1 00:00:01,920 --> 00:00:04,520 Hey everybody. Welcome back to the Elon Musk 2 00:00:04,520 --> 00:00:07,120 Podcast. This is a show where we discuss 3 00:00:07,320 --> 00:00:11,520 the critical crossroads, the Shape, SpaceX, Tesla X, The 4 00:00:11,520 --> 00:00:15,600 Boring Company, and Neuralink. I'm your host, Will Walden. 5 00:00:19,720 --> 00:00:21,920 Hello everybody. My name is Alex Toddling. 6 00:00:21,920 --> 00:00:25,200 I'm the second participant in the Neuralink study, but I'm 7 00:00:25,200 --> 00:00:35,480 here to count us down to the demo in 54321. 8 00:00:46,480 --> 00:00:53,520 Hi everyone, welcome to the Neurolink presentation. 9 00:00:53,520 --> 00:00:57,680 This is an update for the progress of the neural link 10 00:00:57,680 --> 00:00:58,880 team. It's been an incredible amount 11 00:00:58,880 --> 00:01:03,400 of progress. This is we're going to start off 12 00:01:03,400 --> 00:01:07,760 high level, generally describing what neural links doing. 13 00:01:08,160 --> 00:01:11,120 And then we're going to get have a very deep technical dive so 14 00:01:11,120 --> 00:01:14,080 you can actually get an understanding of what exactly 15 00:01:14,080 --> 00:01:18,680 we're doing at a granular level and what we can do to enhance 16 00:01:18,680 --> 00:01:22,320 human capabilities and ultimately build a great future 17 00:01:22,320 --> 00:01:26,080 for humanity. So that's a that's a neural 18 00:01:26,080 --> 00:01:29,240 inspiring. It's funny thing that me talking 19 00:01:29,240 --> 00:01:34,240 right now is a bunch of neurons firing that then result in 20 00:01:34,240 --> 00:01:39,440 speech that you hear that cause neurons to fire in your brain. 21 00:01:42,000 --> 00:01:45,440 Yeah, part of part of this presentation is about 22 00:01:45,440 --> 00:01:50,920 demystifying the the brain. It is a remarkable organ. 23 00:01:50,920 --> 00:01:54,880 I mean, we are the brain basically when you say you, that 24 00:01:54,880 --> 00:01:59,040 really is you're the brain. Like you can, you can get a, a 25 00:01:59,040 --> 00:02:02,760 heart transplant, you can get a kidney transplant, but I don't 26 00:02:02,760 --> 00:02:04,440 know anyone who's gotten a brain transplant. 27 00:02:05,640 --> 00:02:13,320 So you are your brain and your experiences are these neurons 28 00:02:13,320 --> 00:02:20,840 firing with the trillions of of synapses that somehow lead to 29 00:02:20,840 --> 00:02:22,480 conscious comprehension of the world. 30 00:02:23,480 --> 00:02:28,120 This is something that we have only begun to understand. 31 00:02:28,120 --> 00:02:30,600 We're really just barely at the beginning of understanding of 32 00:02:30,600 --> 00:02:32,000 what is the nature of consciousness. 33 00:02:34,000 --> 00:02:36,120 And I've thought a lot about what, what is consciousness? 34 00:02:36,160 --> 00:02:38,680 What is it? Where does consciousness arise? 35 00:02:43,000 --> 00:02:45,160 Because if you start at the beginning of the universe, 36 00:02:46,000 --> 00:02:49,240 assuming physics is true, the standard model of physics is 37 00:02:49,240 --> 00:02:53,880 true, then you have this Big Bang. 38 00:02:54,560 --> 00:02:58,160 You know the matter condensing into stars, those stars 39 00:02:58,160 --> 00:03:01,520 exploding. A lot of the the atoms that are 40 00:03:01,520 --> 00:03:04,360 in your body right now were ones at the centre of stars. 41 00:03:05,400 --> 00:03:07,240 Those stars exploded, recondensed. 42 00:03:08,880 --> 00:03:11,720 Fast forward 13.8 billion years and here we are. 43 00:03:15,240 --> 00:03:21,320 And somewhere along that very long journey, to us at least, 44 00:03:22,200 --> 00:03:27,720 consciousness erodes or the the molecules started talking to 45 00:03:27,720 --> 00:03:35,360 each other. And it begs the question of what 46 00:03:35,360 --> 00:03:39,880 is consciousness is, is everything conscious? 47 00:03:39,880 --> 00:03:43,520 Maybe it's hard to say. We're along that line, that 48 00:03:43,520 --> 00:03:48,480 there's no sort of discreet point where consciousness didn't 49 00:03:48,480 --> 00:03:50,440 exist and then suddenly does exist. 50 00:03:51,480 --> 00:03:56,400 It seems to be maybe you have a condensation of matter that has 51 00:03:56,400 --> 00:04:01,400 a, a density of like, we don't know what the real, the real 52 00:04:01,400 --> 00:04:02,760 answer is. We don't know what consciousness 53 00:04:02,760 --> 00:04:07,760 is. But with the neural link and the 54 00:04:07,760 --> 00:04:10,400 progress that the company's making, we'll begin to 55 00:04:10,400 --> 00:04:14,880 understand a lot more about consciousness and what does it 56 00:04:14,880 --> 00:04:23,360 mean to, to be along the way, We're, we're going to solve a 57 00:04:23,360 --> 00:04:31,680 lot of, a lot of brain issues where the brains get injured or 58 00:04:31,680 --> 00:04:35,400 damaged in some way or didn't develop in quite the right way. 59 00:04:35,880 --> 00:04:39,800 But there's, you know, there's, there's a lot of brain and spine 60 00:04:40,120 --> 00:04:42,320 injuries that go so along the way. 61 00:04:42,320 --> 00:04:45,240 And I do want to emphasize that this is all going to happen 62 00:04:45,240 --> 00:04:48,240 quite slowly, meaning you'll, you'll see it coming. 63 00:04:48,240 --> 00:04:51,960 Sometimes people think that suddenly there will be vast 64 00:04:51,960 --> 00:04:54,960 numbers of neural links all over the place. 65 00:04:55,120 --> 00:04:57,400 This, this is not going to be sudden. 66 00:04:57,400 --> 00:05:00,800 You'll be able to watch it happen, you know, over the 67 00:05:00,800 --> 00:05:06,560 course of several years and, and we go through exhaustive 68 00:05:06,560 --> 00:05:09,960 regulatory approvals. So this is not something that 69 00:05:09,960 --> 00:05:13,440 we're just doing there by ourselves without government 70 00:05:13,440 --> 00:05:16,000 oversight. We're we work closely with the 71 00:05:16,000 --> 00:05:18,040 regulators every step of the way. 72 00:05:18,240 --> 00:05:22,280 We're very cautious with, with neural links in humans. 73 00:05:23,600 --> 00:05:26,040 That's the reason we're not moving faster than we are is 74 00:05:26,040 --> 00:05:28,240 because we're, we're taking great care with, with each 75 00:05:28,240 --> 00:05:31,200 individual to make sure we, we never miss. 76 00:05:31,880 --> 00:05:34,080 And so far we haven't. And I hope that continues into 77 00:05:34,080 --> 00:05:36,480 the future. Every single one of our implants 78 00:05:36,480 --> 00:05:39,600 in humans is working and working quite well. 79 00:05:39,760 --> 00:05:43,360 And you'll get to hear from some of the people that have received 80 00:05:43,360 --> 00:05:46,760 the implants and cured in their words. 81 00:05:47,400 --> 00:05:49,880 So what we're we're we're creating here with a neural link 82 00:05:49,880 --> 00:05:57,840 device is a generalized input output technology for the brain. 83 00:05:57,840 --> 00:06:03,200 So it's how do you get information into or out of the 84 00:06:03,200 --> 00:06:09,560 brain and do so in a way that does not damage the brain or, 85 00:06:09,920 --> 00:06:11,560 you know, cause any negative side effects. 86 00:06:12,440 --> 00:06:17,720 So it's a very hard problem and generally the the reactions I've 87 00:06:17,720 --> 00:06:21,680 seen to this range from it's impossible to it's already been 88 00:06:21,680 --> 00:06:24,960 done before those those people should meet. 89 00:06:25,080 --> 00:06:31,920 Actually the reality is that there actually have been limited 90 00:06:32,480 --> 00:06:37,960 range of computer interfaces for several decades. 91 00:06:38,400 --> 00:06:43,520 On a very basic basis, just what we're doing with neural link is 92 00:06:43,520 --> 00:06:49,160 dramatically increasing the bandwidth by mayors magnitude. 93 00:06:49,440 --> 00:06:54,880 So you can, you can a human bandwidth output is less than 94 00:06:54,880 --> 00:06:56,720 one bit per second over the course of a day. 95 00:06:57,400 --> 00:06:59,760 So there's 86,400 seconds in a day. 96 00:07:00,240 --> 00:07:04,680 It's very rare for a person to do more than 86,400 bits of 97 00:07:04,680 --> 00:07:07,360 output per day. You'd have to be really talking 98 00:07:07,360 --> 00:07:10,360 a lot or typing all day and you might exceed that. 99 00:07:12,600 --> 00:07:16,360 So what we're talking about here is, is going from maybe one bit 100 00:07:16,360 --> 00:07:20,320 per second to ultimately megabits and then gigabits per 101 00:07:20,320 --> 00:07:27,680 second and the ability to do conceptual consensual telepathy. 102 00:07:29,560 --> 00:07:33,200 Now the the input to the brain is much higher because of, 103 00:07:33,200 --> 00:07:37,560 especially because of vision. Depending upon how you count it, 104 00:07:37,560 --> 00:07:41,640 it, it might be on the order of a megabit or in the megabit 105 00:07:41,640 --> 00:07:45,000 range for input primarily due to sight. 106 00:07:47,680 --> 00:07:51,240 So, but even for input, we, we think that can be dramatically 107 00:07:51,240 --> 00:07:54,440 increased to, to the Gigabit plus level. 108 00:07:57,040 --> 00:08:02,040 And, and, and a lot of the, the thinking that we do is which we 109 00:08:02,040 --> 00:08:06,400 take a concept in our mind and we compress that into a small 110 00:08:06,400 --> 00:08:08,520 number of symbols. So when you're trying to 111 00:08:08,520 --> 00:08:12,120 communicate with somebody else, you're actually trying to model 112 00:08:12,120 --> 00:08:17,800 their mind state and, and, and then take perhaps a quite a 113 00:08:17,800 --> 00:08:21,280 complex idea that you have, maybe even a, a complex image 114 00:08:21,280 --> 00:08:25,280 or, or scene or kind of mental video and try to compress that 115 00:08:25,280 --> 00:08:27,320 into a few words or a few keystrokes. 116 00:08:27,880 --> 00:08:30,040 And it's necessarily going to be very lossy. 117 00:08:30,440 --> 00:08:34,520 Your ability to communicate is very limited by how fast you can 118 00:08:34,520 --> 00:08:38,080 talk and how fast you can type. And what we're talking about is 119 00:08:38,080 --> 00:08:43,080 unlocking that potential to enable you to communicate, like 120 00:08:43,080 --> 00:08:46,400 I said, thousands, perhaps millions of times faster than is 121 00:08:46,520 --> 00:08:49,960 is currently possible. This is an incredibly profound 122 00:08:50,480 --> 00:08:53,920 breakthrough. This would this would be a 123 00:08:53,920 --> 00:08:58,760 fundamental change to what it means to be a human. 124 00:08:59,280 --> 00:09:03,520 So we're, we're starting off with reducing human suffering. 125 00:09:03,520 --> 00:09:07,560 So or, or addressing issues that people have, say if they've been 126 00:09:07,560 --> 00:09:13,200 in an accident or they have some neural disease that's 127 00:09:13,200 --> 00:09:16,560 degenerative so they're losing capability to move their body or 128 00:09:17,400 --> 00:09:19,200 some some kind of injury essentially. 129 00:09:20,760 --> 00:09:24,120 So enabling our first product is called telepathy and that 130 00:09:24,200 --> 00:09:30,320 enables someone who has lost the ability to command their body to 131 00:09:30,320 --> 00:09:33,640 be able to communicate with the computer and move the mouse and 132 00:09:33,640 --> 00:09:37,560 and actually operate a computer with roughly the same dexterity, 133 00:09:37,560 --> 00:09:41,600 ultimately much more dexterity than a than a human with working 134 00:09:41,720 --> 00:09:46,920 hands. Then the our next product is, is 135 00:09:46,920 --> 00:09:50,680 blind sight, which will enable those who have total loss of 136 00:09:50,680 --> 00:09:52,960 vision, including if they've lost their eyes or the optic 137 00:09:52,960 --> 00:09:56,480 nerve or maybe have never seen were blind, even blind from 138 00:09:56,480 --> 00:10:00,760 birth to be able to see again, initially low resolution, but 139 00:10:00,760 --> 00:10:04,120 ultimately very high resolution and, and then in multiple 140 00:10:04,120 --> 00:10:05,960 wavelengths. So you could be like Geordie La 141 00:10:05,960 --> 00:10:09,040 Forge in Star Trek and you can see in radar, you can see an 142 00:10:09,040 --> 00:10:12,080 infrared, ultraviolet, superhuman capabilities, 143 00:10:13,040 --> 00:10:14,680 Severnetic enhancement essentially. 144 00:10:17,360 --> 00:10:22,160 And then along the way, this should help us understand a lot 145 00:10:22,160 --> 00:10:25,080 more about consciousness. What does it mean to be a 146 00:10:25,080 --> 00:10:30,000 conscious creature? We'll understand vastly more 147 00:10:30,000 --> 00:10:32,040 about the nature of consciousness as a result of 148 00:10:32,040 --> 00:10:34,360 this. And then ultimately, I think 149 00:10:34,360 --> 00:10:37,600 this helps mitigate the civilizational risk of 150 00:10:37,640 --> 00:10:45,320 artificial intelligence. We're, we are actually already, 151 00:10:45,320 --> 00:10:47,760 we're already sort of have 3 layers of thinking. 152 00:10:48,880 --> 00:10:51,840 There's the limbic system, which is your kind of your instincts, 153 00:10:52,160 --> 00:10:56,000 the cortic, your cortical system, which is your higher 154 00:10:56,000 --> 00:10:59,120 level planning and thinking. And then the tertiary layer, 155 00:10:59,120 --> 00:11:02,960 which is the computers and machines that you interact with, 156 00:11:03,040 --> 00:11:05,480 you like your phone, your all the applications you use. 157 00:11:07,200 --> 00:11:09,960 So people actually are already a Cyborg. 158 00:11:10,040 --> 00:11:15,480 You can maybe have an intuitive sense for this by how much you 159 00:11:15,480 --> 00:11:17,760 miss your phone if you leave it behind. 160 00:11:19,160 --> 00:11:21,360 Leaving your phone behind is like, it's almost like missing 161 00:11:21,360 --> 00:11:26,000 limb syndrome, but your phone is somewhat of an extension of 162 00:11:26,000 --> 00:11:30,840 yourself, as is your computer. So you, you already have this 163 00:11:30,840 --> 00:11:34,640 digital tertiary layer, but the bandwidth between your cortex 164 00:11:34,680 --> 00:11:39,760 and your digital tertiary layer is limited by speech and by and 165 00:11:39,760 --> 00:11:41,960 by how fast you can move your fingers and how fast you can 166 00:11:41,960 --> 00:11:48,800 consume information visually. So, so, but I think it's 167 00:11:48,800 --> 00:11:52,600 actually very important for us to address that input output 168 00:11:52,600 --> 00:11:56,960 bandwidth constraint in order for the collective will of 169 00:11:56,960 --> 00:11:59,840 humanity to match the will of artificial intelligence. 170 00:12:03,400 --> 00:12:09,560 That's my intuition at least. So let's see and, and what what 171 00:12:09,560 --> 00:12:18,440 this presentation is mostly about is attracting smart humans 172 00:12:18,440 --> 00:12:21,200 to come and work with us on this problem. 173 00:12:21,920 --> 00:12:25,640 So this is not a presentation to raise money or anything like 174 00:12:25,640 --> 00:12:28,960 that. We're actually, you know, very 175 00:12:28,960 --> 00:12:30,920 well funded. We have a lot of great 176 00:12:30,920 --> 00:12:33,920 investors. Some of the smartest people in 177 00:12:33,920 --> 00:12:38,280 the world are invested in neural link, but we we need smart 178 00:12:38,280 --> 00:12:40,360 humans to come here and help solve this problem. 179 00:12:41,080 --> 00:12:42,720 So with that, let's let's proceed. 180 00:12:55,920 --> 00:13:01,000 Hey everyone, my name is DJ, my Co founder and president of 181 00:13:01,000 --> 00:13:04,560 Neuro Link. And as Elon mentioned, well, 182 00:13:04,560 --> 00:13:07,520 actually we're standing in the middle of our robot space. 183 00:13:07,920 --> 00:13:11,080 We have a stage set up, but you know, this is actually where 184 00:13:12,000 --> 00:13:14,680 some of the next generation most advanced surgical robots are 185 00:13:14,680 --> 00:13:16,400 being built. So welcome to our space. 186 00:13:24,590 --> 00:13:26,950 It's important to highlight that this technology is not being 187 00:13:26,950 --> 00:13:30,670 built in the dark. This is not a secret lab where 188 00:13:30,670 --> 00:13:32,190 we're not sharing any of the progress. 189 00:13:32,430 --> 00:13:35,760 In fact, we're actually sharing, you know the progress very 190 00:13:35,760 --> 00:13:38,960 openly and as well as also telling you exactly what we're 191 00:13:38,960 --> 00:13:41,800 going to be doing. And we're hoping to progress on 192 00:13:41,800 --> 00:13:46,760 that as as diligently and as safely and as carefully as 193 00:13:46,760 --> 00:13:50,880 possible. So to start off, 2 years ago 194 00:13:50,880 --> 00:13:53,880 when we did our previous fundraising round, we outlined 195 00:13:53,880 --> 00:13:56,160 this path and timeline to 1st Human. 196 00:13:57,200 --> 00:14:00,960 And we currently have a clinical trials in the US for a product 197 00:14:00,960 --> 00:14:04,880 that we call Telepathy, which allows users to control phone or 198 00:14:04,880 --> 00:14:06,800 computer purely with their thoughts. 199 00:14:06,800 --> 00:14:09,440 And you're going to see how we do this and what the impact that 200 00:14:09,440 --> 00:14:13,720 this has had. And not only have we launched 201 00:14:13,720 --> 00:14:18,320 this clinical trial, but as of today, we have not just one, but 202 00:14:18,320 --> 00:14:20,800 7 participants and we have an approval. 203 00:14:27,960 --> 00:14:32,520 And we also have an approval to launch this trial in Canada, UK 204 00:14:32,560 --> 00:14:43,110 and the UAE. So I guess before we dive into 205 00:14:43,230 --> 00:14:47,070 what this technology is and what we built, but I wanted to 206 00:14:47,070 --> 00:14:50,950 quickly share a video with you guys of when our first five 207 00:14:50,950 --> 00:14:53,750 participants met each other for the first time, so. 208 00:14:53,790 --> 00:14:56,310 Here you go. All right, we have everyone 209 00:14:56,310 --> 00:14:58,200 together. What's up guys? 210 00:14:58,400 --> 00:15:01,200 Thanks everybody for joining. Definitely want to introduce all 211 00:15:01,200 --> 00:15:04,280 of you. Yeah, I'm Nolan, AKA P1. 212 00:15:04,560 --> 00:15:07,880 My name's Alex. I am the second participant in 213 00:15:07,880 --> 00:15:12,400 the Neural link study. I am Brad Smith, the ALS Cypler. 214 00:15:12,520 --> 00:15:15,440 P. Three, my name is Mike. 215 00:15:16,400 --> 00:15:27,440 G4 ALS like Francis. Yeah, I'm RJ, I'm P5, and I 216 00:15:27,480 --> 00:15:30,360 just, yes, I'm trying to do this one to the team here. 217 00:15:30,360 --> 00:15:33,680 So yeah, appreciate it. And all them Trailblazer, you 218 00:15:33,680 --> 00:15:35,080 know, somebody's got to get first, man. 219 00:15:35,080 --> 00:15:36,280 That was you. Appreciate that. 220 00:15:36,600 --> 00:15:39,000 What's been your favorite thing you've been able to do with the 221 00:15:39,000 --> 00:15:41,760 neural link so far? I've just had a good time being 222 00:15:41,760 --> 00:15:45,320 able to use it as I travel flying and draw a little 223 00:15:45,320 --> 00:15:47,960 mustache on a cat. Had a lot of fun doing that. 224 00:15:48,000 --> 00:15:50,080 I mean, I've just had a good time playing around with it. 225 00:15:50,400 --> 00:15:53,360 Oh, you know what? I do know what my favorite BCI 226 00:15:53,360 --> 00:15:56,480 feature is? Probably not a feature, but I 227 00:15:56,480 --> 00:16:00,240 just I love web grid more than I love anything in my life 228 00:16:00,320 --> 00:16:02,960 probably. I think I could play that game 229 00:16:03,080 --> 00:16:07,200 non-stop forever. Has to be Fusion 360. 230 00:16:07,720 --> 00:16:11,440 Being able to design parts, design the hat logo with the 231 00:16:11,440 --> 00:16:14,480 BCI. That's what's up. 232 00:16:15,000 --> 00:16:16,160 Pretty sweet. That's sweet. 233 00:16:16,520 --> 00:16:23,560 Yeah, yeah, I have a little Arduino that takes input from my 234 00:16:23,560 --> 00:16:29,880 quad stick, converts it into APPM signal to go to ARC truck. 235 00:16:30,920 --> 00:16:34,920 Cool Little Rock crawler. Well, with the BCII. 236 00:16:36,640 --> 00:16:41,320 Wrote code, can you? To drive the plane with the quad 237 00:16:41,400 --> 00:16:46,080 stick, that's awesome. The best thing I like about an 238 00:16:46,080 --> 00:16:50,440 airline is being able to continue. 239 00:16:51,360 --> 00:16:58,120 To provide for my family and continue working. 240 00:16:58,760 --> 00:17:02,120 I think my favorite thing is probably been able to turn on my 241 00:17:02,120 --> 00:17:05,079 TV. Yeah, like the first time in 2 242 00:17:05,079 --> 00:17:07,400 1/2 years I was able to do that. So it's pretty sweet. 243 00:17:07,400 --> 00:17:10,560 But I like shooting the hobbies. That's that's kind of nice. 244 00:17:11,079 --> 00:17:13,160 Excited to see what BC is got going on. 245 00:17:14,040 --> 00:17:15,760 I got a question. What's your shirt say? 246 00:17:16,640 --> 00:17:18,440 I said. I'd do a thing called whatever I 247 00:17:18,440 --> 00:17:32,040 want. Now, one of the major figure of 248 00:17:32,040 --> 00:17:35,680 merits that we have is to keep track of monthly hours of 249 00:17:35,680 --> 00:17:38,960 independent PCI use. Effectively, are they using the 250 00:17:38,960 --> 00:17:42,600 PCI and not at the clinic but at their home. 251 00:17:42,800 --> 00:17:46,440 And what we have noticed and this is a plot of all of the 252 00:17:46,440 --> 00:17:48,800 different participants, first five participants and their 253 00:17:48,800 --> 00:17:52,880 usage per month over the course of the last year and a half. 254 00:17:53,240 --> 00:17:57,160 And we're averaging around 50 hours a week of usage and in 255 00:17:57,160 --> 00:18:00,080 some cases peak usage of more than 100 hours a week, which is 256 00:18:00,440 --> 00:18:01,800 pretty much every waking moments. 257 00:18:09,800 --> 00:18:12,720 So I think it's been incredible to see all of our participants 258 00:18:13,040 --> 00:18:15,640 demonstrating greater independence through their use 259 00:18:15,640 --> 00:18:20,000 of BCI. Not only that, we've also 260 00:18:20,000 --> 00:18:23,400 accelerated our implantation cadence as we've amassed 261 00:18:23,400 --> 00:18:26,520 evidence of both clinical safety as well as value to our 262 00:18:26,520 --> 00:18:29,560 participants. So to date, we have 4 spinal 263 00:18:29,560 --> 00:18:33,120 cord injury participants as well as three ALS participants with 264 00:18:33,120 --> 00:18:35,920 the last two surgeries happening within one week of each other. 265 00:18:39,800 --> 00:18:43,760 And we're just beginning. This is just tip of the iceberg. 266 00:18:43,920 --> 00:18:47,640 Our end goal is to really build a whole brain interface. 267 00:18:47,720 --> 00:18:50,120 And what do we mean by whole brain interface? 268 00:18:50,360 --> 00:18:54,640 We mean being able to listen to neurons everywhere, be able to 269 00:18:54,640 --> 00:18:58,280 write information to neurons anywhere, be able to have that 270 00:18:58,280 --> 00:19:01,800 fast data wireless transfer to enable that high bandwidth 271 00:19:01,800 --> 00:19:06,080 connection from our biological brain to the external machines 272 00:19:06,600 --> 00:19:11,040 and be able to do all of this with fully automated surgery as 273 00:19:11,040 --> 00:19:13,520 well as enable 24 hours of usage. 274 00:19:16,440 --> 00:19:19,360 And towards that goal, we're really working on three major 275 00:19:19,360 --> 00:19:21,920 product types. Elon mentioned earlier that our 276 00:19:21,920 --> 00:19:25,600 goal is to build a generalized input output platform and 277 00:19:25,600 --> 00:19:29,560 technology to the brain. So to afford the output portion 278 00:19:29,560 --> 00:19:33,680 of it, which is extremely slow through our meat sticks as as 279 00:19:33,680 --> 00:19:39,720 Elon calls them, neat hands that are holding the mics. 280 00:19:41,080 --> 00:19:44,160 We're starting out with helping people with movement disorders 281 00:19:44,640 --> 00:19:47,440 either through where they lost the mind body connection either 282 00:19:47,440 --> 00:19:50,280 through a spinal cord injury, ALS or a stroke. 283 00:19:50,520 --> 00:19:53,880 Be able to regain some of that digital as well as physical 284 00:19:53,880 --> 00:19:55,960 independence through a product that we're building called 285 00:19:55,960 --> 00:19:59,240 Telepathy. And this is our opportunities to 286 00:19:59,240 --> 00:20:01,800 build a high channel read and output device. 287 00:20:03,240 --> 00:20:05,560 On the input side of things, there's. 288 00:20:06,440 --> 00:20:08,880 Opportunities for us to help people that have lost the 289 00:20:08,880 --> 00:20:13,440 ability to to see be able to regain that site again through a 290 00:20:13,440 --> 00:20:15,120 product that we're calling blind sight. 291 00:20:15,480 --> 00:20:18,520 And this is our opportunity to build high channel right 292 00:20:18,760 --> 00:20:22,360 capabilities. And last but not least, be able 293 00:20:22,360 --> 00:20:25,760 to also help people that are suffering from neurological 294 00:20:27,040 --> 00:20:30,360 debilitating dysregulation or psychiatric conditions or 295 00:20:30,840 --> 00:20:35,120 neuropathic pain. By inserting our electrodes in 296 00:20:35,760 --> 00:20:39,680 reaching any brain regions to be able to insert them not just on 297 00:20:39,680 --> 00:20:43,280 the cortical layer but into the sulk eyes as well as deeper 298 00:20:43,280 --> 00:20:46,880 parts of the brain. The so-called limbic system to 299 00:20:47,200 --> 00:20:49,840 really enable better opportunities to just regain 300 00:20:49,840 --> 00:20:54,920 some of that independence. Our North Star metrics is 1. 301 00:20:54,960 --> 00:20:58,200 Increasing the number of neurons that we can interface with. 302 00:20:58,720 --> 00:21:02,360 And 2nd, to expand to many diverse area any parts of the 303 00:21:02,360 --> 00:21:06,040 brain, starting with microfabrication or lithography 304 00:21:06,040 --> 00:21:09,400 to change the way in which we can actually increase the number 305 00:21:09,400 --> 00:21:11,400 of neurons that we can see from a single channel. 306 00:21:12,160 --> 00:21:17,360 And also doing mixed signal chip design to actually increase the 307 00:21:17,360 --> 00:21:22,640 physical channel counts to increase more neurons that we 308 00:21:22,640 --> 00:21:27,200 can interface to, to, to sort of allow more information from the 309 00:21:27,200 --> 00:21:30,440 brain to the outside world. And then you know, everything we 310 00:21:30,440 --> 00:21:34,320 built from day one of the company has always been read and 311 00:21:34,320 --> 00:21:37,040 write capable. And with telepathy, our first 312 00:21:37,040 --> 00:21:40,680 product, the focus has been on the read capabilities or the 313 00:21:40,680 --> 00:21:43,200 output. And we want to hone in on our 314 00:21:43,200 --> 00:21:47,280 write capability and also show that through accessing deeper 315 00:21:47,280 --> 00:21:50,960 regions within the, the, the visual cortex that we can 316 00:21:50,960 --> 00:21:52,320 actually achieve functional vision. 317 00:21:52,320 --> 00:21:55,400 Let's. Go. 318 00:22:04,360 --> 00:22:07,320 So now just to step you through what the product evolution is 319 00:22:07,320 --> 00:22:08,960 going to look like in the next three years. 320 00:22:09,520 --> 00:22:14,080 Today what we have is 1000 electrodes in the motor cortex, 321 00:22:14,080 --> 00:22:16,320 the part of the small part of the brain that you see in this 322 00:22:16,320 --> 00:22:20,400 animation called the hand knob area that allows participants 323 00:22:20,400 --> 00:22:22,760 control computer cursors as well as gaming consoles. 324 00:22:24,040 --> 00:22:28,960 Next quarter, we're planning to implant in the speech cortex to 325 00:22:28,960 --> 00:22:33,920 directly decode attentive words from brain signals to speech. 326 00:22:33,920 --> 00:22:41,400 And in 2026, not only are we going to triple the number of 327 00:22:42,160 --> 00:22:46,640 electrodes from 1000 to 3000 for more capabilities, we're 328 00:22:46,640 --> 00:22:50,040 planning to have our first blind side participant to enable 329 00:22:50,040 --> 00:23:01,800 navigation. And in 2027, we're going to 330 00:23:01,800 --> 00:23:04,960 continue increasing channel counts, probably another triple, 331 00:23:04,960 --> 00:23:10,320 so 10,000 channels and also enable for the first time 332 00:23:10,320 --> 00:23:14,320 multiple implants, so not just one in motor cortex, speech 333 00:23:14,320 --> 00:23:16,920 cortex or visual cortex, but all of the above. 334 00:23:20,280 --> 00:23:25,200 And finally, in 2028, our goal is to get to more than 25,000 335 00:23:25,200 --> 00:23:29,360 channels per implant, have multiple of these, have ability 336 00:23:29,360 --> 00:23:33,200 to access any part of the brain for psychiatric conditions, pain 337 00:23:33,920 --> 00:23:37,240 dysregulation and also start to demonstrate what it would be 338 00:23:37,240 --> 00:23:38,920 like to actually integrate with AI. 339 00:23:47,900 --> 00:23:50,700 And all this is to say that we're really building towards 340 00:23:51,040 --> 00:23:54,440 set of fundamental foundational technology that would allow us 341 00:23:54,440 --> 00:23:57,320 to have hundreds of thousands, if not millions of channels with 342 00:23:57,320 --> 00:23:59,920 multiple implants for whole grain interfaces that could 343 00:23:59,920 --> 00:24:02,640 actually solve not just these debilitating neurological 344 00:24:02,640 --> 00:24:06,320 conditions, but be able to go beyond the the limits of our 345 00:24:06,320 --> 00:24:08,360 biology. And this vertical integration. 346 00:24:08,360 --> 00:24:11,600 And the talent and team that we have at Neural Link has been and 347 00:24:11,600 --> 00:24:14,560 will continue to be the key recipe for rapid progress that 348 00:24:14,560 --> 00:24:17,520 we will be making. Just to recap real quick, Neural 349 00:24:17,520 --> 00:24:20,320 Link is implanted with precision surgical robot. 350 00:24:21,120 --> 00:24:25,440 It's physically invisible and one week later users are able to 351 00:24:25,440 --> 00:24:28,840 see their thoughts, transform into actions and to share more 352 00:24:28,840 --> 00:24:31,000 about what that experience is like. 353 00:24:31,720 --> 00:24:33,720 I'd like to welcome Sahej to the station. 354 00:24:47,720 --> 00:24:50,200 What's up guys? My name is Sahej, I'm from the 355 00:24:50,200 --> 00:24:53,080 Brain Computer Interface team here at Neurolink, and I'm going 356 00:24:53,080 --> 00:24:54,600 to be talking about two things today. 357 00:24:55,160 --> 00:24:58,160 The first thing is, what exactly is the neural link device 358 00:24:58,160 --> 00:25:01,520 capable of doing right now? And the second one is how does 359 00:25:01,520 --> 00:25:04,360 that actually impact the day-to-day lives of our users? 360 00:25:05,640 --> 00:25:10,440 Very Simply put, what the neural link device does right now is it 361 00:25:10,440 --> 00:25:14,240 allows you to control devices simply just by thinking. 362 00:25:14,640 --> 00:25:18,080 Now to put that a bit more concretely, I'm about to play a 363 00:25:18,080 --> 00:25:20,880 video of our first user. His name is Noland, if you 364 00:25:20,880 --> 00:25:25,320 remember from DJ section. And what Noland is doing is he's 365 00:25:25,320 --> 00:25:28,120 looking at a normal off the shelf MacBook Pro. 366 00:25:28,560 --> 00:25:31,840 And with his Neurolink devices, you're going to see he's going 367 00:25:31,840 --> 00:25:34,760 to be able to control the cursor simply with his mind. 368 00:25:34,760 --> 00:25:36,840 No eye tracking, no other sensors. 369 00:25:37,240 --> 00:25:40,880 And what's special about this particular moment is this is the 370 00:25:40,880 --> 00:25:43,920 first time someone is using a Neurolink device to fully 371 00:25:43,920 --> 00:25:50,200 control their cursor. This is not your ordinary brain 372 00:25:50,200 --> 00:25:53,200 controlled cursor. This is actually a 373 00:25:53,200 --> 00:25:56,920 record-breaking control literally on day one, beating 374 00:25:56,920 --> 00:25:58,920 decades of brain computer research. 375 00:25:59,600 --> 00:26:03,800 And I'm about to show you the clip on day one, Nolan breaking 376 00:26:03,800 --> 00:26:20,640 the BCI world record. Oh well done man, He's a new 377 00:26:20,640 --> 00:26:30,000 world record holder. I. 378 00:26:31,600 --> 00:26:34,240 Thought it was higher. I thought I would have to get to 379 00:26:34,240 --> 00:26:37,120 5 or something. Oh my gosh, that's crazy. 380 00:26:39,240 --> 00:26:47,920 It's pretty cool, yeah. Another really fun thing you 381 00:26:47,920 --> 00:26:50,800 could do with the newer Link device, outside of controlling a 382 00:26:50,800 --> 00:26:53,720 computer cursor, is you can actually plug it in through USB 383 00:26:53,720 --> 00:26:55,040 through a lot of different devices. 384 00:26:55,600 --> 00:26:58,440 And here we actually have Nolan playing Mario Kart. 385 00:26:58,880 --> 00:27:02,800 Now, what's special about this particular clip is Nolan is not 386 00:27:02,800 --> 00:27:05,120 the only Cyborg playing Mario Kart in this clip. 387 00:27:05,560 --> 00:27:08,360 We actually have a whole community of users, as mentioned 388 00:27:08,360 --> 00:27:10,760 earlier. And this is literally five of 389 00:27:10,760 --> 00:27:14,080 our first users of Neuralink playing Mario Kart together over 390 00:27:14,080 --> 00:27:26,440 call now. Yeah, Mario Kart is, it's cool. 391 00:27:26,440 --> 00:27:28,440 You know, you're using one joystick and then you're 392 00:27:28,440 --> 00:27:30,360 clicking like a couple buttons to throw items. 393 00:27:31,160 --> 00:27:34,280 What would be even cooler is what if you could control 2 394 00:27:34,280 --> 00:27:37,160 joysticks at once simultaneously with your mind? 395 00:27:37,880 --> 00:27:40,240 What I'm about to show you, and I think this is for the first 396 00:27:40,240 --> 00:27:43,240 time someone playing a first person shooter game with a brain 397 00:27:43,240 --> 00:27:47,680 computer interface. This is Alex and RJ playing Call 398 00:27:47,680 --> 00:27:51,040 of Duty controlling 1 joystick to move and then the other 399 00:27:51,040 --> 00:27:53,440 joystick to like think, point your gun and then shooting 400 00:27:53,440 --> 00:27:57,320 people as a button. Here's the larger Here's Alex 401 00:27:57,880 --> 00:27:59,240 shooting another person. 'S bikes. 402 00:28:00,480 --> 00:28:05,880 Oh dear God, I know I do and I. Want him to freaking. 403 00:28:05,880 --> 00:28:10,400 Choose long. When I do, I know he's shot me 404 00:28:10,400 --> 00:28:14,680 in the face. Now that we have a bit of a 405 00:28:14,680 --> 00:28:18,360 sense of what the BCI can do, a very important question to 406 00:28:18,360 --> 00:28:22,000 answer is how does this impact the day-to-day lives of the 407 00:28:22,000 --> 00:28:26,720 people that use it every day? So I'm about to show you a clip 408 00:28:26,800 --> 00:28:29,760 going back to Nolan for a second where he talks. 409 00:28:29,760 --> 00:28:33,240 We simply just asked him randomly during a day how he 410 00:28:33,240 --> 00:28:36,520 enjoys using the BCI couple months ago and this is his 411 00:28:36,520 --> 00:28:41,240 candid reaction. I work basically all day from 412 00:28:41,520 --> 00:28:45,040 when I wake up. I'm trying to wake up at like 413 00:28:45,440 --> 00:28:49,560 6:00 or 7:00 AM and I'll do work until session. 414 00:28:49,560 --> 00:28:57,080 I'll do session and then I'll work until, you know, 11:12 PM 415 00:28:57,280 --> 00:29:05,800 or 12:00 AM I'm, I'm doing like I'm learning my languages, I'm 416 00:29:07,120 --> 00:29:09,680 learning my math. I'm like relearning all of my 417 00:29:09,680 --> 00:29:14,760 math. I am writing, I am doing the 418 00:29:14,800 --> 00:29:19,880 class that I signed up for and I just I wanted to point out that 419 00:29:20,240 --> 00:29:25,080 like this is not something I would be able to do out like 420 00:29:25,080 --> 00:29:28,480 without the neuro link. Next I want to talk a bit about 421 00:29:28,480 --> 00:29:30,120 Brad. You guys may already know him as 422 00:29:30,120 --> 00:29:34,360 the ALS Cyborg and Brad also has ALS and what separates him from 423 00:29:34,360 --> 00:29:37,360 other users is he's actually non verbal so he can't speak. 424 00:29:38,480 --> 00:29:42,040 Why this is pretty relevant is he relies at least before the 425 00:29:42,040 --> 00:29:45,320 neuro link on an eye gaze machine to communicate and a lot 426 00:29:45,320 --> 00:29:47,520 of eye gaze machines you can't use outdoors. 427 00:29:47,920 --> 00:29:49,480 You really need like a dark room. 428 00:29:49,960 --> 00:29:52,840 So what this means is for the last six years since Brad's been 429 00:29:52,840 --> 00:29:57,160 diagnosed with ALS, he's really unable to leave his house. 430 00:29:57,320 --> 00:29:59,120 Now with the Neuralink device, we're going to show you a clip 431 00:29:59,560 --> 00:30:03,000 of him with his kids at the park, shot by Ashley Vance and 432 00:30:03,000 --> 00:30:07,440 the team Guys ready. No, absolutely. 433 00:30:08,840 --> 00:30:11,600 I am absolutely doing more with Neuralink than I was doing with 434 00:30:11,600 --> 00:30:17,120 eye gaze. I've been a Batman for a long 435 00:30:17,120 --> 00:30:21,320 time, but I go outside now. Going outside has been a huge 436 00:30:21,360 --> 00:30:24,920 blessing for me and I can control the computer with 437 00:30:24,920 --> 00:30:26,600 telepathy. Dad's watching. 438 00:30:27,240 --> 00:30:28,680 OK, He's. Watching on the camera, did he? 439 00:30:30,280 --> 00:30:32,760 Lose. One of the arms the last user I 440 00:30:32,760 --> 00:30:35,440 want to talk about is Alex. You've seen some clips of him 441 00:30:35,440 --> 00:30:37,960 earlier. What's special about Alex to me 442 00:30:38,320 --> 00:30:41,560 is he's a fellow left-handed guy who writes in cursive all the 443 00:30:41,560 --> 00:30:43,600 time. And what he mentioned is since a 444 00:30:43,600 --> 00:30:47,520 spinal cord injury from like, 3-4 years ago, he's been able 445 00:30:47,520 --> 00:30:50,240 unable to just, like, draw or write. 446 00:30:51,240 --> 00:30:53,400 And he always brags about how good his handwriting was. 447 00:30:53,400 --> 00:30:55,040 So we actually got to put in a test. 448 00:30:55,040 --> 00:30:57,560 We gave him a robotic arm. And I think this is the first 449 00:30:57,560 --> 00:31:00,840 time you tried using the robotic arm to write anything. 450 00:31:01,000 --> 00:31:04,200 And this is a spotted version of writing at the convoy trial and 451 00:31:04,200 --> 00:31:15,840 drawing something now. Yeah, controlling A robotic arm 452 00:31:15,840 --> 00:31:18,200 is cool, but this one has a clamp. 453 00:31:18,480 --> 00:31:21,720 And what would be cooler is if you could decode the actual 454 00:31:21,720 --> 00:31:24,520 fingers, the actual wrist, all the muscles of the hand in real 455 00:31:24,520 --> 00:31:28,480 time. Just in the past couple weeks we 456 00:31:28,480 --> 00:31:31,880 were able to do that with Alex and you're about to see him and 457 00:31:31,880 --> 00:31:37,640 his uncle playing a game. Rock, paper, scissors shoot. 458 00:31:39,240 --> 00:31:45,200 Rock, paper, scissors shoot. Rock, paper, scissors shoot 459 00:31:46,520 --> 00:31:47,840 rock. Paper. 460 00:31:48,000 --> 00:31:49,040 Scissors. Shoot. 461 00:32:09,640 --> 00:32:11,120 Cool. Controlling. 462 00:32:11,120 --> 00:32:12,840 Yeah, that's pretty dope. I don't know. 463 00:32:18,040 --> 00:32:22,760 And controlling A robotic hand on screen is obviously not super 464 00:32:22,760 --> 00:32:26,640 helpful for most people. Fortunately, we have connections 465 00:32:26,640 --> 00:32:30,720 with Tesla, who have the optimist hand, and we're 466 00:32:30,720 --> 00:32:36,080 actually actively working on giving Alex an optimist hand so 467 00:32:36,080 --> 00:32:38,240 that you could actually control it in his real life. 468 00:32:39,000 --> 00:32:43,000 And here's actual replay of the end of that video using Alex's 469 00:32:43,000 --> 00:32:46,000 neural signals on an Optus hand. Sean, if you want to play that, 470 00:32:53,120 --> 00:32:59,600 Yeah. Actually, let me maybe add a few 471 00:32:59,600 --> 00:33:05,880 things to that, which is so as we advance the neural link 472 00:33:05,880 --> 00:33:11,600 devices, you should be able to actually have full body control 473 00:33:11,600 --> 00:33:14,000 and sensors from an optimist robot. 474 00:33:14,000 --> 00:33:17,600 So you could basically inhabit an optimist robot. 475 00:33:17,600 --> 00:33:20,760 So not just the hand the whole, the whole thing. 476 00:33:21,800 --> 00:33:26,640 So you could like basically mentally remote into an optimist 477 00:33:27,280 --> 00:33:32,160 robot and and be kind of cool. The future's going to be weird, 478 00:33:33,840 --> 00:33:41,040 but but but pretty cool. And then now I was. 479 00:33:41,640 --> 00:33:43,640 Another thing that could be done also is like for people that 480 00:33:43,640 --> 00:33:47,680 have say, lost a limb, lost an arm or a leg or something like 481 00:33:47,680 --> 00:33:52,440 that, then we think in the future we'll be able to attach 482 00:33:53,000 --> 00:33:57,640 an optimist's arm or legs. And so you kind of like, I 483 00:33:57,640 --> 00:34:02,040 remember that scene from Star Wars where Luke Skywalker gets 484 00:34:02,040 --> 00:34:05,160 his hand, you know, chopped over the lightsaber and he gets kind 485 00:34:05,160 --> 00:34:08,239 of a robot hand. And I think that's the kind of 486 00:34:08,239 --> 00:34:11,920 thing that we'll be able to do in the future working with the 487 00:34:11,920 --> 00:34:16,400 newer Lincoln Tesla. So, so it goes far beyond just 488 00:34:16,400 --> 00:34:19,199 operating a robot hand, but replacing limbs and, and having 489 00:34:19,199 --> 00:34:20,679 kind of a whole body robot experience. 490 00:34:21,280 --> 00:34:25,840 And then I think another thing that will be possible, I think 491 00:34:25,840 --> 00:34:30,080 it's very likely in the future is to be able to bridge the, the 492 00:34:30,159 --> 00:34:33,120 where the damaged neurons are. So you can take the signal from 493 00:34:33,120 --> 00:34:38,000 the brain and, and transmit that signal past where the neurons 494 00:34:38,000 --> 00:34:40,960 are damaged or strained to the rest of the body. 495 00:34:40,960 --> 00:34:44,440 So you could reanimate the body so that if you have a neural 496 00:34:44,440 --> 00:34:48,120 link implant in the brain and then one in the spinal cord, 497 00:34:48,520 --> 00:34:52,800 then you you can actually bridge the signals and you could walk 498 00:34:52,800 --> 00:34:55,560 again and have full body functionality. 499 00:34:55,560 --> 00:34:57,160 Obviously that's what people would prefer. 500 00:34:57,720 --> 00:35:00,320 To be clear, we realized that that would be the preferred 501 00:35:00,320 --> 00:35:06,440 outcome and and so that even if you have a broken neck or you 502 00:35:06,440 --> 00:35:09,600 could. So we believe I'm, I'm actually 503 00:35:09,800 --> 00:35:11,920 at this point I'd say fairly confident that at some point in 504 00:35:11,920 --> 00:35:14,920 the future we'll be able to restore full body functionality. 505 00:35:23,040 --> 00:35:24,680 Yes. So hello, hello everyone. 506 00:35:24,680 --> 00:35:27,520 My name is Nir and I'm leading the BCR application group. 507 00:35:28,200 --> 00:35:32,600 And I think the videos just that Sahel just shared with you, I 508 00:35:32,600 --> 00:35:36,280 probably watch them maybe thousands of time, but still I 509 00:35:36,280 --> 00:35:38,120 get the goosebump every time I watch them. 510 00:35:38,760 --> 00:35:41,120 And I think this is one of the cool perks here at New Orleans 511 00:35:41,120 --> 00:35:44,040 when you get a job is that you might get goosebump every week 512 00:35:44,080 --> 00:35:45,880 or maybe every few days in good weeks. 513 00:35:46,440 --> 00:35:53,080 And, and this is really fun as an engineer, it's really cool 514 00:35:53,080 --> 00:35:54,600 because you can build a new feature. 515 00:35:54,600 --> 00:35:57,360 You can build a new machine learning model and use software 516 00:35:57,360 --> 00:36:00,560 feature and test it on the same day with the participant and get 517 00:36:00,560 --> 00:36:04,320 feedback. And you already saw with our 518 00:36:04,320 --> 00:36:08,800 first device Telepathy that we can address a very diverse needs 519 00:36:08,920 --> 00:36:13,480 of the different users that we have for moving a cursor to 520 00:36:13,480 --> 00:36:16,760 playing games, to move a robotic arm with multiple fingers. 521 00:36:17,920 --> 00:36:20,320 And we could not have done it without the neural link device. 522 00:36:20,600 --> 00:36:23,560 The neural link device gives us something that no other device 523 00:36:23,560 --> 00:36:27,840 can give us, which is in a single neuron recording from 524 00:36:27,840 --> 00:36:29,480 thousands of channels simultaneously. 525 00:36:29,840 --> 00:36:32,600 The telepathy products is basically recording the neural 526 00:36:32,600 --> 00:36:36,720 activity from the small area in the motor cortex that involve an 527 00:36:36,720 --> 00:36:38,640 execution of hand and arm movements. 528 00:36:39,320 --> 00:36:42,520 But if we go only about two or three inches below, there's 529 00:36:42,520 --> 00:36:45,400 another brain area that's involved in execution of speech. 530 00:36:46,360 --> 00:36:49,680 And with the same device, with the same machine learning model 531 00:36:49,720 --> 00:36:52,920 architecture, the same software pipeline, the same surgical 532 00:36:52,920 --> 00:36:55,560 robot, we can have a new application and we can do it 533 00:36:55,560 --> 00:36:57,920 very quickly. It's really interesting that if 534 00:36:57,920 --> 00:37:03,040 we can decode someone intention to speak silently and non vocal 535 00:37:03,040 --> 00:37:06,800 communication, we can use that to revolutionize the way we 536 00:37:06,800 --> 00:37:10,080 interact with computers, with technology and with information. 537 00:37:10,800 --> 00:37:13,920 Instead of typing with your finger or like moving the mouse 538 00:37:14,360 --> 00:37:17,120 or talking to your phone, you'll be able to interact with 539 00:37:17,120 --> 00:37:18,720 computer with the speed of thoughts. 540 00:37:19,840 --> 00:37:22,760 It will make this interaction much more, much faster and much 541 00:37:22,760 --> 00:37:25,360 more intuitive. The computers will understand 542 00:37:25,360 --> 00:37:28,080 what you want to do. And we can also expand that to 543 00:37:28,080 --> 00:37:31,040 AI. We can now build an interface 544 00:37:31,560 --> 00:37:35,680 with AI that you will be able to achieve information, will be 545 00:37:35,680 --> 00:37:40,120 able to store our thoughts anywhere, anytime, privately and 546 00:37:40,120 --> 00:37:42,800 silently. Again, because we build a 547 00:37:42,800 --> 00:37:46,400 fundamental technology, a platform, and we do everything 548 00:37:46,400 --> 00:37:49,200 in house. We own the entire stack from 549 00:37:49,200 --> 00:37:51,480 neurons to pixels on the user's computer. 550 00:37:52,280 --> 00:37:56,120 Now I'll pass pass it to RUSE to talk about UI for Visa. 551 00:38:03,050 --> 00:38:06,490 Thank you, NIA. Each spike that our implant 552 00:38:06,490 --> 00:38:09,890 detects goes on a fairly remarkable journey to ultimately 553 00:38:09,890 --> 00:38:12,250 form a pixel on a participant's display. 554 00:38:12,250 --> 00:38:18,210 And that experience starts with, of course, unboxing, the very 555 00:38:18,210 --> 00:38:21,650 first time that a participant pairs to and meets their 556 00:38:21,650 --> 00:38:26,520 implant, this invisible part of their body, and sees their own 557 00:38:26,520 --> 00:38:28,960 spikes materialize across the display. 558 00:38:30,960 --> 00:38:34,280 From there, they'll go into body mapping and actually imagine 559 00:38:34,280 --> 00:38:38,040 moving their arm again and get a feel for what feels natural to 560 00:38:38,040 --> 00:38:41,760 them and what doesn't. And they'll take that into 561 00:38:41,760 --> 00:38:47,640 calibration, using one of those motions to actually move a 562 00:38:47,640 --> 00:38:52,360 cursor again, iteratively refining their control as they 563 00:38:52,360 --> 00:38:57,520 go throughout this process, until finally they're teleported 564 00:38:57,760 --> 00:39:02,640 back to their desktop and can experience the magic of neural 565 00:39:02,640 --> 00:39:07,640 control for the very first time. And our control interfaces is 566 00:39:07,640 --> 00:39:12,560 where the OS integration that we do really shines, letting us 567 00:39:12,560 --> 00:39:17,360 adapt both control and feedback for every interaction. 568 00:39:17,720 --> 00:39:21,400 So for familiar interactions like scrolling, we can surface 569 00:39:21,400 --> 00:39:25,480 an indicator over the scrollable parts of the display, add a 570 00:39:25,480 --> 00:39:27,680 touch of gravity to automatically pop a 571 00:39:27,680 --> 00:39:32,840 participant's cursor onto that indicator as they approach, show 572 00:39:32,840 --> 00:39:36,040 the actual velocities that we decode inside of it, and add a 573 00:39:36,040 --> 00:39:39,040 bit of momentum to those velocities to carry them forward 574 00:39:39,040 --> 00:39:43,400 as they glide across the page. There are also unique 575 00:39:43,400 --> 00:39:45,760 interactions that we need to solve for in this space. 576 00:39:46,280 --> 00:39:49,160 For example, when a participant is watching a movie or just 577 00:39:49,160 --> 00:39:53,360 talking to somebody next to them, the brain is very active 578 00:39:53,360 --> 00:39:56,400 still, and that activity can actually induce motion in the 579 00:39:56,400 --> 00:39:58,560 cursor, distracting them from that moment. 580 00:39:59,240 --> 00:40:01,800 So when a participant wants to just get their cursor out of the 581 00:40:01,800 --> 00:40:05,240 way, they can push it into the edge of the display to park it 582 00:40:05,240 --> 00:40:07,840 there. And of course we add gravity to 583 00:40:07,840 --> 00:40:12,400 sort of hold it still, but they can push it out with either just 584 00:40:12,400 --> 00:40:14,680 a firm push or in this case, a gesture. 585 00:40:16,480 --> 00:40:19,280 And of course, it's when it goes without saying that all of these 586 00:40:19,280 --> 00:40:21,840 control interfaces are designed hand in hand with our 587 00:40:21,840 --> 00:40:24,360 participants. So huge shout out to both Noland 588 00:40:24,360 --> 00:40:26,240 and Brad for helping us design these two. 589 00:40:27,240 --> 00:40:30,040 And those control interfaces, of course, extend typing. 590 00:40:30,560 --> 00:40:33,040 We have a great software keyboard that does everything 591 00:40:33,040 --> 00:40:35,920 you'd expect it to, popping up when a participant clicks on a 592 00:40:35,920 --> 00:40:39,080 text field, giving them feedback about the click on the surface 593 00:40:39,080 --> 00:40:41,880 of the key, and supporting both dictation and swipe. 594 00:40:49,440 --> 00:40:52,400 Hi everyone, I'm Harrison and ML Engineer here at Neural Link. 595 00:40:52,880 --> 00:40:55,720 And I must say, being an ML engineer at Neural Link is a bit 596 00:40:55,720 --> 00:40:57,280 like being a kid in a candy store. 597 00:40:59,640 --> 00:41:02,640 When you think of the inputs to most ML systems out there, you 598 00:41:02,640 --> 00:41:06,200 might think of pixels of tokens or of a user's Netflix watch 599 00:41:06,200 --> 00:41:09,320 history. The input to our systems is a 600 00:41:09,320 --> 00:41:13,000 little different. It is pure raw brain power. 601 00:41:13,240 --> 00:41:15,720 And when we think about the ML systems we can build here at 602 00:41:15,720 --> 00:41:18,400 Neuralink, really we're limited by our imagination and our 603 00:41:18,400 --> 00:41:21,320 creativity. There's no reason our ML systems 604 00:41:21,320 --> 00:41:24,960 can't do anything that the human brain can do, such as 605 00:41:25,080 --> 00:41:28,360 controlling a phone, typing, or even gaming. 606 00:41:29,640 --> 00:41:32,640 Right here to my left is actual footage of Alex, one of our 607 00:41:32,640 --> 00:41:36,400 participants, playing a first person shooter against RJ, 608 00:41:36,400 --> 00:41:39,800 another one of our participants. Now, for those unfamiliar with 609 00:41:39,800 --> 00:41:42,320 first person shooters, this is not a trivial feat. 610 00:41:42,440 --> 00:41:46,640 It requires 2 fully independent joysticks or 4 continuous 611 00:41:46,640 --> 00:41:50,400 degrees of control, as well as multiple reliable buttons. 612 00:41:51,600 --> 00:41:55,240 Now contrary to popular belief, the neural link does not simply 613 00:41:55,320 --> 00:41:59,520 read people's minds, it's simply reading neuronal activations 614 00:41:59,520 --> 00:42:02,800 corresponding to motor intent. So one of the fun challenges 615 00:42:02,800 --> 00:42:05,720 with this project was figuring out which motions were going to 616 00:42:05,720 --> 00:42:09,120 be mapped to the joystick. We started with the typical left 617 00:42:09,120 --> 00:42:11,560 thumb and right thumb, but quickly found that the dominant 618 00:42:11,560 --> 00:42:13,360 hand overshadowed the non dominant hand. 619 00:42:14,240 --> 00:42:16,880 My personal favorite is we had one of our participants imagine 620 00:42:16,880 --> 00:42:19,480 walking for the left joystick and aiming for the right 621 00:42:19,480 --> 00:42:21,560 joystick. So in game, they were simply 622 00:42:21,560 --> 00:42:25,560 doing naturalistic motions like you might do in virtual reality 623 00:42:25,560 --> 00:42:28,240 in Ready Player 1, and that was really cool to watch. 624 00:42:28,840 --> 00:42:32,280 What we ended up on was the thumb for the left joystick and 625 00:42:32,280 --> 00:42:34,040 the wrist for the right joystick. 626 00:42:34,320 --> 00:42:37,240 And I challenge the audience to try to replicate their emotions. 627 00:42:37,440 --> 00:42:39,520 I'm really in all of them being able to pull this off. 628 00:42:40,160 --> 00:42:42,360 I want to talk a bit about the progress to our cursor 629 00:42:42,360 --> 00:42:45,920 calibration experience. To my left, here you can see RJ 630 00:42:45,920 --> 00:42:49,400 completing his first ever cursor calibration with a redesigned 631 00:42:49,400 --> 00:42:51,680 Open the Flow, where he first gather information about his 632 00:42:51,680 --> 00:42:54,440 intent and how to map the neural activity to the first time he 633 00:42:54,440 --> 00:42:57,440 controls a cursor, to the final product where he has smooth and 634 00:42:57,440 --> 00:43:02,160 fluid control of his computer. And most remarkably, this 635 00:43:02,160 --> 00:43:06,800 experience took only 15 minutes from start to finish, 15 minutes 636 00:43:06,800 --> 00:43:15,880 from not 15 minutes from no control to fluid computer use. 637 00:43:17,400 --> 00:43:20,360 Contrast that to a year and a half ago with P1, where that was 638 00:43:20,360 --> 00:43:23,000 multiple hours to get to the same level of control and 639 00:43:23,000 --> 00:43:25,440 several engineers standing around a table pulling their 640 00:43:25,440 --> 00:43:27,840 hair out. There was virtually no need for 641 00:43:27,840 --> 00:43:29,680 neural link engineers to even be at this session. 642 00:43:30,000 --> 00:43:32,160 This was basically an out-of-the-box experience for 643 00:43:32,160 --> 00:43:38,280 our participants. And even more remarkably, we're 644 00:43:38,280 --> 00:43:41,560 continuing to smash day one records, with RJ being able to 645 00:43:41,560 --> 00:43:45,560 achieve 7 BPS on his very first day with a neural link. 646 00:43:49,840 --> 00:43:52,640 Now such an effective and efficient calibration process is 647 00:43:52,640 --> 00:43:56,040 only made possible by high fidelity estimations of a user 648 00:43:56,040 --> 00:44:00,480 intention or labels. And to briefly illustrate just 649 00:44:00,480 --> 00:44:03,240 how challenging of a problem that is, this is an animation of 650 00:44:03,240 --> 00:44:06,720 myself trying to draw circles on my desktop with a mouse. 651 00:44:07,360 --> 00:44:10,920 Now the task was simple, draw uniform circles at a constant 652 00:44:10,920 --> 00:44:13,720 speed, repeatedly. And as you can see by that 653 00:44:13,720 --> 00:44:15,840 animation, I am horrible at that. 654 00:44:16,160 --> 00:44:19,200 Even though my intent was pretty obvious, unambiguous, the 655 00:44:19,200 --> 00:44:23,120 execution was really poor. There is a ton of variation in 656 00:44:23,120 --> 00:44:27,840 both speed and the shape itself. To visualize this a little 657 00:44:27,840 --> 00:44:31,040 differently, each row here is one of those circles unwound in 658 00:44:31,040 --> 00:44:34,040 time with synchronized starts, and you can just see how much 659 00:44:34,040 --> 00:44:36,600 variation there is in the timing of each circle as well as I'm 660 00:44:36,600 --> 00:44:38,280 doing at any given point in time. 661 00:44:40,840 --> 00:44:43,320 Orthogonal to the labeling problem is neural non 662 00:44:43,320 --> 00:44:46,120 stationarity, or the tendency of neural signals to drift over 663 00:44:46,120 --> 00:44:48,840 time. And I think that's honestly a 664 00:44:48,840 --> 00:44:51,400 beautiful thing, right? If you if your neural signals 665 00:44:51,400 --> 00:44:54,520 didn't drift, you couldn't grow. When you wake up the next day, 666 00:44:54,800 --> 00:44:56,560 you're not the same person you were the day before. 667 00:44:56,600 --> 00:44:59,960 You've learned, you've grown, you've changed, and so too must 668 00:44:59,960 --> 00:45:02,440 your neural data change. This animation. 669 00:45:02,440 --> 00:45:05,160 Here is a simple illustration of the learned representation by 670 00:45:05,160 --> 00:45:08,040 the decoder and how it drifts the further away we get from the 671 00:45:08,040 --> 00:45:10,440 day it was trained on. This is one of the key 672 00:45:10,440 --> 00:45:13,160 challenges we need to solve here at Neurolink to unlock fluid and 673 00:45:13,160 --> 00:45:14,840 product level experience for our users. 674 00:45:23,730 --> 00:45:29,770 Hey everyone, Hey everyone, my name is Joey. 675 00:45:31,690 --> 00:45:35,090 Blind Sight is our project to build a visual prosthesis to 676 00:45:35,090 --> 00:45:39,450 help the blind see again. Users would wear a pair of 677 00:45:39,450 --> 00:45:45,400 glasses with an embedded camera and receive an implant in their 678 00:45:45,400 --> 00:45:49,080 visual cortex. Scenes from the environment are 679 00:45:49,080 --> 00:45:52,400 recorded by the camera and processed in the patterns of 680 00:45:52,400 --> 00:45:57,840 stimulation delivered to the brain, causing visual perception 681 00:45:58,680 --> 00:46:02,600 and restoring functionality. Now blind sight will be enabled 682 00:46:02,600 --> 00:46:04,840 by placing our implant into visual cortex. 683 00:46:04,960 --> 00:46:07,640 This is a new brain area for us, and this brings new 684 00:46:07,640 --> 00:46:12,600 opportunities and challenges. So the surface of the brain for 685 00:46:12,600 --> 00:46:16,360 visual cortex represents just a few degrees of angle in the 686 00:46:16,360 --> 00:46:20,320 center of the visual field. Larger fields of view are 687 00:46:20,320 --> 00:46:23,320 represented deep within the cortical folds of the calcarine 688 00:46:23,320 --> 00:46:27,120 fissure. Our threads are able to access 689 00:46:27,120 --> 00:46:30,600 these deeper structures, providing the possibility of 690 00:46:30,600 --> 00:46:33,360 restoring vision over a functional, useful visual field. 691 00:46:34,320 --> 00:46:37,360 So the N1 implant has had experimental stimulation 692 00:46:37,360 --> 00:46:41,960 capabilities for quite some time, but our new S2 chip is 693 00:46:41,960 --> 00:46:44,160 designed from the ground up for stimulation. 694 00:46:44,960 --> 00:46:49,640 It provides over 1600 channels of electrical stimulation, high 695 00:46:49,640 --> 00:46:53,240 dynamic range recording capabilities and a wide range of 696 00:46:53,240 --> 00:46:55,480 micro stimulation currents and voltages. 697 00:46:56,800 --> 00:46:59,240 We can achieve these capabilities because we are 698 00:46:59,240 --> 00:47:02,480 vertically integrated and we designed this custom ASIC in 699 00:47:02,480 --> 00:47:05,400 house. Similarly, we design and 700 00:47:05,400 --> 00:47:08,640 fabricate our electrode threads in house, and here you can see 701 00:47:08,640 --> 00:47:12,760 one of our standard threads designed for recording in an 702 00:47:12,760 --> 00:47:15,800 electron micrograph for blind sight. 703 00:47:15,800 --> 00:47:18,640 Our requirements are a little different, and our vertical 704 00:47:18,640 --> 00:47:21,560 integration allows us to rapidly iterate on the design and 705 00:47:21,560 --> 00:47:24,200 manufacturing of these threads for this new purpose. 706 00:47:25,480 --> 00:47:28,400 So here I'm using Red Arrows to highlight the electrode 707 00:47:28,400 --> 00:47:32,040 contacts, which are optimized for stimulation. 708 00:47:32,480 --> 00:47:35,000 And as you can see, they're a little bit larger, which results 709 00:47:35,000 --> 00:47:38,400 in a lower electrical impedance for safe and effective charge 710 00:47:38,400 --> 00:47:40,360 delivery, which is important for blind sight. 711 00:47:43,080 --> 00:47:45,480 Now, how can we calibrate our implant for blind sight? 712 00:47:46,200 --> 00:47:50,920 So here's one way we stimulate on the array, picking say three 713 00:47:50,920 --> 00:47:54,200 different channels. The user perceives something, 714 00:47:54,200 --> 00:47:57,840 say 3 spots of light somewhere in their visual field and points 715 00:47:57,840 --> 00:48:00,480 at them. We track their arm and eye 716 00:48:00,480 --> 00:48:03,120 movements and repeat this process for each of the channels 717 00:48:03,120 --> 00:48:08,120 on the array. And here's what a simulated 718 00:48:08,120 --> 00:48:10,840 example of a blind sight vision could look like after 719 00:48:10,840 --> 00:48:24,440 calibration. Now I showed you how for blind 720 00:48:24,440 --> 00:48:26,920 sight, we need to insert threads deeper into the brain than we 721 00:48:26,920 --> 00:48:30,640 have previously, and doing this requires state-of-the-art 722 00:48:30,640 --> 00:48:34,000 medical imaging. So we worked with Siemens to get 723 00:48:34,000 --> 00:48:35,400 some of the best scanners on Earth. 724 00:48:36,040 --> 00:48:40,160 We built out our imaging core from scratch in the past year. 725 00:48:40,560 --> 00:48:41,840 Actually, it was faster than that. 726 00:48:41,840 --> 00:48:43,880 It was about four months from dirt to done. 727 00:48:45,200 --> 00:48:47,680 Since bringing the scanners online, we've scanned over 50 728 00:48:47,680 --> 00:48:50,480 internal participants, building out a database of human 729 00:48:50,480 --> 00:48:52,080 structural and functional anatomy. 730 00:48:52,840 --> 00:48:55,760 What can we do with the imaging information from these scanners 731 00:48:56,320 --> 00:48:59,080 so medical imaging can be used for surgical placement? 732 00:48:59,360 --> 00:49:02,600 It lets us parcel it out brain regions by their function and we 733 00:49:02,600 --> 00:49:05,560 use our imaging capabilities to refine the placement for 734 00:49:05,560 --> 00:49:08,120 telepathy. It also gives us the capability 735 00:49:08,120 --> 00:49:11,000 to target new brain regions for future products such as blind 736 00:49:11,000 --> 00:49:14,160 sight or speech prosthesis. And we're working towards more 737 00:49:14,160 --> 00:49:17,040 capabilities. So one click, automated planning 738 00:49:17,040 --> 00:49:19,720 of surgery from functional images to robot insertion 739 00:49:19,720 --> 00:49:21,880 targets. Here you can see a screen 740 00:49:21,880 --> 00:49:24,520 capture from one of our in house tooling to do end to end 741 00:49:24,520 --> 00:49:27,080 surgical planning. You can see a region of motor 742 00:49:27,080 --> 00:49:31,120 cortex known as hand knob and the thread trajectory plans that 743 00:49:31,120 --> 00:49:32,880 will be sent directly to the robot. 744 00:49:34,080 --> 00:49:37,440 This is a really incredible degree of automation that's only 745 00:49:37,440 --> 00:49:39,680 possible because we're controlling the system from one 746 00:49:39,680 --> 00:49:46,680 end to the other. My name is John and I lead the 747 00:49:46,880 --> 00:49:50,560 robot mechanical team. This is our current R1 robot. 748 00:49:50,560 --> 00:49:53,080 It was used to implant the first 7 participants. 749 00:49:53,680 --> 00:49:56,360 This robot works really well, but it has a few flaws. 750 00:49:56,360 --> 00:49:58,800 One of which is the cycle time is rather slow. 751 00:49:59,160 --> 00:50:03,280 So to insert each thread it takes in a best case scenario 17 752 00:50:03,280 --> 00:50:05,800 seconds. And many cases external 753 00:50:05,800 --> 00:50:09,560 disturbances cause us to have to retry to reinsert, grasp that 754 00:50:09,560 --> 00:50:13,480 thread and then reinsert it. To scale our number of neuron or 755 00:50:13,480 --> 00:50:16,360 neurons access through higher channel count, increased numbers 756 00:50:16,360 --> 00:50:19,000 of threads, we need to have a much faster cycle time. 757 00:50:19,480 --> 00:50:23,800 So let me introduce our next generation robot, which is right 758 00:50:23,800 --> 00:50:34,520 here. Through rethinking the way that 759 00:50:34,520 --> 00:50:36,720 we hold the implant in front of the robot, by holding it 760 00:50:36,720 --> 00:50:39,560 directly in front on on the robot head, we will achieve an 761 00:50:39,560 --> 00:50:43,720 11 times cycle time improvement. So each thread takes 1 1/2 762 00:50:43,720 --> 00:50:46,640 seconds. We also scale up a lot of 763 00:50:46,800 --> 00:50:49,720 surgery. Workflow process improvements 764 00:50:49,720 --> 00:50:52,600 through deleting the separate operator station and implant 765 00:50:52,600 --> 00:50:55,200 stand. Now the outside of the robot 766 00:50:55,560 --> 00:50:58,160 looks pretty similar between the two, but it's what's inside that 767 00:50:58,160 --> 00:51:01,000 really counts. Each system has been redesigned 768 00:51:01,000 --> 00:51:04,240 from the the ground up with the focus on reliability, 769 00:51:04,240 --> 00:51:07,080 manufacturability, serviceability and using a lot 770 00:51:07,080 --> 00:51:09,000 of our vertical integration techniques. 771 00:51:09,000 --> 00:51:11,600 It's enabled us to have a lot more control of the system end 772 00:51:11,600 --> 00:51:15,680 to end. Now that fast cycle time doesn't 773 00:51:15,680 --> 00:51:18,120 mean much if it's not compatible with a significant portion of 774 00:51:18,120 --> 00:51:22,040 the human population. Prior to each surgery we scan 775 00:51:22,280 --> 00:51:24,560 our participants anatomy and ensure that they will be 776 00:51:24,560 --> 00:51:26,960 compatible with the robot and vice versa. 777 00:51:27,440 --> 00:51:30,280 Unfortunately, the robot isn't compatible with everyone so we 778 00:51:30,280 --> 00:51:33,240 had to extend the reach of the needle in the next generation 779 00:51:33,240 --> 00:51:36,080 robot and now we're compatible with more than 99% of the human 780 00:51:36,080 --> 00:51:38,800 population. We've also increased the depth 781 00:51:38,800 --> 00:51:40,160 of the needle can insert threads. 782 00:51:40,440 --> 00:51:43,000 Now we can reach more than 50mm from the surface of the brain, 783 00:51:43,200 --> 00:51:45,680 accessing and enabling new indications. 784 00:51:46,240 --> 00:51:49,600 We have to produce a ton of custom sterile components for 785 00:51:49,600 --> 00:51:51,880 each surgery. We actually supply more than 20 786 00:51:51,880 --> 00:51:54,040 of these parts. Many of these parts are made 787 00:51:54,040 --> 00:51:56,680 through traditional CNC manufacturing capabilities, 788 00:51:56,680 --> 00:51:59,680 which we do just on the other side of this wall actually, and 789 00:51:59,680 --> 00:52:02,400 some custom developed processes like this femtosecond laser 790 00:52:02,400 --> 00:52:04,640 milling used to manufacture the tip of the needle. 791 00:52:05,280 --> 00:52:09,360 Now these processes take quite a bit of time, effort and cost. 792 00:52:09,480 --> 00:52:12,800 So let's take a look at how we're going to reduce costs and 793 00:52:12,800 --> 00:52:16,560 time for one of the components. So the current needle cartridge 794 00:52:16,880 --> 00:52:20,280 has a total cycle time of about 24 hours and the machine 795 00:52:20,280 --> 00:52:24,200 components cost about $350.00. The final assembly is performed 796 00:52:24,200 --> 00:52:26,040 by a set of like highly skilled technicians. 797 00:52:26,280 --> 00:52:30,080 They have to glue 150 Micron diameter Canyon onto this wire 798 00:52:30,080 --> 00:52:32,760 EDM machined stainless steel base plate. 799 00:52:32,760 --> 00:52:36,080 They have to Electro Polish a 40 Micron wire into a sharp taper 800 00:52:36,480 --> 00:52:39,200 and then they have to thread that 40 Micron wire into the a 801 00:52:39,200 --> 00:52:43,800 60 Micron hole in the Canyon. This is done manually and then 802 00:52:43,800 --> 00:52:46,080 they finally have to laser Weld all the components together. 803 00:52:48,440 --> 00:52:51,640 Next generation needle cartridge takes only 30 minutes of cycle 804 00:52:51,640 --> 00:52:55,240 time and $15 in component. We were able to delete the wire 805 00:52:55,240 --> 00:52:58,880 EDM machined base plate and the Kenya gluing step by switching 806 00:52:58,880 --> 00:53:01,880 to an insert molded component. So we get a box of these base 807 00:53:01,880 --> 00:53:04,800 plates with the Kenya was already installed for like 1000 808 00:53:04,800 --> 00:53:07,120 of them for like a couple 5-10 dollars a piece. 809 00:53:07,640 --> 00:53:10,280 We also deleted the Electro polishing step with the revised 810 00:53:10,280 --> 00:53:13,640 needle tip geometry, which is also compatible with inserting 811 00:53:13,640 --> 00:53:17,040 the threads through the dura. We have a few revised 812 00:53:17,040 --> 00:53:19,720 manufacturing techniques to delete the manual threading 813 00:53:19,720 --> 00:53:22,600 through a basically a funnel. Rather simple, but it has been a 814 00:53:22,600 --> 00:53:24,720 big impact. And then we're able to delete 815 00:53:24,720 --> 00:53:26,200 the laser alding through using crimping. 816 00:53:32,440 --> 00:53:38,640 Hi. Hi. 817 00:53:38,640 --> 00:53:40,600 I'm Julian. I'm one of the leads on the 818 00:53:40,600 --> 00:53:44,240 implant team. So the way humans communicate 819 00:53:44,240 --> 00:53:47,480 today, if they want to output information, is by using their 820 00:53:47,480 --> 00:53:49,600 hands and their voice, as I'm doing right now. 821 00:53:50,200 --> 00:53:53,160 And if you want to receive information, you use your ears 822 00:53:53,160 --> 00:53:55,480 and your eyes. And of course, that's how you're 823 00:53:55,480 --> 00:54:00,720 receiving this very talk. But we've built this implant and 824 00:54:00,720 --> 00:54:04,200 this implant is very special because it is the first time 825 00:54:04,480 --> 00:54:07,560 that we're able to add a completely new mode of data 826 00:54:07,560 --> 00:54:09,880 transfer into and out of the brain. 827 00:54:11,080 --> 00:54:14,000 If you look at this device in a nutshell, it's really just 828 00:54:14,000 --> 00:54:17,120 sampling voltages in the brain and sending them over radio. 829 00:54:17,720 --> 00:54:21,600 But if you zoom out and look at the system from end to end, what 830 00:54:21,600 --> 00:54:25,680 you actually see is that we're connecting your brain or 831 00:54:25,760 --> 00:54:29,840 biological neural net to a machine learning model or a 832 00:54:29,840 --> 00:54:32,000 silicon neural net on the right hand side. 833 00:54:32,520 --> 00:54:36,920 And I actually think this is really elegant because the 834 00:54:36,920 --> 00:54:39,520 machine learning model on the right hand side is in fact 835 00:54:39,520 --> 00:54:42,440 inspired by neurons on the left hand side. 836 00:54:42,760 --> 00:54:46,080 And so in some sense, we're really extending the fundamental 837 00:54:46,080 --> 00:54:48,920 substrate of the brain. For the first time, we're able 838 00:54:48,920 --> 00:54:51,720 to do this in a mass market product that's a very, very 839 00:54:51,720 --> 00:55:02,600 special piece of hardware. So these are some of the first 840 00:55:02,600 --> 00:55:07,280 implants that we ever built. There are electrodes that were 841 00:55:07,280 --> 00:55:09,880 made with our in house lithography tools. 842 00:55:10,200 --> 00:55:12,680 We have custom ASICS that we also designed in house. 843 00:55:13,000 --> 00:55:16,520 And this was really a platform for us to develop the technology 844 00:55:16,720 --> 00:55:20,160 that allows us to sense micro level volts in the brain across 845 00:55:20,160 --> 00:55:21,960 thousands of channels simultaneously. 846 00:55:22,760 --> 00:55:25,480 We learnt a lot from this. But as you'll notice in the 847 00:55:25,480 --> 00:55:29,800 right to images, there are USBC connectors on these devices. 848 00:55:30,080 --> 00:55:32,760 These were not really the most implantable implants. 849 00:55:34,520 --> 00:55:38,640 This next set of images are the wireless implants, and it was a 850 00:55:38,640 --> 00:55:42,200 complete evolution that we went through to add the battery, the 851 00:55:42,200 --> 00:55:46,200 antenna, the radio, and to make it actually fully implantable. 852 00:55:46,320 --> 00:55:48,960 Once it's implanted, it's completely invisible. 853 00:55:49,120 --> 00:55:52,440 It's very compact, it's modular, and it's a general platform that 854 00:55:52,440 --> 00:55:54,000 you can use in many places in the brain. 855 00:55:55,480 --> 00:55:59,920 Going from that top row to the bottom row is very challenging. 856 00:56:00,600 --> 00:56:03,200 The implant you see on the bottom right here is in fact the 857 00:56:03,200 --> 00:56:06,800 device that we have working in seven participants today, and 858 00:56:06,800 --> 00:56:10,120 it's augmenting their brain every day and restoring their 859 00:56:10,120 --> 00:56:12,080 autonomy. But getting to that point 860 00:56:12,440 --> 00:56:15,280 involved a huge number of formidable engineering 861 00:56:15,280 --> 00:56:17,400 challenges. We first had to make a hermetic 862 00:56:17,400 --> 00:56:21,280 enclosure, passing 1000 separate conductors through the enclosure 863 00:56:21,280 --> 00:56:23,840 of the device. We had to figure out how to make 864 00:56:23,840 --> 00:56:26,720 charging seamless and work with very tight thermal constraints 865 00:56:26,720 --> 00:56:29,920 in a very, very small area. And then we also had to scale up 866 00:56:30,120 --> 00:56:32,680 our testing infrastructure so that we could support large 867 00:56:32,680 --> 00:56:35,920 scale manufacturing and very safe devices and have confidence 868 00:56:35,960 --> 00:56:39,080 in our iteration cycle. So what's next? 869 00:56:39,880 --> 00:56:43,280 We're going to be increasing our manufacturing so that we don't 870 00:56:43,280 --> 00:56:46,160 just produce, you know, a certain like a small number of 871 00:56:46,160 --> 00:56:48,520 implants per year, but thousands and then eventually millions of 872 00:56:48,520 --> 00:56:50,800 implants per year. We're also going to be 873 00:56:50,800 --> 00:56:55,000 increasing channel count. More channels means more neurons 874 00:56:55,000 --> 00:56:57,160 are sensed, which means more capabilities. 875 00:56:57,760 --> 00:57:01,400 In some sense, We often think a lot about the the Moore's Law of 876 00:57:01,400 --> 00:57:02,920 neurons that we're interacting with. 877 00:57:03,160 --> 00:57:06,920 And in the same way that Moore's Law propelled forward many 878 00:57:06,920 --> 00:57:10,080 subsequent revolutions in computing, we think that sensing 879 00:57:10,080 --> 00:57:13,280 more and more neurons will also completely redefine how we 880 00:57:13,280 --> 00:57:15,960 interact with computers and reality at large. 881 00:57:17,800 --> 00:57:19,360 I want to leave you with one final thought. 882 00:57:20,360 --> 00:57:24,760 When I was a child I used a 56 KB modem to access the Internet. 883 00:57:26,760 --> 00:57:29,240 If you remember what it's like, you would go to a website. 884 00:57:29,560 --> 00:57:32,000 You're lucky. You're lucky. 885 00:57:32,000 --> 00:57:35,120 Bastard. Yeah, When I was a child we had 886 00:57:35,120 --> 00:57:37,000 acoustic couplers. Oh yeah, OK. 887 00:57:37,520 --> 00:57:39,640 So just beep. Just beep at each other. 888 00:57:39,800 --> 00:57:43,240 Yeah, the the first modem was the acoustic coupler, incredible 889 00:57:43,240 --> 00:57:47,040 device honestly. But then if you, I guess if 890 00:57:47,040 --> 00:57:55,760 you're my age, you started with 56 K bit modem and you, you 891 00:57:55,760 --> 00:57:58,920 would go to a website and and like there would be an image and 892 00:57:58,920 --> 00:58:02,600 it would, it would scroll like slowly it was loading pixel by 893 00:58:02,600 --> 00:58:05,480 pixel on the screen. So that that's what it's like to 894 00:58:05,480 --> 00:58:08,800 be bandwidth limited. Now imagine using the current 895 00:58:08,800 --> 00:58:12,160 Internet with that same modem. It it's like it's inconceivable. 896 00:58:12,160 --> 00:58:17,680 It would be impossible to do so. What broadband Internet did to 897 00:58:17,680 --> 00:58:21,680 the 56 KB modem is what this hardware is going to do to the 898 00:58:21,680 --> 00:58:23,920 brain. We are trying to drastically 899 00:58:23,920 --> 00:58:27,360 expand the amount of bandwidth that you have access to, to have 900 00:58:27,360 --> 00:58:30,280 a much richer experience and superhuman capabilities. 901 00:58:37,820 --> 00:58:42,500 So I guess just to kind of close out and to recap today Neurolink 902 00:58:42,500 --> 00:58:47,160 is working reliably and has already changed the lives of 903 00:58:47,160 --> 00:58:49,520 seven participants and making a real impact. 904 00:58:51,640 --> 00:58:55,040 And our next milestone is to go to market and enable scaling of 905 00:58:55,040 --> 00:58:59,600 this technology to thousands of people and as well as expand 906 00:58:59,600 --> 00:59:03,000 functionalities beyond just the movement to enable robotic, 907 00:59:03,080 --> 00:59:07,640 sophisticated robotic arm control, speech, vision, give 908 00:59:07,640 --> 00:59:10,680 sight back and even getting to the speed of thought. 909 00:59:11,360 --> 00:59:15,960 I, I hope you got a good sort of sample of our technology stack 910 00:59:15,960 --> 00:59:19,800 and the challenges that we have and I'd like to hand over the 911 00:59:19,800 --> 00:59:22,400 mic to Elon for any closing remarks. 912 00:59:24,400 --> 00:59:30,040 Well, we're trying to give you a sense of the the depth of talent 913 00:59:30,320 --> 00:59:32,240 at Neurolink. There's a lot of really smart 914 00:59:32,240 --> 00:59:35,360 people working on a lot of important problems. 915 00:59:36,160 --> 00:59:40,360 This is one of the most difficult things to to actually 916 00:59:40,360 --> 00:59:43,400 succeed in creating and have it work and work at scale and be 917 00:59:43,400 --> 00:59:46,880 reliable and available for millions of people at an 918 00:59:46,880 --> 00:59:53,280 affordable price. So super hard problem and would 919 00:59:53,280 --> 00:59:56,320 like to have you come join and help us solve it. 920 00:59:56,320 --> 01:00:15,680 Thank you. Hey, thank you so much for 921 01:00:15,680 --> 01:00:17,880 listening today. I really do appreciate your 922 01:00:17,880 --> 01:00:19,440 support. If you could take a second and 923 01:00:19,440 --> 01:00:22,360 hit this subscribe or the follow button on whatever podcast 924 01:00:22,360 --> 01:00:25,400 platform that you're listening on right now, I greatly 925 01:00:25,400 --> 01:00:27,040 appreciate it. It helps out the show 926 01:00:27,040 --> 01:00:29,680 tremendously and you'll never miss an episode. 927 01:00:29,880 --> 01:00:33,240 And each episode is about 10 minutes or less to get you 928 01:00:33,240 --> 01:00:36,000 caught up quickly. And please, if you want to 929 01:00:36,000 --> 01:00:41,480 support the show even more, go to patreon.com/stagezero. 930 01:00:41,960 --> 01:00:44,160 And please take care of yourselves and each other, and 931 01:00:44,160 --> 01:00:45,080 I'll see you tomorrow.