things about Tesla. You guys right

231
00:12:06,519 --> 00:12:09,279
here in Silicon Valley, you're rich with software engineers and

232
00:12:09,720 --> 00:12:13,240
you have that you' have that computer sensibility about architecting

233
00:12:13,279 --> 00:12:17,440
a computer properly, designing the software, properly designing the software

234
00:12:17,440 --> 00:12:19,519
for many generations of cars, so where he refines and

235
00:12:19,519 --> 00:12:21,519
gets better and better. And it has been getting better.

236
00:12:21,559 --> 00:12:24,039
I mean the software from the first time you send

237
00:12:24,080 --> 00:12:29,320
me my Tesla to the now it's just like unrecognizable software.

238
00:12:28,919 --> 00:12:30,240
Speaker 2: Right, big improvement.

239
00:12:30,240 --> 00:12:32,200
Speaker 3: So that's why the first thing we try to do

240
00:12:32,440 --> 00:12:36,320
is establish the hardware platform, make sure that we have

241
00:12:36,440 --> 00:12:40,759
the sensors and compute power. And so we do that first,

242
00:12:40,799 --> 00:12:43,440
even though the software is only taking advantage of a

243
00:12:43,480 --> 00:12:47,039
small percentage of the sensors and compute power. And then

244
00:12:47,480 --> 00:12:51,159
we do continuous updates to make the car more and

245
00:12:51,159 --> 00:12:51,840
more capable.

246
00:12:52,159 --> 00:12:54,159
Speaker 2: And we're going to see a lot of that happen

247
00:12:54,559 --> 00:12:55,240
later this year.

248
00:12:57,639 --> 00:12:59,399
Speaker 3: If I didn't have an announcement on Thursday morning, I

249
00:12:59,440 --> 00:13:00,840
would be saying a lot more of it.

250
00:13:00,960 --> 00:13:06,080
Speaker 1: Yeah, the audience doesn't understand why they have to wait

251
00:13:06,120 --> 00:13:09,320
until Thursday morning. You tweeted it already. You're announce that

252
00:13:09,360 --> 00:13:11,360
you're going to do an Ota. What kind of announcement

253
00:13:11,399 --> 00:13:13,679
is that I'm going to do an Ota on Thursday.

254
00:13:14,080 --> 00:13:15,519
That's like a new product announcement.

255
00:13:15,759 --> 00:13:17,840
Speaker 3: These days, it's just it's just a well, it's just

256
00:13:17,879 --> 00:13:20,159
saying that there's going to be a cool on Thursday morning,

257
00:13:20,159 --> 00:13:22,639
and I'll describe what's going to be in version six

258
00:13:22,679 --> 00:13:24,559
point two for anyone who's interested.

259
00:13:25,080 --> 00:13:27,360
Speaker 1: That's so awesome. Though I'm interested. I get excited every

260
00:13:27,399 --> 00:13:29,720
time I get an Ota, and it's you know, one

261
00:13:29,720 --> 00:13:32,080
of the things that was really interesting is in the beginning,

262
00:13:32,080 --> 00:13:36,679
when we first built the first Tesla together, the tegra

263
00:13:36,799 --> 00:13:41,440
in it, we thought was more than enough. And recently

264
00:13:41,480 --> 00:13:44,159
you said, can we just squeeze more performance out of

265
00:13:44,159 --> 00:13:46,960
that platform? And it just happened in literally two years,

266
00:13:47,360 --> 00:13:49,759
you know, several versions of your software updates all of

267
00:13:49,799 --> 00:13:52,759
a sudden, the computing platform is not powerful enough, right,

268
00:13:53,000 --> 00:13:54,519
And it's because you want to add more features, and

269
00:13:54,559 --> 00:13:56,519
a lot of features these days are based on software.

270
00:13:56,679 --> 00:13:56,919
Speaker 2: True.

271
00:13:57,039 --> 00:14:01,200
Speaker 1: Yeah, And so one last question, and it's it has

272
00:14:01,240 --> 00:14:04,559
to do with I guess something that a lot of

273
00:14:04,559 --> 00:14:07,039
people are very concerned about, which is your car becomes

274
00:14:07,039 --> 00:14:10,799
a software platform and software platforms get hacked. How do

275
00:14:10,879 --> 00:14:12,720
you think about that, How do you think about security?

276
00:14:12,759 --> 00:14:13,879
And what are some of the things that we could

277
00:14:13,879 --> 00:14:17,080
do to try to make make make the car more

278
00:14:17,080 --> 00:14:19,759
resilient to security attacks.

279
00:14:20,279 --> 00:14:22,600
Speaker 3: Yeah, I think that that becomes really important when the

280
00:14:22,639 --> 00:14:25,720
cars are fully autonomous. I mean, the way the cars

281
00:14:25,759 --> 00:14:29,519
work right now, every system of the car, it's assumed,

282
00:14:29,559 --> 00:14:32,159
could actually have a mechanical failure of some kind or

283
00:14:32,320 --> 00:14:35,120
a logic failure, a fundamental logic failure. So you can

284
00:14:35,159 --> 00:14:38,919
always overwhelm the breaking of the car with your foot,

285
00:14:39,080 --> 00:14:42,000
and you can overwhelm the steering wheel with your hands. So,

286
00:14:43,240 --> 00:14:44,960
but but when when there is a steering wheel, or

287
00:14:44,960 --> 00:14:46,799
there isn't you know, a brake, pedal or something in

288
00:14:46,840 --> 00:14:49,720
that like you know many is from now, then it's

289
00:14:49,799 --> 00:14:54,879
really really dangerous, you know, because but even as it

290
00:14:54,919 --> 00:14:57,080
is right now, what we spend most of our time

291
00:14:57,120 --> 00:14:59,600
on is making sure that it's it's very difficult to

292
00:14:59,639 --> 00:15:03,320
do a multi car hack. Like if you have direct

293
00:15:03,320 --> 00:15:05,000
access to a car, just like if you've got direct

294
00:15:05,000 --> 00:15:08,200
access to a computer or even a conventional car, you

295
00:15:08,240 --> 00:15:10,879
can do a lot of things to it.

296
00:15:11,120 --> 00:15:12,200
Speaker 2: But but that's less of.

297
00:15:12,200 --> 00:15:14,799
Speaker 3: A concern than somebody being able to hack an arbitrary

298
00:15:14,840 --> 00:15:18,200
car or multiple cars. So that's what we focus our

299
00:15:18,279 --> 00:15:21,080
energy on is making sure that that in that ways

300
00:15:21,080 --> 00:15:23,159
it's it's it's a lot like a like a cell

301
00:15:23,200 --> 00:15:26,960
phone or a laptop, you know, you focus on making

302
00:15:26,960 --> 00:15:29,159
sure that they can't or that it's very difficult for

303
00:15:29,200 --> 00:15:33,159
that to be any kind of system wide hack. So

304
00:15:33,159 --> 00:15:35,559
we put a lot of effort into that, and we

305
00:15:35,600 --> 00:15:40,399
have third parties try to attack it. And in certain

306
00:15:40,440 --> 00:15:43,679
parts of the car at the very fundamental level, like

307
00:15:43,759 --> 00:15:47,720
the drive unit controller or the steering controller have an

308
00:15:47,759 --> 00:15:52,759
additional level of security. So somebody may be able to uh,

309
00:15:53,320 --> 00:15:57,919
you know, hack something that's cosmetic, but it's much harder

310
00:15:57,919 --> 00:16:00,440
to hack something that's that's actually physically agerous.

311
00:16:01,679 --> 00:16:02,759
Speaker 2: There's multiple levels of.

312
00:16:02,720 --> 00:16:04,720
Speaker 1: Secure and so this way, if you if you weren't

313
00:16:04,759 --> 00:16:08,039
able to penetrate maybe the entertainment system, it doesn't allow

314
00:16:08,080 --> 00:16:09,320
you quickly as a result of.

315
00:16:09,240 --> 00:16:12,519
Speaker 3: That, right, I may display a funny message or something,

316
00:16:12,559 --> 00:16:14,240
but it would not you would not be able to

317
00:16:14,240 --> 00:16:16,399
then control the steering or the motor.

318
00:16:16,600 --> 00:16:19,519
Speaker 1: Yeah. Well, the future of cars is so exciting and

319
00:16:19,559 --> 00:16:21,320
the work that you guys are doing are so exciting,

320
00:16:21,320 --> 00:16:23,759
and it's it's it's great to see you guys pioneering

321
00:16:24,559 --> 00:16:26,879
these computer rized cars. I mean a lot of people

322
00:16:26,879 --> 00:16:30,840
think about think about Tesla as the electric car, but

323
00:16:30,879 --> 00:16:33,320
I think it's obviously more than that. It's an electric car,

324
00:16:33,360 --> 00:16:35,639
but it's a whole computer platform on top of that.

325
00:16:36,320 --> 00:16:38,399
Speaker 3: Yeah, I think I think Tess is sort of the

326
00:16:38,480 --> 00:16:40,799
leader in electric cars, but I think we'll also sort

327
00:16:40,799 --> 00:16:43,919
of be the leader in autonomous cars, at least autonomous

328
00:16:43,919 --> 00:16:48,240
cars that people can buy. And and so we're I mean,

329
00:16:48,279 --> 00:16:50,519
if there's anybody interested in working on autonomous cars, would

330
00:16:50,559 --> 00:16:54,519
love to have you work at Tesla, by the way.

331
00:16:56,320 --> 00:16:59,080
So we're gonna put a lot of effort into automotive

332
00:17:00,000 --> 00:17:02,919
autonomous driving because it's just going to be the default

333
00:17:02,919 --> 00:17:05,400
thing and it could save a lot of lives.

334
00:17:05,559 --> 00:17:07,960
Speaker 1: Yeah, so save a lot of lives and hopefully, hopefully

335
00:17:08,000 --> 00:17:10,000
one of these days, it would be nice if Nvidia's

336
00:17:10,039 --> 00:17:13,119
campus has no parking lot, yeah right, that it drops

337
00:17:13,160 --> 00:17:15,039
us off and the meanders off to a place where

338
00:17:15,240 --> 00:17:17,640
the land's a little cheaper and you know, and parks

339
00:17:17,680 --> 00:17:19,799
a whole bunch of cars there and when it's time

340
00:17:19,839 --> 00:17:21,400
to go home, were someone at to.

341
00:17:21,319 --> 00:17:26,000
Speaker 3: Come, it will be extremely transformative, that's for sure. But yeah,

342
00:17:26,000 --> 00:17:27,359
I mean when it comes to AI, I'm not really

343
00:17:27,359 --> 00:17:32,240
worried about narrow AI like like autonomous cars or like

344
00:17:32,440 --> 00:17:36,000
you know, a smart air conditioning unit at the house

345
00:17:36,079 --> 00:17:38,920
or something. It's more like sort of the deep intelligent

346
00:17:39,000 --> 00:17:43,880
stuff that is where we need to be cautious. I

347
00:17:43,920 --> 00:17:48,519
actually think there's many potential flavors of AI. And you know,

348
00:17:49,279 --> 00:17:51,160
it's odd that we're at we're so close to the

349
00:17:51,200 --> 00:17:53,960
advent of AI, Like it's it seems strange that we

350
00:17:53,960 --> 00:17:55,720
would be alive in this in this time.

351
00:17:57,240 --> 00:17:59,559
Speaker 1: We'll come back every year, come back every year, and

352
00:17:59,599 --> 00:18:02,480
you'll see the work that this group was going to do.

353
00:18:02,799 --> 00:18:05,000
I mean, there's so much deep learning work being done here.

354
00:18:05,519 --> 00:18:07,200
You have a lot of engineers here as well, and

355
00:18:07,720 --> 00:18:11,839
there is fantastics to see the whole community focused on

356
00:18:12,119 --> 00:18:14,240
advancing this field. And along the way, we're gonna spin

357
00:18:14,279 --> 00:18:16,000
off a whole bunch of new capabilities. As you know,

358
00:18:16,400 --> 00:18:18,799
that's gonna make cars just safer and more fun to drive.

359
00:18:19,599 --> 00:18:22,680
Long before we have to get to essentially a self

360
00:18:22,759 --> 00:18:24,759
driving car, there's going to be a lot of versions

361
00:18:24,799 --> 00:18:26,839
along the way that's just going to bring joy to

362
00:18:26,880 --> 00:18:27,440
a lot of people.

363
00:18:28,000 --> 00:18:30,119
Speaker 2: Yeah. Absolutely, I just hope there's something left for used

364
00:18:30,160 --> 00:18:30,720
humans to do.

365
00:18:32,400 --> 00:18:34,519
Speaker 1: Well. I'm not gonna let let go of my steeringwell,

366
00:18:34,559 --> 00:18:37,440
you know, I've got mine on the craziness mode and

367
00:18:37,480 --> 00:18:40,200
the sports steering mode. Is that the way you have it?

368
00:18:41,319 --> 00:18:42,759
You get driven to work? Now?

369
00:18:42,880 --> 00:18:47,200
Speaker 3: No, well I drive half the time actually, And which

370
00:18:47,240 --> 00:18:48,640
mode do you have it in? I always have it

371
00:18:48,720 --> 00:18:49,400
insane mode?

372
00:18:49,559 --> 00:18:52,720
Speaker 1: Yeah right, all right, thank you,