1
00:00:00,080 --> 00:00:03,680
Speaker 1: Welcome to the Neuralink presentation. This is an update for

2
00:00:04,839 --> 00:00:07,559
the progress from the neuralink team. It's been an incredible

3
00:00:07,559 --> 00:00:08,279
amount of progress.

4
00:00:08,960 --> 00:00:09,480
Speaker 2: This is.

5
00:00:11,560 --> 00:00:15,240
Speaker 1: We're going to start off high level generally describing what

6
00:00:15,599 --> 00:00:18,559
neuralink is doing, and then we're going to have a

7
00:00:18,679 --> 00:00:21,039
very deep technical dive so you can actually get an

8
00:00:21,120 --> 00:00:24,480
understanding of what exactly we're doing at a granule level

9
00:00:24,679 --> 00:00:29,160
and what we can do to enhance human capabilities and

10
00:00:29,320 --> 00:00:35,479
ultimately build a great future for humanity. So that's a neuronspiring.

11
00:00:35,960 --> 00:00:39,119
It's funny thing that me talking right now is a

12
00:00:39,119 --> 00:00:43,880
bunch of neurons firing that then result in speech.

13
00:00:45,039 --> 00:00:48,359
Speaker 2: That you here. Of course, neurons to fire in your brain.

14
00:00:51,159 --> 00:00:51,479
Speaker 3: Part of.

15
00:00:53,240 --> 00:00:58,399
Speaker 1: This presentation is about demystifying the brain. It is a

16
00:00:58,399 --> 00:01:01,560
remarkable organ. I mean we are the brain. Basically when

17
00:01:01,600 --> 00:01:06,079
you say you that really is uh, you're the brain.

18
00:01:06,159 --> 00:01:08,879
Like you can you can get a heart transplant, you

19
00:01:08,920 --> 00:01:11,920
can get a kidney transplant, but I don't know anyone

20
00:01:11,920 --> 00:01:13,359
who's gotten a brain transplant.

21
00:01:14,640 --> 00:01:16,480
Speaker 2: So you are your brain.

22
00:01:19,560 --> 00:01:24,760
Speaker 1: And your experiences are these neurons firing with the trillions

23
00:01:24,760 --> 00:01:27,439
of synapses.

24
00:01:28,680 --> 00:01:31,280
Speaker 2: That somehow lead to conscious comprehension of the world.

25
00:01:32,439 --> 00:01:36,959
Speaker 1: This is something that we have only begun to understand.

26
00:01:37,000 --> 00:01:39,439
We're really just barely at the beginning of understanding of

27
00:01:39,519 --> 00:01:43,280
what is the nature of consciousness. And I've thought a

28
00:01:43,319 --> 00:01:45,000
lot about what what is consciousness?

29
00:01:45,079 --> 00:01:47,239
Speaker 2: What is it? Where does consciousness arise?

30
00:01:51,799 --> 00:01:53,799
Speaker 1: Because if you start at the beginning of the universe,

31
00:01:54,799 --> 00:01:57,519
assuming physics is true, the physic current, the standard model

32
00:01:57,560 --> 00:02:00,359
of physics is true, then you have this.

33
00:02:01,959 --> 00:02:06,120
Speaker 2: Big bang, the matter condensing.

34
00:02:05,680 --> 00:02:10,159
Speaker 1: Into stars, those stars exploding, a lot of the atoms

35
00:02:10,199 --> 00:02:12,120
that are in your body. Right now, we're once at

36
00:02:12,159 --> 00:02:18,479
the center of stars, those stars exploded, recondensed. Fast forward

37
00:02:18,719 --> 00:02:24,240
thirteen point eight billion years, and here we are. And

38
00:02:24,360 --> 00:02:30,159
somewhere along that very long journey to us, at least

39
00:02:31,159 --> 00:02:37,039
consciousness arose or the molecules started talking to each other.

40
00:02:38,759 --> 00:02:47,639
And it begs the question of what is consciousness? Is

41
00:02:47,639 --> 00:02:50,439
is everything conscious? Maybe it's hard to say where along

42
00:02:50,479 --> 00:02:54,199
that line that there's no sort of discrete point where

43
00:02:54,319 --> 00:03:00,439
consciousness didn't exist, and then suddenly it does exist. It

44
00:03:00,479 --> 00:03:04,800
seems to be maybe you have a condensation of matter

45
00:03:04,879 --> 00:03:09,360
that has a density of like we don't know what

46
00:03:09,759 --> 00:03:11,919
really the real answer is, we don't know what consciousness is.

47
00:03:13,599 --> 00:03:17,159
But with the neural link and the progress of the

48
00:03:17,199 --> 00:03:20,439
company's making, will begin to understand a lot more about

49
00:03:20,840 --> 00:03:30,560
consciousness and what does it mean to to be. Along

50
00:03:30,599 --> 00:03:32,599
the way, we're we're going to solve a lot of

51
00:03:34,840 --> 00:03:38,719
a lot of brain issues where the brains get uh

52
00:03:40,120 --> 00:03:43,240
injury or damaged in some way or didn't develop in

53
00:03:43,400 --> 00:03:47,400
quite the right way. But there's yeah, there's there's a

54
00:03:47,439 --> 00:03:50,960
lot of brain in spine injuries that will so along

55
00:03:51,000 --> 00:03:53,199
the way. And I do want to have size that

56
00:03:53,240 --> 00:03:56,199
this is all going to happen quite slowly, meaning you'll

57
00:03:56,400 --> 00:03:59,759
you'll see it coming. Sometimes people think that suddenly there

58
00:03:59,759 --> 00:04:03,680
will be vast numbers of neural links all over the place.

59
00:04:04,120 --> 00:04:06,800
This is not going to be sudden. You'll be able

60
00:04:06,840 --> 00:04:10,039
to watch it happen, you know, over the course of

61
00:04:10,080 --> 00:04:17,240
several years. And we go through exhaustive regulatory approvals. So

62
00:04:17,319 --> 00:04:20,439
this is not something that we're just doing there by

63
00:04:20,439 --> 00:04:25,560
ourselves without governor oversight. We work closely with the regulators

64
00:04:26,000 --> 00:04:30,519
every sepular way. We're very cautious with the neural links

65
00:04:30,519 --> 00:04:34,600
and humans. That's the reason we're not moving faster than

66
00:04:34,600 --> 00:04:37,079
we are is because we're taking great care with each

67
00:04:37,120 --> 00:04:41,360
individual to make sure we never miss and so far

68
00:04:41,399 --> 00:04:43,439
we haven't, and I hope that continues into the future.

69
00:04:43,480 --> 00:04:46,920
Every single one of our implants and humans is working

70
00:04:47,439 --> 00:04:50,199
and working quite well, and you'll get to hear from

71
00:04:50,360 --> 00:04:53,639
some of the people that have received the implants and

72
00:04:54,720 --> 00:04:58,040
care it in their words. So what we're creating here

73
00:04:58,079 --> 00:05:02,959
with a neuralink device is a generalized input output.

74
00:05:04,759 --> 00:05:06,720
Speaker 2: Technology for the brain.

75
00:05:06,839 --> 00:05:11,399
Speaker 1: So it's how do you get information into or out

76
00:05:11,399 --> 00:05:16,000
of the brain and do so in a way that

77
00:05:16,079 --> 00:05:19,199
does not damage the brain or you know, of course

78
00:05:19,279 --> 00:05:22,639
any negative side effects. So it's a very hard problem

79
00:05:23,480 --> 00:05:28,759
and generally the reactions I've seen to this range from

80
00:05:28,920 --> 00:05:33,319
it's impossible to it's already been done before those people

81
00:05:33,360 --> 00:05:38,839
should meet. Actually, the reality is that there actually have

82
00:05:39,040 --> 00:05:47,360
been limited brain to computer interfaces for several decades on

83
00:05:47,399 --> 00:05:51,680
a very basic basis. Just what we're doing with neurlink

84
00:05:52,279 --> 00:05:58,079
is dramatically increasing the bandwidth by many or as the magnitude,

85
00:05:58,360 --> 00:06:03,040
so you can you can a human bandwidth output is

86
00:06:03,319 --> 00:06:05,240
less than one bit per second over the course of

87
00:06:05,240 --> 00:06:08,560
a day, so there's eighty six seconds in a day.

88
00:06:09,160 --> 00:06:11,800
It's very rare for a person to do more than

89
00:06:11,839 --> 00:06:14,920
eighty six thousand, four hundred bits of output per day.

90
00:06:15,199 --> 00:06:17,079
You'd have to be really talking a lot or typing

91
00:06:17,120 --> 00:06:22,839
all day and you might exceed that. So what we're

92
00:06:22,879 --> 00:06:25,439
talking about here is going from maybe one bit per

93
00:06:25,480 --> 00:06:31,800
second to ultimately megabits and then gigabits per second, and

94
00:06:31,920 --> 00:06:39,560
the ability to do conceptual consensual tlepathy. Now, the input

95
00:06:39,639 --> 00:06:43,240
to the rate is much higher because especially because of vision,

96
00:06:44,360 --> 00:06:47,279
depending upon how you counted, it might be on the

97
00:06:47,360 --> 00:06:52,319
order of a megabit or in the megabt range for input,

98
00:06:52,399 --> 00:06:59,199
primarily due to site, but even for input, we think

99
00:06:59,240 --> 00:07:02,319
that can be traumatic increased to the.

100
00:07:02,240 --> 00:07:03,279
Speaker 2: Gig of it plus level.

101
00:07:05,920 --> 00:07:09,839
Speaker 1: And a lot of the thinking that we do is

102
00:07:10,519 --> 00:07:12,959
which we take a concept in our mind and we

103
00:07:13,160 --> 00:07:16,959
compress that into a small number of symbols. So when

104
00:07:17,000 --> 00:07:20,600
you're trying to communicate with somebody else, you're actually trying

105
00:07:20,639 --> 00:07:26,680
to model their mind state and then take peraps quite

106
00:07:26,680 --> 00:07:29,720
a complex idea that you have, maybe even a complex

107
00:07:29,800 --> 00:07:33,560
image or scene or kind of mental video, and try

108
00:07:33,560 --> 00:07:36,160
to compress that into a few woods or a few keystrokes,

109
00:07:36,800 --> 00:07:38,240
and it's necessarily.

110
00:07:37,839 --> 00:07:38,839
Speaker 2: Going to be very lossy.

111
00:07:39,360 --> 00:07:43,199
Speaker 1: Your ability to communicate is very limited by how fast

112
00:07:43,199 --> 00:07:45,839
you can talk and how fast you can type, and

113
00:07:45,920 --> 00:07:49,480
what we're talking about is unlocking that potential to enable

114
00:07:49,519 --> 00:07:53,279
you to communicate, like I said, thousands, perhaps millions of

115
00:07:53,319 --> 00:07:57,879
times faster than is currently possible. This is an incredibly

116
00:07:58,040 --> 00:08:03,639
profound breakthrough. This would this would be a fundamental change

117
00:08:03,680 --> 00:08:08,879
to what it means to be a human. So we're

118
00:08:08,920 --> 00:08:15,120
starting off with reducing human suffering, so or addressing issues

119
00:08:15,120 --> 00:08:17,199
that people have, save if they've been in an accident,

120
00:08:17,439 --> 00:08:22,800
or they have some uh neural disease that's degenerative, so

121
00:08:22,839 --> 00:08:26,680
they're losing capability to move their body, or some some

122
00:08:26,800 --> 00:08:31,879
kind of injury. Essentially, so enabling the first products is

123
00:08:31,879 --> 00:08:37,320
called telepathy, and that enables someone who has lost the

124
00:08:37,360 --> 00:08:40,879
ability to command their body to be able to communicate

125
00:08:40,879 --> 00:08:43,320
with a computer and move the mouse and actually operate

126
00:08:43,320 --> 00:08:47,279
a computer with roughly the same dexterity, ultimately much more

127
00:08:47,320 --> 00:08:55,039
dexterity than a human with working hands. Then our next

128
00:08:55,080 --> 00:08:58,279
product is blind sight, which will enable those who have

129
00:08:58,960 --> 00:09:01,559
total loss of visioning they've lost their eyes or the

130
00:09:01,559 --> 00:09:05,240
optic nerve or maybe i've never seen, we're even blind

131
00:09:05,240 --> 00:09:09,440
from birth, to be able to see again, initially low resolution,

132
00:09:09,519 --> 00:09:13,639
but ultimately very high resolution, and then in multiple wavelengths,

133
00:09:13,799 --> 00:09:16,440
so you could be like Jordie LaForge in Star Trek

134
00:09:16,440 --> 00:09:19,200
and you can see in radar, you can see infrared, ultraviolet,

135
00:09:19,759 --> 00:09:28,279
superhuman capabilities, sebernetic enhancement essentially, and then along the way

136
00:09:28,399 --> 00:09:32,799
this should help us understand a lot more about consciousness,

137
00:09:32,799 --> 00:09:37,559
what does it mean to be a conscious creature? Well,

138
00:09:37,799 --> 00:09:40,559
understand vastly more about the nature of consciousness as a

139
00:09:40,600 --> 00:09:43,679
result of this, and then ultimately I think this helps

140
00:09:43,720 --> 00:09:53,480
mitigate the civilizational risk of artificial intelligence. We are actually

141
00:09:53,480 --> 00:09:56,519
already we're ready to sort of have three layers of thinking.

142
00:09:57,759 --> 00:10:00,679
There's the olympic system, which is your kind of your instincts,

143
00:10:02,600 --> 00:10:06,080
your cortical system, which is your higher level planning and thinking.

144
00:10:06,679 --> 00:10:10,639
And then the tertiary layer, which is the computers and

145
00:10:10,720 --> 00:10:13,120
machines that you interact with, like your phone, all the

146
00:10:13,240 --> 00:10:14,240
applications you use.

147
00:10:16,120 --> 00:10:18,799
Speaker 2: So people actually are already a cyborg.

148
00:10:19,000 --> 00:10:23,480
Speaker 1: You can maybe have an intuitive sense for this by

149
00:10:23,799 --> 00:10:27,000
how much you miss your phone if you leave it behind.

150
00:10:28,039 --> 00:10:30,240
Leaving your phone behind is like it's almost like missing

151
00:10:30,240 --> 00:10:30,879
limbs syndrome.

152
00:10:32,879 --> 00:10:35,440
Speaker 2: Your phone is somewhat of an extension of yourself, as

153
00:10:35,480 --> 00:10:36,120
is your computer.

154
00:10:36,240 --> 00:10:37,320
Speaker 4: So you.

155
00:10:38,960 --> 00:10:42,279
Speaker 1: Already have this digital tertie layer, but the bandwidth between

156
00:10:42,840 --> 00:10:46,679
your cortex and your digital tertie layer is limited by

157
00:10:46,720 --> 00:10:49,440
speech and by and by half fast you can move

158
00:10:49,480 --> 00:10:52,799
your fingers, and how fast you can consume information visually.

159
00:10:53,440 --> 00:10:58,639
So so, but I think it's actually very important for

160
00:10:58,679 --> 00:11:03,360
us to address that input output boundwidth constraint in order

161
00:11:03,639 --> 00:11:07,399
for the collective will of humanity to match the will

162
00:11:07,440 --> 00:11:16,559
of artificial intelligence. That's my intuition at least. So let's see,

163
00:11:17,759 --> 00:11:26,919
and what this presentation is mostly about is attracting smart

164
00:11:27,000 --> 00:11:30,240
humans to come and work with us on this problem.

165
00:11:30,840 --> 00:11:34,039
So this is not a presentation to raise money or

166
00:11:34,080 --> 00:11:34,759
anything like that.

167
00:11:36,200 --> 00:11:38,320
Speaker 2: We're actually, you know, very well funded.

168
00:11:38,360 --> 00:11:42,159
Speaker 1: We have a lot of great investors, some of the

169
00:11:42,159 --> 00:11:44,679
smartest people in the world are invested in neur Link.

170
00:11:45,679 --> 00:11:48,200
But we we need smart humans to come here and

171
00:11:48,240 --> 00:11:51,559
help solve this problem. So with that, let's uh, let's proceed.

172
00:12:04,799 --> 00:12:09,080
Speaker 5: Hey everyone, my name is DJ I'm my co founder

173
00:12:09,120 --> 00:12:13,720
and president of Neuralink. And as Elon mentioned, well, actually

174
00:12:13,759 --> 00:12:16,919
we're standing in the middle of our robust space. We

175
00:12:16,960 --> 00:12:19,159
have a stage set up, but you know, this is

176
00:12:19,200 --> 00:12:23,159
actually where some of the next generation, most advanced surgical

177
00:12:23,200 --> 00:12:33,639
robots are being built, so welcome to our space. It's

178
00:12:33,679 --> 00:12:36,159
important to highlight that this technology is not being built

179
00:12:36,320 --> 00:12:39,480
in the dark. This is not a secret lab where

180
00:12:39,519 --> 00:12:41,879
we're not sharing any of the progress. In fact, we're

181
00:12:41,879 --> 00:12:46,159
actually sharing you know, the progress very openly and as

182
00:12:46,159 --> 00:12:48,159
well as also telling you exactly what we're going to

183
00:12:48,159 --> 00:12:52,759
be doing, and we're hoping to progress on that as

184
00:12:52,919 --> 00:12:56,080
as diligently and as safely and as carefully as possible.

185
00:12:56,159 --> 00:13:00,360
So to start off, two years ago, when we did

186
00:13:00,399 --> 00:13:04,320
our previous fundraising round, we outlined this path and timeline

187
00:13:04,320 --> 00:13:08,720
to first human and we currently have a clinical trials

188
00:13:08,720 --> 00:13:11,000
in the US for a product that we call Telepathy,

189
00:13:11,159 --> 00:13:15,159
which allows users to control phone or computer purely with

190
00:13:15,200 --> 00:13:17,039
their thoughts. And you're going to see how we do

191
00:13:17,120 --> 00:13:20,919
this and what the impact that this has had. And

192
00:13:21,279 --> 00:13:24,559
not only have we launched this clinical trial, but as

193
00:13:24,559 --> 00:13:27,799
of today we have not just one, but seven.

194
00:13:27,639 --> 00:13:30,200
Speaker 6: Participants and we have an approval.

195
00:13:36,840 --> 00:13:39,200
Speaker 5: And we also have an approval to launch this trial

196
00:13:39,720 --> 00:13:51,279
in Canada, UK and the UAE. So I guess before

197
00:13:51,320 --> 00:13:54,879
we dive into what this technology is and what we built,

198
00:13:55,360 --> 00:13:57,559
but I wanted to quickly share a video with you

199
00:13:57,600 --> 00:14:01,399
guys of when our first five parts participants met each

200
00:14:01,440 --> 00:14:02,320
other for the first time.

201
00:14:02,399 --> 00:14:03,159
Speaker 2: So here you go.

202
00:14:03,759 --> 00:14:06,759
Speaker 6: All right, we have everyone together up.

203
00:14:06,720 --> 00:14:10,279
Speaker 2: Guys, thanks everybody joining. Want to use of you.

204
00:14:10,759 --> 00:14:13,000
Speaker 7: Yeah, I'm Nolan aka P one.

205
00:14:13,480 --> 00:14:16,720
Speaker 8: My name is Alex. I am the second participant in

206
00:14:16,759 --> 00:14:17,720
the Neuralink study.

207
00:14:18,240 --> 00:14:21,919
Speaker 2: I am Brad Smith. Let's cycle P three.

208
00:14:22,759 --> 00:14:27,639
Speaker 9: My name and my G four have.

209
00:14:29,080 --> 00:14:29,879
Speaker 4: Less like.

210
00:14:31,799 --> 00:14:37,440
Speaker 3: Uh yeah, I'm marj I'm P five and I just yes,

211
00:14:37,519 --> 00:14:40,159
I'm kind of to one of the team heres. So yeah,

212
00:14:40,200 --> 00:14:43,279
appreciate it. No trail blazer, you know, somebody's got to

213
00:14:43,279 --> 00:14:44,320
get first. Man, that was you.

214
00:14:44,360 --> 00:14:44,919
Speaker 2: Appreciate that.

215
00:14:45,559 --> 00:14:47,600
Speaker 10: What's been your favorite thing you've been able to do

216
00:14:47,639 --> 00:14:48,639
with the neuralink so far.

217
00:14:49,039 --> 00:14:51,159
Speaker 7: I've just had a good time being able to use

218
00:14:51,200 --> 00:14:54,159
it as I travel buying and drawn.

219
00:14:53,919 --> 00:14:55,279
Speaker 2: A little mustache on a cat.

220
00:14:55,440 --> 00:14:57,399
Speaker 7: Had a lot of fun doing that. I mean, I've

221
00:14:57,399 --> 00:14:58,879
just had a good time playing around with it.

222
00:14:59,279 --> 00:15:00,679
Speaker 2: Oh you know what, But I.

223
00:15:00,600 --> 00:15:03,399
Speaker 7: Do know what My favorite BCI feature is probably not

224
00:15:03,480 --> 00:15:07,279
a feature, but I just I love web Grid more

225
00:15:07,320 --> 00:15:10,600
than I love anything in my life. Probably I think

226
00:15:10,639 --> 00:15:13,799
I could play that game NonStop forever.

227
00:15:14,360 --> 00:15:18,279
Speaker 8: Has to be Fusion three sixty being able to design parts,

228
00:15:18,840 --> 00:15:20,759
design the hat logo with the BCI.

229
00:15:22,679 --> 00:15:23,360
Speaker 2: That's what's up?

230
00:15:23,840 --> 00:15:24,440
Speaker 11: Pretty sweet?

231
00:15:24,480 --> 00:15:25,039
Speaker 2: That's sweet?

232
00:15:25,480 --> 00:15:29,519
Speaker 8: Yeah, Yeah, I have a little uh are do we know?

233
00:15:30,120 --> 00:15:35,000
That takes input from my quad stick converts it into

234
00:15:35,080 --> 00:15:38,679
a PPM signal to go to an RC truck.

235
00:15:39,600 --> 00:15:41,600
Speaker 9: Cool little rock crawler.

236
00:15:42,120 --> 00:15:48,360
Speaker 8: Well with the BCI, I room code to drive the

237
00:15:48,360 --> 00:15:50,559
plane with the quad stick.

238
00:15:51,240 --> 00:15:52,000
Speaker 6: That's awesome.

239
00:15:52,679 --> 00:15:57,960
Speaker 1: The best thing I like Abunerline is being able to

240
00:15:58,200 --> 00:16:02,559
continue to provide.

241
00:16:02,279 --> 00:16:06,840
Speaker 7: For my family and continue working.

242
00:16:07,679 --> 00:16:10,799
Speaker 3: I think my favorite things probably been able to turn

243
00:16:10,840 --> 00:16:13,840
on my TV. Yeah, like the first time in two

244
00:16:13,840 --> 00:16:15,360
and a half years. I was able to do that.

245
00:16:15,559 --> 00:16:18,120
That's pretty sweet with a right shooting the all base.

246
00:16:18,240 --> 00:16:21,480
That's that's sound nice. Excited to see it. BCIs got

247
00:16:21,559 --> 00:16:21,879
going on?

248
00:16:22,879 --> 00:16:23,360
Speaker 11: They got a car?

249
00:16:23,679 --> 00:16:24,600
Speaker 8: What's your shirts say?

250
00:16:25,639 --> 00:16:25,799
Speaker 11: Is it?

251
00:16:25,919 --> 00:16:32,200
Speaker 3: I do a thing called whatever I want.

252
00:16:39,159 --> 00:16:39,440
Speaker 6: Now.

253
00:16:39,679 --> 00:16:41,840
Speaker 5: One of the major figure of merits that we have

254
00:16:42,159 --> 00:16:46,960
is to keep track of monthly hours of independent DCI use. Effectively,

255
00:16:47,080 --> 00:16:50,720
are they using the BCI and not at the clinic

256
00:16:50,720 --> 00:16:53,559
but at their home? And what we have noticed, and

257
00:16:53,600 --> 00:16:56,200
this is a plot of all of the different participants,

258
00:16:56,240 --> 00:17:00,320
first five participants and their usage per month over the

259
00:17:00,360 --> 00:17:03,159
course of the last year and a half and we're

260
00:17:03,159 --> 00:17:06,039
averaging around fifty hours a week of usage and in

261
00:17:06,039 --> 00:17:08,359
some cases peak usage of more than one hundred hours

262
00:17:08,400 --> 00:17:18,720
a week, which is pretty much every waking moments. So

263
00:17:18,799 --> 00:17:20,839
I think it's been incredible to see all of our

264
00:17:20,880 --> 00:17:28,039
participants demonstrating greater independence through their use of PCI. Not

265
00:17:28,079 --> 00:17:31,680
only that, we've also accelerated our implantation cadence as we've

266
00:17:31,799 --> 00:17:35,160
amassed evidence of both clinical safety as well as value

267
00:17:35,200 --> 00:17:38,440
to our participants. So to date, we have four spinal

268
00:17:38,440 --> 00:17:41,880
cord injury participants as well as three ALS participants, with

269
00:17:42,039 --> 00:17:44,799
the last two surgeries happening within one week of each other,

270
00:17:48,799 --> 00:17:50,519
and we're just beginning.

271
00:17:50,880 --> 00:17:52,599
Speaker 6: This is just tip of the iceberg.

272
00:17:52,839 --> 00:17:56,480
Speaker 5: Our end goal is to really build a whole brand interface.

273
00:17:56,640 --> 00:17:59,359
And what do we mean by whole brand interface. We

274
00:17:59,480 --> 00:18:03,119
mean being able to listen to neurons everywhere, be able

275
00:18:03,160 --> 00:18:06,920
to write information to neurons anywhere, be able to have

276
00:18:07,000 --> 00:18:10,680
that fast data wireless transfer, to enable that high bandwidth

277
00:18:10,680 --> 00:18:15,680
connection from our biological brain to the external machines, and

278
00:18:15,839 --> 00:18:19,000
be able to do all of this with fully automated surgery,

279
00:18:19,759 --> 00:18:25,400
as well as enable twenty four hours of usage, and

280
00:18:25,480 --> 00:18:29,039
towards that goal, we're really working on three major product types.

281
00:18:29,400 --> 00:18:32,880
Elon mentioned earlier that our goals to build a generalized

282
00:18:32,920 --> 00:18:36,759
input output platform and technology to the brain so to

283
00:18:37,079 --> 00:18:40,240
afford the output portion of it, which is extremely slow,

284
00:18:40,319 --> 00:18:49,920
through our meat sticks, as Elon calls them.

285
00:18:46,880 --> 00:18:48,640
Speaker 6: Meet hands that are holding the mics.

286
00:18:50,039 --> 00:18:53,880
Speaker 5: We're starting out with helping people with movement disorders, either

287
00:18:53,920 --> 00:18:56,519
through where they lost a mind body connection, either through

288
00:18:56,519 --> 00:18:59,759
a spinal cor injury ALS or a stroke, be able

289
00:18:59,839 --> 00:19:02,839
to regain some of that digital as well as physical

290
00:19:02,839 --> 00:19:06,079
independence through a product that we're building called Telepathy, and

291
00:19:06,200 --> 00:19:09,599
this is our opportunities to build a high channel read

292
00:19:09,680 --> 00:19:14,359
and output device. On the input side of things, there's

293
00:19:15,359 --> 00:19:17,759
opportunities for us to help people that have lost the

294
00:19:17,799 --> 00:19:21,279
ability to see be able to regain that site again

295
00:19:21,839 --> 00:19:24,680
through a product that we're calling blind Site, and this

296
00:19:24,720 --> 00:19:29,200
is our opportunity to build high channel right capabilities and

297
00:19:29,400 --> 00:19:32,279
last but not least, be able to also help people

298
00:19:32,319 --> 00:19:38,920
that are suffering from neurological debilitating dysregulation or psychiatric conditions

299
00:19:39,039 --> 00:19:45,279
or neuropathic pain by inserting our electros in reaching any

300
00:19:45,359 --> 00:19:48,440
brain regions to be able to insert them not just

301
00:19:48,480 --> 00:19:51,359
on the cortical layer, but into the sulk guys as

302
00:19:51,359 --> 00:19:54,119
well as deeper parts of the brain, the so called

303
00:19:54,160 --> 00:19:58,720
limbic system, to really enable better opportunities to just regain

304
00:19:58,759 --> 00:20:03,680
some of that independence. Our north Star metrics is one

305
00:20:03,880 --> 00:20:07,039
increasing the number of neurons that we can interface with,

306
00:20:07,680 --> 00:20:11,079
and second to expand to many diverse areas any parts

307
00:20:11,119 --> 00:20:15,880
of the brain, starting with microfabrication or lithography to change

308
00:20:15,960 --> 00:20:18,279
the way in which we can actually increase the number

309
00:20:18,279 --> 00:20:20,240
of neurons that we can see from a single channel,

310
00:20:21,079 --> 00:20:26,160
and also doing mixed signal chip design to actually increase

311
00:20:26,200 --> 00:20:31,519
the physical channel counts to increase more neurons that we

312
00:20:31,559 --> 00:20:35,839
can interface to to sort of allow more information from

313
00:20:35,960 --> 00:20:37,640
the brain to the outside world.

314
00:20:37,960 --> 00:20:38,799
Speaker 2: And then you.

315
00:20:38,799 --> 00:20:41,480
Speaker 5: Know, everything we built from day one of the company

316
00:20:41,599 --> 00:20:45,200
has always been read and write capable, and with Telepathy

317
00:20:45,359 --> 00:20:48,720
our first product, the focus has been on the read

318
00:20:48,759 --> 00:20:51,720
capabilities or the output and we want to hone in

319
00:20:51,759 --> 00:20:55,799
on our right capability and also show that through accessing

320
00:20:55,839 --> 00:21:00,000
deeper regions within the visual cortex that we can actually

321
00:21:00,200 --> 00:21:14,960
achieve functional vision. So now just to step you through

322
00:21:15,000 --> 00:21:16,920
what the product evolution is going to look like in

323
00:21:16,960 --> 00:21:20,440
the next three years, today, what we have is one

324
00:21:20,480 --> 00:21:23,599
thousand electrodes in the motor cortex, the part of the

325
00:21:23,640 --> 00:21:25,160
small part of the brain that you see in this

326
00:21:25,240 --> 00:21:29,599
animation called the hand knob area that allows participants control

327
00:21:29,680 --> 00:21:31,680
computer cursors as well as gaming consoles.

328
00:21:33,039 --> 00:21:36,200
Speaker 6: Next quarter, we're planning to implant in.

329
00:21:36,240 --> 00:21:41,480
Speaker 5: The speech cortex to directly decode attentive words from brain

330
00:21:41,559 --> 00:21:48,279
signals to speech. And in twenty twenty six, not only

331
00:21:48,359 --> 00:21:52,440
are we going to triple the number of electrodes from

332
00:21:52,519 --> 00:21:56,039
one thousand to three thousand for more capabilities, we're planning

333
00:21:56,039 --> 00:21:59,519
to have our first blind site participant to enable navigation.

334
00:22:08,920 --> 00:22:11,559
And in twenty twenty seven we're going to continue increasing

335
00:22:11,640 --> 00:22:16,000
channel counts, probably another triple, so ten thousand channels, and

336
00:22:16,279 --> 00:22:20,920
also enable for the first time multiple implants, so not

337
00:22:21,039 --> 00:22:24,480
just one in mortal cortex, speech cortex, or visual cortex,

338
00:22:24,519 --> 00:22:30,960
but all of the above. And finally, in twenty twenty eight,

339
00:22:31,240 --> 00:22:34,000
our goals to get to more than twenty five thousand

340
00:22:34,119 --> 00:22:38,400
channels per implant, have multiple of these, have ability to

341
00:22:38,440 --> 00:22:43,559
access any part of the brain for psychiatric conditions, pain dysregulation,

342
00:22:43,960 --> 00:22:46,039
and also start to demonstrate.

343
00:22:45,559 --> 00:22:47,720
Speaker 6: What it would be like to actually integrate with AI,

344
00:22:56,799 --> 00:22:57,279
and all of.

345
00:22:57,319 --> 00:23:00,279
Speaker 5: This is to say that we're really building towards set

346
00:23:00,319 --> 00:23:03,640
of fundamental foundational technology that would allow us to have

347
00:23:04,000 --> 00:23:06,640
hundreds of thousands, if not millions, of channels with multiple

348
00:23:06,680 --> 00:23:10,000
implants for whole bit interfaces that could actually solve not

349
00:23:10,079 --> 00:23:13,160
just these devoltating neurological conditions, but be able to go

350
00:23:13,240 --> 00:23:15,799
beyond the limits of our biology.

351
00:23:15,920 --> 00:23:17,680
Speaker 6: And this vertical integration.

352
00:23:17,279 --> 00:23:20,039
Speaker 5: And the talented team that we have at Neuralink has

353
00:23:20,119 --> 00:23:22,319
been and will continue to be the key recipe for

354
00:23:22,680 --> 00:23:25,319
rapid progress that we will be making. Just to recap

355
00:23:25,359 --> 00:23:30,119
real quick, neuralink is implanted with precision surgical robot. It's

356
00:23:30,160 --> 00:23:34,359
physically invisible, and one week later users are able to

357
00:23:34,359 --> 00:23:37,720
see their thoughts transform into actions and to share more

358
00:23:37,799 --> 00:23:41,559
about what that experience is like. I like to welcome

359
00:23:41,599 --> 00:23:53,680
to Hedge to the stage.

360
00:23:56,599 --> 00:23:59,039
Speaker 12: What's up, guys. My name is Sahedge. I'm from the

361
00:23:59,079 --> 00:24:01,920
Brain Computer Inner team here at Neuralink, and I'm going

362
00:24:01,960 --> 00:24:04,720
to be talking about two things today. The first thing

363
00:24:04,839 --> 00:24:07,880
is what exactly is a neuralink device capable of doing

364
00:24:07,960 --> 00:24:10,519
right now? And the second one is how does that

365
00:24:10,599 --> 00:24:13,200
actually impact the day to day lives of our users?

366
00:24:14,640 --> 00:24:18,799
Very simply put, what the neuralink device does right now

367
00:24:19,119 --> 00:24:23,720
is it allows you to control devices simply just by thinking. Now,

368
00:24:23,720 --> 00:24:26,720
to put that a bit more concretely, I'm about to

369
00:24:26,720 --> 00:24:29,519
play a video of our first user. His name is Noland,

370
00:24:29,519 --> 00:24:32,880
if you remember from DJ section, And what Nolan is

371
00:24:32,920 --> 00:24:36,039
doing is he's looking at a normal off the shelf

372
00:24:36,119 --> 00:24:39,799
MacBook Pro and with his neuralink devices, you're going to

373
00:24:39,839 --> 00:24:42,200
see he's going to be able to control the cursor

374
00:24:42,640 --> 00:24:45,640
simply with his mind, no eye tracking, no other sensors.

375
00:24:46,160 --> 00:24:49,680
And what's special about this particular moment is this is

376
00:24:49,720 --> 00:24:52,400
the first time someone is using a neuralink device to

377
00:24:52,480 --> 00:24:59,079
fully control their cursor. This is not your ordinary brain

378
00:24:59,160 --> 00:25:04,599
controlled cursor. This is actually a record breaking control, literally

379
00:25:04,640 --> 00:25:08,519
on day one, beating decades of brain computer research. And

380
00:25:08,519 --> 00:25:11,039
I'm about to show you the clip on day one,

381
00:25:11,680 --> 00:25:14,039
Nolan breaking the b c I world record.

382
00:25:20,640 --> 00:25:24,359
Speaker 9: Ship good.

383
00:25:26,519 --> 00:25:30,279
Speaker 12: Man, he's a new world record holder.

384
00:25:38,000 --> 00:25:38,839
Speaker 2: There's one of surprise.

385
00:25:40,400 --> 00:25:41,359
Speaker 9: I thought it was higher.

386
00:25:41,799 --> 00:25:44,519
Speaker 7: I thought I would have to get to five or something.

387
00:25:44,599 --> 00:25:46,799
Speaker 9: Oh my gosh, that's crazy.

388
00:25:48,039 --> 00:25:48,680
Speaker 11: It's pretty cool.

389
00:25:52,640 --> 00:25:55,359
Speaker 2: Uh yeah.

390
00:25:55,559 --> 00:25:58,079
Speaker 12: Another really fun thing you could do with the neuralink

391
00:25:58,079 --> 00:26:01,160
device outside of controlling a computer cursors. You can actually

392
00:26:01,640 --> 00:26:03,880
plug it in through USB through a lot of different devices.

393
00:26:04,480 --> 00:26:08,000
And here we actually have Nolan playing Mario Kart. Now,

394
00:26:08,000 --> 00:26:11,640
what's special about this particular clip is Nolan is not

395
00:26:11,720 --> 00:26:13,440
the only cyborg playing Mario Kart.

396
00:26:13,480 --> 00:26:14,039
Speaker 2: In this clip.

397
00:26:14,480 --> 00:26:17,640
Speaker 12: We actually have a whole community of users as mentioned earlier,

398
00:26:17,839 --> 00:26:20,680
and this is literally five of our first users of

399
00:26:20,720 --> 00:26:33,400
neuraling playing Mario Kart together over call. Yeah, now, yeah,

400
00:26:33,880 --> 00:26:34,359
Mario Kart.

401
00:26:34,480 --> 00:26:35,279
Speaker 2: Is it's cool?

402
00:26:35,319 --> 00:26:37,680
Speaker 12: You know you're using one joystick and then you're clicking

403
00:26:37,680 --> 00:26:40,319
like a couple of buttons to throw items. What would

404
00:26:40,359 --> 00:26:43,119
be even cooler is what if you could control two

405
00:26:43,200 --> 00:26:47,319
joysticks at once simultaneously with your mind. When I'm about

406
00:26:47,359 --> 00:26:48,839
to show you, and I think this is for the

407
00:26:48,880 --> 00:26:51,599
first time someone playing a first person shooter game with

408
00:26:51,720 --> 00:26:55,440
a brain computer interface. This is Alex and r J

409
00:26:55,960 --> 00:26:59,440
playing Call of Duty, controlling one joystick to move and

410
00:26:59,480 --> 00:27:01,759
then the other stick to I think, point your gun,

411
00:27:01,799 --> 00:27:04,119
and then shooting people as a button. Uh, here's the

412
00:27:04,240 --> 00:27:07,559
larger here's Alex shooting another person.

413
00:27:09,079 --> 00:27:13,839
Speaker 3: Oh dear God, I don't know what to do, and

414
00:27:13,880 --> 00:27:16,920
I'm wanting to freaking shoot you. When I do, I

415
00:27:18,400 --> 00:27:19,960
notice shot in the face.

416
00:27:22,720 --> 00:27:24,000
Speaker 12: Now that we have a bit of a sense of

417
00:27:24,000 --> 00:27:27,119
what the bc I can do, a very important question

418
00:27:27,160 --> 00:27:29,559
to answer is how does this impact the day to

419
00:27:29,640 --> 00:27:33,279
day lives of the people that use it every day.

420
00:27:33,440 --> 00:27:36,200
So I'm about to show you a clip going back

421
00:27:36,240 --> 00:27:39,160
to Nolan for a second, where he talks. We simply

422
00:27:39,200 --> 00:27:42,519
just asked him randomly during the day how he enjoys

423
00:27:42,599 --> 00:27:44,960
using the bc I a couple of months ago, and

424
00:27:45,000 --> 00:27:46,200
this is his candid reaction.

425
00:27:46,720 --> 00:27:52,039
Speaker 13: I work basically all day from when I wake up

426
00:27:52,759 --> 00:27:55,400
trying to wake up at like six or seven am,

427
00:27:56,240 --> 00:27:59,920
and I'll do work until session. I'll do session, and

428
00:28:00,119 --> 00:28:06,279
then I'll work until you know, eleven, twelve pm or

429
00:28:06,319 --> 00:28:13,880
twelve am. While I'm doing, like I'm learning my languages,

430
00:28:14,279 --> 00:28:18,920
I'm learning my math, I'm like relearning all of my math.

431
00:28:19,640 --> 00:28:25,079
I am writing, I am doing the class that I

432
00:28:25,160 --> 00:28:28,079
sign up for, And I just I wanted to point

433
00:28:28,079 --> 00:28:31,880
out that, like, this is not something I would be

434
00:28:31,920 --> 00:28:35,079
able to do out like without the NEURALNK.

435
00:28:35,880 --> 00:28:37,839
Speaker 12: Next, I want to talk a bit about Brad. You

436
00:28:37,839 --> 00:28:40,559
guys may already know him as the ALS cyborg, and

437
00:28:40,799 --> 00:28:43,440
Brad also has ALS and what separates him from our other

438
00:28:43,480 --> 00:28:47,640
users is he's actually nonverbal, so he can't speak. Why

439
00:28:47,680 --> 00:28:50,799
this is pretty relevant is he relies, at least before

440
00:28:50,839 --> 00:28:54,000
the neuralink on an igaze machine to communicate, and a

441
00:28:54,000 --> 00:28:56,920
lot of I Gaze machines you can't use outdoors. You

442
00:28:57,079 --> 00:28:59,680
really need like a dark room. So what this means

443
00:28:59,720 --> 00:29:02,119
is for the last six years since Brad's been diagnosed

444
00:29:02,119 --> 00:29:06,359
with alas, he's really unable to leave his house. Now

445
00:29:06,359 --> 00:29:08,079
with the neuralink device, We're gonna show you a clip

446
00:29:08,400 --> 00:29:11,160
of him with his kids at the park, shot by

447
00:29:11,200 --> 00:29:12,240
Ashley Vance and the team.

448
00:29:12,440 --> 00:29:16,400
Speaker 3: Okay here ready, you absolutely do.

449
00:29:17,720 --> 00:29:20,119
Speaker 14: I am absolutely doing more with Neuralink than I was

450
00:29:20,119 --> 00:29:25,599
doing with I Games. I have been a batman for

451
00:29:25,640 --> 00:29:29,519
a long time, but I go outside now. Going outside

452
00:29:29,519 --> 00:29:32,759
has been a huge blessing for me, and I can

453
00:29:32,799 --> 00:29:34,359
control the computer with telepody.

454
00:29:34,720 --> 00:29:35,400
Speaker 9: Dad's watching.

455
00:29:36,039 --> 00:29:37,680
Speaker 15: Look, he's watching on the camera.

456
00:29:39,319 --> 00:29:40,200
Speaker 14: One of the arms.

457
00:29:40,880 --> 00:29:43,079
Speaker 12: The last user I want to talk about is Alex.

458
00:29:43,279 --> 00:29:46,079
You've seen some clips of him earlier. What's specially about

459
00:29:46,079 --> 00:29:49,160
Alex to me is he's a fellow left handed guy

460
00:29:49,160 --> 00:29:51,680
who writes some cursive all the time. And what he

461
00:29:51,799 --> 00:29:54,279
mentioned is since his spinal cord injury from like three

462
00:29:54,400 --> 00:29:58,480
four years ago. He's been unable to just like draw

463
00:29:58,680 --> 00:30:01,759
or write, and he always brags about how good his

464
00:30:01,759 --> 00:30:03,599
handwriting was. So we actually got to put it in

465
00:30:03,640 --> 00:30:05,640
the test. We gave him a robotic arm. And I

466
00:30:05,640 --> 00:30:07,519
think this is the first time you tried using the

467
00:30:07,599 --> 00:30:10,559
robotic arm to write anything. And this is a spouta

468
00:30:10,680 --> 00:30:14,039
version of writing at the Convoy trial and drawing something.

469
00:30:16,640 --> 00:30:17,200
Speaker 2: Sounds like that.

470
00:30:22,640 --> 00:30:26,000
Speaker 12: Now, Yeah, controlling a robotic arm is cool. Uh, but

471
00:30:26,079 --> 00:30:28,480
this one has a clamp. And what would be cooler

472
00:30:29,079 --> 00:30:31,759
is if you could decode the actual fingers, the actual wrist,

473
00:30:31,839 --> 00:30:33,720
all the muscles of the hand in real time.

474
00:30:35,119 --> 00:30:36,440
Speaker 2: Just in the past couple of weeks.

475
00:30:37,000 --> 00:30:39,640
Speaker 12: Uh, we were able to do that with Alex, and

476
00:30:39,759 --> 00:30:42,599
you're about to see him and his uncle, uh playing

477
00:30:42,640 --> 00:30:43,240
a game.

478
00:30:43,359 --> 00:30:50,720
Speaker 16: Rock Favor Scissors, Shoot damn It, rock vapor saissors, Shoot

479
00:30:51,839 --> 00:31:07,319
rock paper says or shoot rock paper saissor shoot that.

480
00:31:07,720 --> 00:31:21,279
Speaker 12: Some more cool controlling. Yeah, that's pretty dope.

481
00:31:21,279 --> 00:31:21,640
Speaker 11: I don't know.

482
00:31:26,960 --> 00:31:30,880
Speaker 12: And uh, controlling a robotic hand on screen is obviously

483
00:31:30,920 --> 00:31:35,440
not super helpful for most people. Fortunately, we have connections

484
00:31:35,599 --> 00:31:40,000
with Tesla who have the Optimist hand, and we're actually

485
00:31:40,119 --> 00:31:45,000
actively working on giving Alex an Optimist hand so that

486
00:31:45,079 --> 00:31:48,000
you could actually control it in his real life. And

487
00:31:48,160 --> 00:31:51,319
here's actual replay of the end of that video using

488
00:31:51,440 --> 00:31:53,480
Alex's neural signals on an optist hand.

489
00:31:53,799 --> 00:31:54,880
Speaker 2: Sean, if you want to play that.

490
00:32:02,160 --> 00:32:07,480
Speaker 11: Yeah, actually let me.

491
00:32:07,559 --> 00:32:10,319
Speaker 1: Let me maybe add a few things to that, which

492
00:32:10,359 --> 00:32:16,559
is so as we advance the neural link devices, your

493
00:32:16,720 --> 00:32:21,680
Chevelte actually have a full body control and sensors from

494
00:32:21,839 --> 00:32:26,400
an Optimist robot, So you could basically inhabit an Optimist robot.

495
00:32:26,440 --> 00:32:29,519
It's not just the hand, the whole the whole thing.

496
00:32:30,640 --> 00:32:35,359
So you could like basically mentally remote into an Optimist

497
00:32:36,200 --> 00:32:36,759
robot and.

498
00:32:37,000 --> 00:32:40,200
Speaker 2: And uh, be kind of cool. The future is going

499
00:32:40,279 --> 00:32:46,000
to be weird, but but but pretty cool.

500
00:32:47,599 --> 00:32:51,680
Speaker 1: And then now another thing that can be down also

501
00:32:51,799 --> 00:32:54,480
is like for people that have say lost a limb,

502
00:32:55,279 --> 00:32:59,039
lost an armor like or something like that, then we

503
00:32:59,079 --> 00:33:02,000
think in the future will be able to attach an

504
00:33:02,039 --> 00:33:06,599
optimus armor legs. And so you kind of like, I

505
00:33:06,720 --> 00:33:10,960
remember that scene from Star Wars where Luke Skywalker gets

506
00:33:10,960 --> 00:33:13,400
his hand, you know, chopped over with the lightslaber and

507
00:33:13,480 --> 00:33:16,400
he gets kind of a robot hand, and I think

508
00:33:16,480 --> 00:33:18,599
that's the kind of thing that we'll be able to

509
00:33:18,680 --> 00:33:21,720
do in the future working with the neuralink in Tesla,

510
00:33:22,559 --> 00:33:26,519
so that it goes far beyond just operating a robot hand,

511
00:33:26,519 --> 00:33:28,720
but replacing limbs and having kind of a whole body

512
00:33:28,799 --> 00:33:29,559
robot experience.

513
00:33:30,160 --> 00:33:30,880
Speaker 2: And then I think.

514
00:33:30,960 --> 00:33:35,279
Speaker 1: Another thing that will be possible, I think is very

515
00:33:35,400 --> 00:33:37,720
likely in the future, is to be able to bridge

516
00:33:39,119 --> 00:33:41,160
where the damaged neurons are, so you can take the

517
00:33:41,319 --> 00:33:46,000
signal from the brain and transmit that signal past where

518
00:33:46,039 --> 00:33:49,400
the neurons are damaged or strained to the rest of

519
00:33:49,440 --> 00:33:52,200
the body, so you could reanimate the body, so that

520
00:33:52,599 --> 00:33:54,839
if you have a neurallink implant in the brain and

521
00:33:54,880 --> 00:33:58,759
then one in the spinal cord, then you can actually

522
00:33:58,839 --> 00:34:03,119
bridge the signals and you could walk again and have

523
00:34:03,400 --> 00:34:04,400
full body functionality.

524
00:34:04,400 --> 00:34:05,880
Speaker 2: Obviously that's what people would prefer.

525
00:34:06,519 --> 00:34:08,840
Speaker 1: To be clear, we realize that that would be the

526
00:34:08,880 --> 00:34:13,480
preferred outcome, and so that even if you have a

527
00:34:13,519 --> 00:34:18,519
broken neck or if you saw we believe I'm actually

528
00:34:18,599 --> 00:34:20,519
at this point I'd say fairly confident that at some

529
00:34:20,559 --> 00:34:22,920
point in the future, well, we're able to restore full

530
00:34:22,920 --> 00:34:23,800
body functionality.

531
00:34:31,920 --> 00:34:34,559
Speaker 4: Yeah, so hello, hello everyone. My name is Near and

532
00:34:34,599 --> 00:34:38,400
I am leading the BCA application group. And I think

533
00:34:38,440 --> 00:34:41,960
the video is that just shared with you. I probably

534
00:34:42,320 --> 00:34:45,400
watch them maybe thousands of time, but still I get

535
00:34:45,400 --> 00:34:47,840
a goose bump every time I watch them. And I

536
00:34:47,880 --> 00:34:49,639
think this is one of the cool perks here at

537
00:34:49,639 --> 00:34:51,599
new Link when you get the job, is that you

538
00:34:51,679 --> 00:34:53,719
might get goose bump every week or maybe every few

539
00:34:53,760 --> 00:34:58,039
days in good weeks. And and this is really fun

540
00:34:59,360 --> 00:35:02,880
as an engine. It's really cool because you can build

541
00:35:02,880 --> 00:35:04,639
a new feature. You can build a new machine learning

542
00:35:04,719 --> 00:35:07,840
model and new software feature and test it on the

543
00:35:07,880 --> 00:35:12,079
same day with a participant and get feedback. And you

544
00:35:12,159 --> 00:35:15,880
already saw with our first device, Telepathy, that we can

545
00:35:16,079 --> 00:35:19,719
address a very diverse needs of the different users that

546
00:35:19,760 --> 00:35:23,679
we have for moving a cursor, to playing games, to

547
00:35:23,800 --> 00:35:27,360
move everybody can with multiple fingers, and we could not

548
00:35:27,440 --> 00:35:29,840
have done it without the new Link device. The neural

549
00:35:29,840 --> 00:35:32,679
Link device gives us something that no other device can

550
00:35:32,800 --> 00:35:37,119
give us, which is a single neuron recording from thousands

551
00:35:37,159 --> 00:35:41,239
of channels simultaneously. The Telepathy products is basically recording the

552
00:35:41,239 --> 00:35:44,679
neural activity from the small area and the motor context

553
00:35:44,920 --> 00:35:48,280
that involve in execution of hand and arm movements. But

554
00:35:48,360 --> 00:35:50,639
if we go only about two or three inches below

555
00:35:51,119 --> 00:35:55,320
there's another branaria that's involved in execution of speech and

556
00:35:55,440 --> 00:35:59,239
with the same device, with the same machine learning model architecture,

557
00:35:59,360 --> 00:36:02,159
the same soft through a pipeline, the same surgical robot.

558
00:36:02,440 --> 00:36:04,360
We can have a new application and we can do

559
00:36:04,400 --> 00:36:07,119
it very quickly. It's really interesting that if we can

560
00:36:07,159 --> 00:36:12,599
decode someone intention to speak silently and non vocal communication,

561
00:36:13,079 --> 00:36:16,159
we can use that to revolutionize the way we interact

562
00:36:16,159 --> 00:36:20,760
with computers, with technology and with information. Instead of typing

563
00:36:20,800 --> 00:36:23,719
with your finger or like moving the mouse or talking

564
00:36:23,719 --> 00:36:26,440
to your phone, you'll be able to interact with computer

565
00:36:26,599 --> 00:36:29,920
with the speed of thoughts. It will make this interaction

566
00:36:30,039 --> 00:36:33,639
much more, much faster, and much more intuitive. The computers

567
00:36:33,639 --> 00:36:36,079
will understand what you want to do. And we can

568
00:36:36,119 --> 00:36:39,159
also expand that to AI. We can now build an

569
00:36:39,199 --> 00:36:44,039
interface with AI that you will be able to achieve information,

570
00:36:44,239 --> 00:36:48,719
will be able to start our thoughts anywhere, anytime, privately

571
00:36:48,840 --> 00:36:53,440
and silently. Again, because we build a fundamental technology a

572
00:36:53,480 --> 00:36:56,639
platform and we do everything in house, we own the

573
00:36:56,679 --> 00:37:01,360
intense stack from neons to pixels on the users. Now

574
00:37:01,400 --> 00:37:04,840
I'll pass passy to rouse to talk about UI for be.

575
00:37:04,800 --> 00:37:15,880
Speaker 17: Said, Thank you nil each spike that our implant detects

576
00:37:15,960 --> 00:37:19,159
goes on a fairly remarkable journey to ultimately form a

577
00:37:19,199 --> 00:37:23,639
pixel on a participant's display, and that experience starts with,

578
00:37:24,000 --> 00:37:28,639
of course, unboxing, the very first time that a participant

579
00:37:28,920 --> 00:37:32,840
pairs to and meets their implant, this invisible part of

580
00:37:32,880 --> 00:37:38,880
their body and sees their own spikes materialize across the display.

581
00:37:39,840 --> 00:37:43,239
From there, they'll go into body mapping and actually imagine

582
00:37:43,239 --> 00:37:46,079
moving their arm again and get a feel for what

583
00:37:46,119 --> 00:37:49,800
feels natural to them and what doesn't, and they'll take

584
00:37:49,880 --> 00:37:55,760
that into calibration, using one of those motions to actually

585
00:37:56,000 --> 00:38:01,280
move a cursor again, iteratively refine their control as they

586
00:38:01,280 --> 00:38:07,199
go throughout this process, until finally they're teleported back to

587
00:38:07,280 --> 00:38:12,000
their desktop and can experience the magic of neural control

588
00:38:12,159 --> 00:38:16,559
for the very first time. And our control interfaces is

589
00:38:16,599 --> 00:38:21,320
where the OS integration that we do really shines, letting

590
00:38:21,400 --> 00:38:26,800
us adapt both control and feedback for every interaction. So

591
00:38:27,039 --> 00:38:30,960
for familiar interactions like scrolling, we can surface an indicator

592
00:38:31,039 --> 00:38:34,679
over the scrollable parts of the display at a touch

593
00:38:34,679 --> 00:38:37,960
of gravity to automatically pop a participant's cursor onto that

594
00:38:38,039 --> 00:38:39,199
indicator as they approach.

595
00:38:41,280 --> 00:38:43,400
Speaker 9: Show the actual velocities.

596
00:38:42,840 --> 00:38:45,039
Speaker 17: That we decode inside of it and add a bit

597
00:38:45,079 --> 00:38:48,000
of momentum to those velocities to carry them forward as

598
00:38:48,000 --> 00:38:52,960
they glide across the page. They are also unique interactions

599
00:38:52,960 --> 00:38:55,760
that we need to solve for in this space. For example,

600
00:38:55,760 --> 00:38:58,400
when a participant is watching a movie or just talking

601
00:38:58,440 --> 00:39:02,199
to somebody next to them, the brain is very active

602
00:39:02,280 --> 00:39:05,719
still and that activity can actually induce motion in the cursor,

603
00:39:05,960 --> 00:39:09,360
distracting them from that moment. So when a participant wants

604
00:39:09,400 --> 00:39:11,360
to just get their cursor out of the way, they

605
00:39:11,360 --> 00:39:13,719
can push it into the edge of the display to

606
00:39:13,760 --> 00:39:16,679
park it there, and of course we add gravity to

607
00:39:16,760 --> 00:39:19,639
sort of hold it still, but they can push it

608
00:39:19,719 --> 00:39:22,920
out with either just a firm push or in this case,

609
00:39:22,960 --> 00:39:27,119
a gesture, and of course it's what it goes. Without

610
00:39:27,119 --> 00:39:30,000
saying that, all of these control interfaces are designed hand

611
00:39:30,000 --> 00:39:32,559
in hand with our participants, so huge shout out to

612
00:39:32,599 --> 00:39:35,880
both Nolan and Brad for helping us design these two

613
00:39:36,079 --> 00:39:39,719
and those control interfaces, of course extend typing. We have

614
00:39:39,760 --> 00:39:42,679
a great software keyboard that does everything you'd expect it to,

615
00:39:43,079 --> 00:39:45,440
popping up when a participant clicks on a text field,

616
00:39:45,679 --> 00:39:48,039
giving them feedback about the click along the surface of

617
00:39:48,079 --> 00:39:57,119
the key and supporting both dictation and swipe.

618
00:39:58,280 --> 00:40:01,800
Speaker 18: Hi everyone, I'm Harrison and engineer here at Neuralink, and

619
00:40:01,880 --> 00:40:04,320
I must say being an mL engineer at Neuralink is

620
00:40:04,320 --> 00:40:06,079
a bit like being a kid in a candy store.

621
00:40:08,480 --> 00:40:11,000
When you think of the inputs to most mL systems

622
00:40:11,000 --> 00:40:13,800
out there, you might think of pixels of tokens or

623
00:40:13,840 --> 00:40:17,480
of a user's Netflix watch history. The input to our

624
00:40:17,519 --> 00:40:21,800
systems is a little different. It is pure raw brain power.

625
00:40:22,199 --> 00:40:24,119
And when we think about the mL systems we can

626
00:40:24,119 --> 00:40:27,000
build here at Neuralink, really we're limited by our imagination

627
00:40:27,119 --> 00:40:30,480
and our creativity. There's no reason our mL systems can't

628
00:40:30,559 --> 00:40:33,760
do anything that the human brain can do, such as

629
00:40:33,920 --> 00:40:37,199
controlling a phone, typing, or even gaming.

630
00:40:38,559 --> 00:40:39,280
Speaker 6: Right here to my.

631
00:40:39,280 --> 00:40:42,159
Speaker 18: Left is actual footage of Alex, one of our participants,

632
00:40:42,199 --> 00:40:45,840
playing at first person shooter against RJ, another one of

633
00:40:45,880 --> 00:40:49,639
our participants. Now, for those unfamiliar with first person shooters,

634
00:40:49,880 --> 00:40:52,840
this is not a trivial feat. It requires two fully

635
00:40:52,840 --> 00:40:57,360
independent joysticks or four continuous degrees of control, as well

636
00:40:57,360 --> 00:41:02,599
as multiple reliable buttons. Now, contrary to popular belief, the

637
00:41:02,639 --> 00:41:07,159
Neuralink does not simply read people's minds it's simply reading

638
00:41:07,239 --> 00:41:09,880
neuronal activations corresponding to motor intent.

639
00:41:10,400 --> 00:41:11,400
Speaker 9: So one of the fun.

640
00:41:11,320 --> 00:41:14,320
Speaker 18: Challenges with this project was figuring out which motions we're

641
00:41:14,360 --> 00:41:17,000
going to give out to the joystick. We started with

642
00:41:17,039 --> 00:41:19,840
the typical left thumb and right thumb, but quickly found

643
00:41:19,920 --> 00:41:23,760
the dominant hand overshadow the non dominant hand. My personal

644
00:41:23,760 --> 00:41:26,199
favorite is we had one of our participants imagine walking

645
00:41:26,320 --> 00:41:28,880
for the left joystick and aiming for the right joystick,

646
00:41:29,159 --> 00:41:33,000
so in game, they were simply doing naturalistic motions like

647
00:41:33,039 --> 00:41:35,239
you might do in virtual reality in Ready Player one,

648
00:41:35,599 --> 00:41:37,079
and that was really cool to watch.

649
00:41:37,719 --> 00:41:39,920
Speaker 9: What we ended up on was the thumb for.

650
00:41:39,880 --> 00:41:42,880
Speaker 18: The left joystick and the wrist for the right joystick,

651
00:41:43,239 --> 00:41:46,079
and I challenged the audience to try to replicate their motions.

652
00:41:46,320 --> 00:41:47,920
I'm really in all of them being able to pull

653
00:41:47,960 --> 00:41:50,159
this off. I want to talk a bit about the

654
00:41:50,199 --> 00:41:52,360
progress to our cursor calibration experience.

655
00:41:53,079 --> 00:41:53,639
Speaker 2: To my left.

656
00:41:53,679 --> 00:41:56,559
Speaker 18: Here you can see RJ completing his first ever cursor

657
00:41:56,599 --> 00:41:59,440
calibration with a redesigned open up flow where he first

658
00:41:59,519 --> 00:42:01,519
gathering for about his intent and how to map the

659
00:42:01,559 --> 00:42:04,199
neural activity to the first time he controls a cursor

660
00:42:04,480 --> 00:42:06,719
to the final product where he has smooth and fluid

661
00:42:06,920 --> 00:42:11,840
control of his computer. And most remarkably, this experience took

662
00:42:11,880 --> 00:42:15,679
only fifteen minutes from a start to finish, fifteen minutes

663
00:42:15,679 --> 00:42:24,599
from not fifteen minutes from no control to fluid computer use.

664
00:42:26,320 --> 00:42:28,000
Contrast that to a year and a half ago with

665
00:42:28,039 --> 00:42:30,679
P one, where that was multiple hours to get to

666
00:42:30,679 --> 00:42:33,480
the same level of control and several engineers standing around

667
00:42:33,519 --> 00:42:36,199
a table pulling their hair out. There was virtually no

668
00:42:36,320 --> 00:42:38,559
need for neuralink engineers to even be at this session.

669
00:42:38,920 --> 00:42:41,039
This was basically it out of the box experience for

670
00:42:41,079 --> 00:42:48,199
our participant, and even more remarkably, we're continuing to smash

671
00:42:48,280 --> 00:42:53,119
day one records, with RJ being able to achieve seven BPS.

672
00:42:52,719 --> 00:42:54,440
Speaker 9: On his very first day with a neuralink.

673
00:42:58,760 --> 00:43:01,760
Speaker 18: Now, such an effective and fishing calibration process is only

674
00:43:01,760 --> 00:43:05,320
made possible by high fidelity estimations of a user intention

675
00:43:05,920 --> 00:43:10,119
or labels. And to briefly illustrate just how challenging a

676
00:43:10,159 --> 00:43:13,000
problem that is, this is an animation of myself trying

677
00:43:13,000 --> 00:43:16,440
to draw circles on my desktop with a mouse. Now,

678
00:43:16,519 --> 00:43:20,159
the task was simple, draw uniform circles at a constant speed,

679
00:43:20,320 --> 00:43:23,519
repeatedly and as you can see by that animation, I

680
00:43:23,559 --> 00:43:24,480
am horrible at that.

681
00:43:25,000 --> 00:43:25,599
Speaker 2: Even though my.

682
00:43:25,559 --> 00:43:29,599
Speaker 18: Intent was pretty obvious unambiguous, the execution was really poor.

683
00:43:29,920 --> 00:43:33,039
There is a ton of variation in both speed and

684
00:43:33,920 --> 00:43:37,679
the shape itself. To visualize this a little differently, each

685
00:43:37,800 --> 00:43:40,239
row here is one of those circles unwound in time

686
00:43:40,480 --> 00:43:42,920
with synchronized starts, and you can just see how much

687
00:43:43,000 --> 00:43:45,079
variation there is in the timing of each circle as

688
00:43:45,079 --> 00:43:47,119
well as what I'm doing at any given point in time.

689
00:43:49,760 --> 00:43:53,199
Orthogonal to the labeling problem is neural nonstationarity, or the

690
00:43:53,239 --> 00:43:56,960
tendency of neural signals to drift over time. And I

691
00:43:56,960 --> 00:43:59,960
think that's honestly a beautiful thing, right If your neurals

692
00:44:00,000 --> 00:44:02,719
signals didn't drift, you couldn't grow. When you wake up

693
00:44:02,760 --> 00:44:04,800
the next day, you're not the same person you were

694
00:44:04,800 --> 00:44:07,920
the day before. You've learned, you've grown, You've changed, and

695
00:44:08,000 --> 00:44:11,519
so too must your neural data change. This animation here

696
00:44:11,599 --> 00:44:14,119
is a simple illustration of the learned representation by the

697
00:44:14,159 --> 00:44:16,679
decoder and how it drifts the further away we get

698
00:44:16,719 --> 00:44:18,920
from the day it was trained on. This is one

699
00:44:18,920 --> 00:44:20,480
of the key challenges we need to solve here At

700
00:44:20,519 --> 00:44:23,079
neuralink to unlock fluid and a product level experience for

701
00:44:23,119 --> 00:44:24,119
our users.

702
00:44:32,760 --> 00:44:41,599
Speaker 15: Everyone here, everyone, My name is Joey. Blind Sight is

703
00:44:41,639 --> 00:44:44,280
our project to build a visual prosthesis to help the

704
00:44:44,320 --> 00:44:48,920
blind see again. Users would wear a pair of glasses

705
00:44:50,559 --> 00:44:54,239
with an embedded camera and receive an implant in their

706
00:44:54,320 --> 00:44:58,679
visual cortex. Scenes from the environment are recorded by the

707
00:44:58,719 --> 00:45:03,639
camera and processed in the patterns of stimulation delivered to

708
00:45:03,760 --> 00:45:10,679
the brain, causing visual perception and restoring functionality. Now blind

709
00:45:10,679 --> 00:45:13,599
Site will be enabled by placing our implant into visual cortex.

710
00:45:13,880 --> 00:45:15,960
This is a new brain area for us, and this

711
00:45:16,000 --> 00:45:21,119
brings new opportunities and challenges. So the surface of the

712
00:45:21,159 --> 00:45:24,280
brain for visual cortex represents just a few degrees of

713
00:45:24,360 --> 00:45:28,559
angle in the center of the visual field. Larger fields

714
00:45:28,559 --> 00:45:31,559
of view are represented deep within the cortical folds of

715
00:45:31,599 --> 00:45:36,239
the calcarine fissure. Our threads are able to access these

716
00:45:36,320 --> 00:45:41,119
deeper structures, providing the possibility of restoring vision over a functional,

717
00:45:41,199 --> 00:45:44,679
useful visual field. So the N one implant has had

718
00:45:44,880 --> 00:45:49,440
experimental stimulation capabilities for quite some time, but our new

719
00:45:49,760 --> 00:45:52,880
S two chip is designed from the ground up for stimulation.

720
00:45:53,880 --> 00:45:58,440
It provides over sixteen hundred channels of electrical stimulation, high

721
00:45:58,599 --> 00:46:02,920
dynamic range recording abilities, and a wide range of microstimulation

722
00:46:03,039 --> 00:46:08,079
currents and voltages. We can achieve these capabilities because we're

723
00:46:08,159 --> 00:46:13,360
vertically integrated and we design this CUSTOMASIC in house. Similarly,

724
00:46:13,599 --> 00:46:16,639
we design and fabricate our electrode threads in house, and

725
00:46:16,719 --> 00:46:20,519
here you can see one of our standard threads designed

726
00:46:20,519 --> 00:46:24,840
for recording in an electron micrograph. For blind site, our

727
00:46:24,840 --> 00:46:28,239
requirements are a little different and our vertical integration allows

728
00:46:28,280 --> 00:46:31,440
us to rapidly iterate on the design and manufacturing of

729
00:46:31,440 --> 00:46:35,559
these threads for this new purpose. So here I'm using

730
00:46:35,599 --> 00:46:40,159
red arrows to highlight the electrode contacts which are optimized

731
00:46:40,159 --> 00:46:42,559
for stimulation, and as you can see, they're a little

732
00:46:42,599 --> 00:46:45,519
bit larger, which results in a lower electrical impedance for

733
00:46:45,639 --> 00:46:52,159
safe and effective charge delivery, which is important for blind site. Now,

734
00:46:52,159 --> 00:46:55,159
how can we calibrate our implant for blind site? So

735
00:46:55,239 --> 00:46:59,360
here's one way. We stimulate on the array, picking say

736
00:46:59,480 --> 00:47:04,039
three different channels. The user perceives something say three spots

737
00:47:04,039 --> 00:47:07,079
of light somewhere in their visual field and points at them.

738
00:47:07,920 --> 00:47:10,599
We track their arm and eye movements and repeat this

739
00:47:10,679 --> 00:47:15,280
process for each of the channels on the array, and

740
00:47:15,360 --> 00:47:18,679
here's what a simulated example of a blind site vision

741
00:47:18,719 --> 00:47:32,880
could look like after calibration. Now I showed you how

742
00:47:32,880 --> 00:47:35,239
for blind site we need to insert threads deeper into

743
00:47:35,280 --> 00:47:38,639
the brain than we have previously and doing this requires

744
00:47:38,800 --> 00:47:42,280
state of the art medical imaging. So we worked with

745
00:47:42,280 --> 00:47:44,280
Semens to get some of the best scanners on earth.

746
00:47:45,000 --> 00:47:48,280
We built out our imaging core from scratch in the

747
00:47:48,320 --> 00:47:50,920
past year. Actually it was faster than that. It was

748
00:47:50,960 --> 00:47:54,760
about four months from dirt to done. Since bringing the

749
00:47:54,760 --> 00:47:58,079
scanners online, we scanned over fifty internal participants, building out

750
00:47:58,079 --> 00:48:02,039
a database of human structural and functional anatomy. What can

751
00:48:02,079 --> 00:48:05,360
we do with the imaging information from these scanners? So

752
00:48:05,519 --> 00:48:08,599
medical imaging can be used for surgical placement. It lets

753
00:48:08,679 --> 00:48:11,440
us parcelate out brain regions by their function, and we

754
00:48:11,559 --> 00:48:15,079
use our imaging capabilities to refine the placement for telepathy.

755
00:48:15,559 --> 00:48:17,920
It also gives us the capability of target new brain

756
00:48:18,039 --> 00:48:20,719
regions for future products such as blind site or speech

757
00:48:20,760 --> 00:48:24,719
pres thesis, and we're working towards more capabilities. So one

758
00:48:24,760 --> 00:48:28,159
click automated planning of surgery from functional images to robot

759
00:48:28,239 --> 00:48:31,400
insertion targets. Here you can see a screen capture from

760
00:48:31,440 --> 00:48:33,159
one of our in house tooling to do end to

761
00:48:33,239 --> 00:48:35,960
end surgical planning. You can see a region of motor

762
00:48:35,960 --> 00:48:39,960
cortex known as hand knob and the thread trajectory plans

763
00:48:39,960 --> 00:48:43,960
that'll be sent directly to the robot. This is a

764
00:48:44,000 --> 00:48:47,280
really incredible degree of automation that's only possible because we're

765
00:48:47,280 --> 00:48:49,599
controlling the system from one end to the other.

766
00:48:54,159 --> 00:48:57,039
Speaker 10: My name is John and I lead the robot mechanical team.

767
00:48:57,159 --> 00:49:00,000
This is our current R one robot. It was used

768
00:49:00,079 --> 00:49:04,039
to implant the first seven participants. This robot works really well,

769
00:49:04,079 --> 00:49:06,119
but it has a few flaws, one of which is

770
00:49:06,159 --> 00:49:09,360
the cycle time is rather slow, so to insert each

771
00:49:09,360 --> 00:49:12,679
thread it takes in the best case scenario seventeen seconds,

772
00:49:13,119 --> 00:49:16,519
and many cases external disturbances cause us to have to

773
00:49:16,559 --> 00:49:20,280
retry to grasp that thread and then reinsert it. To

774
00:49:20,360 --> 00:49:24,400
scale our number of neurons access through higher channel acount

775
00:49:24,559 --> 00:49:26,800
increase numbers of threads, we need to have a much

776
00:49:26,840 --> 00:49:30,960
faster cycle time. So let me introduce our next generation robot,

777
00:49:31,800 --> 00:49:43,480
which is right here. Through rethinking the way that we

778
00:49:43,519 --> 00:49:45,480
hold the implant in front of the robot, by holding

779
00:49:45,519 --> 00:49:48,039
it directly in front on the robot, head. We will

780
00:49:48,039 --> 00:49:51,840
achieve an eleven times cycle time improvement, so each thread

781
00:49:51,880 --> 00:49:52,400
takes one and.

782
00:49:52,360 --> 00:49:53,239
Speaker 2: A half seconds.

783
00:49:53,719 --> 00:49:58,039
Speaker 10: We also scale up a lot of surgery workflow process

784
00:49:58,079 --> 00:50:01,519
improvements through deleting the separate an operator station and implant

785
00:50:01,559 --> 00:50:05,320
stand Now the outside of the robot looks pretty similar

786
00:50:05,360 --> 00:50:08,480
between the two, but what's inside that really counts. Each

787
00:50:08,519 --> 00:50:11,239
system has been redesigned from the ground up with a

788
00:50:11,280 --> 00:50:16,000
focus on re liability, manufacturability, serviceability, and using a lot

789
00:50:16,000 --> 00:50:18,960
of our vertical integration techniques. It's enabled us to have

790
00:50:19,000 --> 00:50:21,840
a lot more control of the system and to end

791
00:50:22,800 --> 00:50:25,199
now that fast cycle time doesn't mean much if it's

792
00:50:25,199 --> 00:50:27,880
not compatible with a significant portion of the human population.

793
00:50:28,679 --> 00:50:32,519
Prior to each surgery, we scan our participants anatomy and

794
00:50:32,639 --> 00:50:35,159
ensure that they will be compatible with the robot and

795
00:50:35,239 --> 00:50:38,960
vice versa. Unfortunately, the robot isn't compatible with everyone, so

796
00:50:39,000 --> 00:50:41,280
we had to extend the reach of the needle in

797
00:50:41,320 --> 00:50:43,679
the next generation robot, and now we're compatible with more

798
00:50:43,719 --> 00:50:46,519
than ninety nine percent of the human population. We've also

799
00:50:46,599 --> 00:50:49,400
increased the depth that the needle can insert threads. Now

800
00:50:49,400 --> 00:50:51,440
we can reach more than fifty millimeters from the surface

801
00:50:51,440 --> 00:50:55,360
of the brain, accessing and enabling new indications. We have

802
00:50:55,440 --> 00:50:59,039
to produce a ton of custom sterile components for each surgery.

803
00:50:59,079 --> 00:51:01,440
Speaker 6: We actually supple more than twenty of these parts.

804
00:51:01,960 --> 00:51:04,159
Speaker 10: Many of these parts are made through traditional C and

805
00:51:04,199 --> 00:51:06,840
C manufacturing capabilities, which we do just on the other

806
00:51:06,880 --> 00:51:10,039
side of this wall actually, and some custom developed processes

807
00:51:10,119 --> 00:51:13,000
like this pemptosecond laser milling used to manufacture the tip

808
00:51:13,039 --> 00:51:16,800
of the needle. Now, these processes take quite a bit

809
00:51:16,840 --> 00:51:19,159
of time, effort, and cost, So let's take a look

810
00:51:19,199 --> 00:51:22,440
at how we're going to reduce costs and time for

811
00:51:22,559 --> 00:51:25,920
one of the components. So the current needle cartridge has

812
00:51:25,960 --> 00:51:28,440
a total cycle time of about twenty four hours and

813
00:51:28,639 --> 00:51:31,199
the machine components cost about three hundred and fifty dollars.

814
00:51:31,800 --> 00:51:33,639
The final assembly is performed by a set of like

815
00:51:33,719 --> 00:51:36,360
highly skilled technicians. They have to glue a one hundred

816
00:51:36,400 --> 00:51:40,880
and fifty micron diameter canua onto this wireedm machined, seainless

817
00:51:40,880 --> 00:51:43,360
steeled base plate. They have to electro polish a forty

818
00:51:43,400 --> 00:51:45,840
micron wire into a sharp taper, and then they have

819
00:51:45,880 --> 00:51:49,119
to thread that forty micron wire into a sixty micron

820
00:51:49,159 --> 00:51:52,559
hole in the canua. This is done manually and then

821
00:51:52,599 --> 00:51:54,920
they finally have to laser weld all the components together.

822
00:51:57,320 --> 00:52:00,559
Next generation needle cartridge takes only thirty minute of cycle

823
00:52:00,599 --> 00:52:03,199
time and fifteen dollars in component. We were able to

824
00:52:03,239 --> 00:52:06,599
delete the wire idium machined base plate and the canuo

825
00:52:06,639 --> 00:52:09,280
Gulian step by switching to an insert mold it component.

826
00:52:09,360 --> 00:52:11,239
So we get a box of these base plates with

827
00:52:11,320 --> 00:52:13,920
the canules already installed, for like a thousand of them

828
00:52:13,960 --> 00:52:17,000
for like a couple five ten dollars apiece. We also

829
00:52:17,119 --> 00:52:20,159
deleted the electropolishing step with the revised needle tip geometry,

830
00:52:20,519 --> 00:52:23,719
which is also compatible with inserting the threads through the dura.

831
00:52:24,679 --> 00:52:27,960
We have a few revised manufacturing techniques to delete the

832
00:52:28,000 --> 00:52:31,079
manual threading through basically a funnel. Rather simple, but it

833
00:52:31,079 --> 00:52:33,320
has been a big impact. And then we're able to

834
00:52:33,320 --> 00:52:35,119
delete the laser welding through using crimping.

835
00:52:41,360 --> 00:52:47,599
Speaker 2: By Hi.

836
00:52:47,679 --> 00:52:50,199
Speaker 19: I'm Julian. I'm one of the leads on the implant team.

837
00:52:51,039 --> 00:52:54,280
So the way humans communicate today, if they want to

838
00:52:54,400 --> 00:52:57,440
output information is by using their hands and their voice

839
00:52:57,480 --> 00:52:59,880
as I'm doing right now, And if you want to

840
00:53:00,000 --> 00:53:02,800
receive information, you use your ears and your eyes, and

841
00:53:02,840 --> 00:53:06,599
of course that's how you're receiving this very torque but

842
00:53:07,079 --> 00:53:11,039
we've built this implant, and this implant is very special

843
00:53:11,280 --> 00:53:14,159
because it is the first time that we're able to

844
00:53:14,199 --> 00:53:17,920
add a completely new mode of data transfer into and

845
00:53:17,960 --> 00:53:20,880
out of the brain. If you look at this device

846
00:53:21,000 --> 00:53:24,039
in a nutshell, it's really just sampling voltages in the

847
00:53:24,079 --> 00:53:27,760
brain and sending them over radio. But if you zoom

848
00:53:27,760 --> 00:53:29,840
out and look at the system from end to end,

849
00:53:30,280 --> 00:53:34,079
what you actually see is that we're connecting your brain

850
00:53:34,320 --> 00:53:37,920
or a biological neural net to a machine learning model

851
00:53:38,199 --> 00:53:41,000
or a silicon neural net on the right hand side.

852
00:53:41,440 --> 00:53:45,719
And I actually think this is really elegant because the

853
00:53:45,800 --> 00:53:48,119
machine learning model on the right hand side is in

854
00:53:48,159 --> 00:53:51,760
fact inspired by neurons on the left hand side, and

855
00:53:51,800 --> 00:53:55,519
so in some sense, we're really extending the fundamental substrate

856
00:53:55,679 --> 00:53:57,960
of the brain for the first time. We're able to

857
00:53:58,000 --> 00:54:00,360
do this in a mass market product. That's a very

858
00:54:00,440 --> 00:54:11,079
very special piece of hardware. So these are some of

859
00:54:11,159 --> 00:54:15,679
the first implants that we ever built. There are electrodes

860
00:54:15,800 --> 00:54:19,159
that were made with our in house lithography tools. We

861
00:54:19,280 --> 00:54:21,519
have custom A six that we also designed in house,

862
00:54:21,960 --> 00:54:24,519
and this was really a platform for us to develop

863
00:54:24,559 --> 00:54:27,960
the technology that allows us to sense micro level vaults

864
00:54:28,000 --> 00:54:32,079
in the brain across thousands of channels simultaneously. We learn

865
00:54:32,159 --> 00:54:34,360
a lot from this. But as you'll notice in the

866
00:54:34,440 --> 00:54:38,639
right to images, there are USBC connectors on these devices.

867
00:54:38,960 --> 00:54:43,880
These were not really the most implantable implants. This next

868
00:54:43,920 --> 00:54:47,480
set of images are the wireless implants, and there was

869
00:54:47,519 --> 00:54:50,719
a complete evolution that we went through to add the battery,

870
00:54:50,960 --> 00:54:55,079
the antenna, the radio, and to make it actually fully implantable.

871
00:54:55,280 --> 00:54:59,880
Once it's implanted, it's completely invisible. It's very compact, it's modular,

872
00:55:00,079 --> 00:55:01,840
and it's a general platform that you can use in

873
00:55:01,840 --> 00:55:06,519
many places in the brain. Going from that top row

874
00:55:06,840 --> 00:55:10,159
to the bottom row is very challenging. The implant you

875
00:55:10,199 --> 00:55:12,079
see on the bottom right here is in fact the

876
00:55:12,119 --> 00:55:15,599
device that we have working in seven participants today and

877
00:55:15,639 --> 00:55:19,559
it's augmenting their brain every day and restoring their autonomy.

878
00:55:19,840 --> 00:55:23,199
But getting to that point involved a huge number of

879
00:55:23,199 --> 00:55:26,880
formidable engineering challenges. We first had to make hometic enclosure,

880
00:55:27,159 --> 00:55:30,760
passing one thousand separate conductors through the enclosure of the device.

881
00:55:31,239 --> 00:55:33,639
We had to figure out how to make charging seamless

882
00:55:33,639 --> 00:55:35,960
and work with very tight thermal constraints in a very

883
00:55:36,039 --> 00:55:38,679
very small area, and then we also had to scale

884
00:55:38,760 --> 00:55:41,559
up our testing infrastructure so that we could support large

885
00:55:41,559 --> 00:55:44,960
scale manufacturing and very safe devices and have confidence in

886
00:55:45,000 --> 00:55:46,000
our iteration cycle.

887
00:55:47,199 --> 00:55:48,000
Speaker 9: So what's next.

888
00:55:48,760 --> 00:55:51,960
Speaker 19: We're going to be increasing our manufacturing so that we

889
00:55:52,000 --> 00:55:55,039
don't just produce a certain like a small number of

890
00:55:55,039 --> 00:55:57,440
implants per year, but thousands and then eventually millions of

891
00:55:57,440 --> 00:56:00,280
implants per year. We're also going to be increase seeing

892
00:56:00,440 --> 00:56:04,880
channel count. More channels means more neurons are sensed, which

893
00:56:04,920 --> 00:56:08,400
means more capabilities. In some sense, we often think a

894
00:56:08,440 --> 00:56:11,559
lot about the Moore's law of neurons that were interacting

895
00:56:11,599 --> 00:56:14,760
with and in the same way that Moore's law propelled

896
00:56:14,800 --> 00:56:18,920
forward many subsequent revolutions in computing, we think that sensing

897
00:56:18,960 --> 00:56:22,159
more and more neurons will also completely redefine how we

898
00:56:22,239 --> 00:56:24,800
interact with computers and reality at large.

899
00:56:26,599 --> 00:56:28,199
Speaker 9: I want to leave you with one final thought.

900
00:56:29,239 --> 00:56:31,800
Speaker 19: When I was a child, I used a fifty six

901
00:56:31,880 --> 00:56:36,199
kilo bit moodem to access the Internet. If you remember

902
00:56:36,239 --> 00:56:38,239
what it's like, you would go to a website.

903
00:56:38,519 --> 00:56:43,199
Speaker 2: You're lucky, You're lucky, bastard. Yeah, when I was a child,

904
00:56:43,719 --> 00:56:45,000
we had acoustic couplers.

905
00:56:45,119 --> 00:56:48,480
Speaker 1: Oh, yeah, okay, so just just beat at each other.

906
00:56:48,760 --> 00:56:49,519
Speaker 9: Yeah.

907
00:56:49,000 --> 00:56:53,000
Speaker 19: The first modem was the acoustic Coppler. Incredible device honestly.

908
00:56:54,519 --> 00:56:56,599
But then if you I guess, if you're my age,

909
00:56:56,599 --> 00:57:04,760
you started with fifty six kit modem and you would

910
00:57:04,800 --> 00:57:06,800
go to a website and like there would be an

911
00:57:06,840 --> 00:57:10,400
image and it would it would scroll like slowly, it

912
00:57:10,440 --> 00:57:12,400
was loading pixel by pixel on the screen.

913
00:57:12,880 --> 00:57:15,199
Speaker 9: So that's what it's like to be bandwidth limited.

914
00:57:15,760 --> 00:57:19,119
Speaker 19: Now I imagine using the current Internet with that same modem.

915
00:57:19,440 --> 00:57:22,280
It's like it's inconceivable, it would be impossible to do.

916
00:57:23,320 --> 00:57:26,559
Speaker 9: So what broadband Internet did to.

917
00:57:26,519 --> 00:57:29,400
Speaker 19: The fifty six kilo bit modem is what this hardware

918
00:57:29,519 --> 00:57:32,000
is going to do to the brain. We're trying to

919
00:57:32,079 --> 00:57:35,320
drastically expand the amount of bandwidth that you have access

920
00:57:35,360 --> 00:57:39,159
to to have a much richer experience and superhuman capabilities.

921
00:57:46,719 --> 00:57:48,920
Speaker 5: So I guess, just to kind of close out and

922
00:57:49,119 --> 00:57:54,760
to recap today, neuralink is working reliably and has already

923
00:57:54,840 --> 00:57:58,400
changed the lives of seven participants and making a real impact.

924
00:58:00,519 --> 00:58:03,000
And our next milestone is to go to market and

925
00:58:03,119 --> 00:58:07,320
enable scaling of this technology to thousands of people, as

926
00:58:07,360 --> 00:58:11,119
well as expand functionalities beyond just the movement to enable

927
00:58:11,519 --> 00:58:17,159
robotic sophisticated robotic arm control, speech vision, give site back,

928
00:58:17,360 --> 00:58:18,320
and even.

929
00:58:18,440 --> 00:58:19,599
Speaker 6: Getting to the speed of thought.

930
00:58:20,440 --> 00:58:22,559
Speaker 5: I hope you got a good sort of sample of

931
00:58:22,639 --> 00:58:26,960
our technology stack and the challenges that we have. And

932
00:58:27,280 --> 00:58:30,280
I like to handle over the mic to Elon for

933
00:58:30,320 --> 00:58:31,360
any closing remarks.

934
00:58:33,280 --> 00:58:37,800
Speaker 1: Well, we're trying to give you a sense of the

935
00:58:38,119 --> 00:58:38,599
depth of.

936
00:58:38,559 --> 00:58:39,840
Speaker 2: Talent at your link.

937
00:58:39,880 --> 00:58:42,679
Speaker 1: There's a lot of really smart people working on a

938
00:58:42,679 --> 00:58:46,679
lot of important problems. This is one of the most

939
00:58:46,679 --> 00:58:50,480
difficult things to to actually succeed in creating and have

940
00:58:50,599 --> 00:58:53,360
at work and work at scale and be reliable and

941
00:58:53,400 --> 00:58:58,039
available for millions of people at an affordable price. So

942
00:58:59,679 --> 00:59:03,719
super heart problem and we'd like to have you come

943
00:59:03,880 --> 00:59:04,719
join and help.

944
00:59:04,599 --> 00:59:06,400
Speaker 2: Us solve it. Thank you.