WEBVTT

1
00:00:00.120 --> 00:00:05.120
How the brain works is very limited. You know, we've got fMRI,

2
00:00:06.040 --> 00:00:09.759
which is that that's kind of like
putting us, you know, a stethoscope

3
00:00:10.160 --> 00:00:13.480
on the outside of a factory wall
and been putting it like all over the

4
00:00:13.519 --> 00:00:16.239
factory wall, and you can sort
of hear the sounds, but you don't

5
00:00:16.239 --> 00:00:20.640
know what the machines are doing.
Really it's hot. You can infer a

6
00:00:20.640 --> 00:00:24.559
few things, but it's very broad
breaststroke. In order to really know what's

7
00:00:24.559 --> 00:00:26.920
going on in the brain, you
really need you have to have high precision

8
00:00:27.000 --> 00:00:31.120
sensors, and then you want to
have stimulus and response, Like if you

9
00:00:31.280 --> 00:00:34.159
trigger a neuron, what, how
do you feel? What do you see?

10
00:00:34.679 --> 00:00:38.039
How does the change of perception of
the world you're speaking to physically?

11
00:00:38.200 --> 00:00:40.880
Just getting close to the brain,
being able to measure signals from the brain

12
00:00:41.359 --> 00:00:46.799
will give us sort of open the
door inside the factory. Yes, exactly.

13
00:00:47.039 --> 00:00:52.600
Being able to have high precision sensors
that tell you what individual neurons are

14
00:00:52.600 --> 00:00:57.439
doing, and then being able to
trigger the neuron and see what the response

15
00:00:57.560 --> 00:01:00.840
is in the brain, so you
can see the consequences of of of a

16
00:01:02.799 --> 00:01:04.879
if you fire this neuron, what
happens? How do you feel? What

17
00:01:04.920 --> 00:01:10.239
has changed. It's it'll be really
profound to have this in people, because

18
00:01:10.280 --> 00:01:15.640
people can articulate their change, like
if there's a change in mood or if

19
00:01:15.680 --> 00:01:19.359
they you know, if they can
tell you if they can see better or

20
00:01:19.519 --> 00:01:26.040
hear better, or be able to
form sentences better or worse, or you

21
00:01:26.079 --> 00:01:30.319
know, their memories are jogged or
that kind of thing. So on the

22
00:01:30.400 --> 00:01:34.480
on the human side, there's this
incredible general malleability plasticity of the human brain.

23
00:01:34.599 --> 00:01:38.640
The human brain adapts, adjusts,
and so on. So that's not

24
00:01:38.719 --> 00:01:42.200
that plastic you're totally frank, so
there's a firm structure, but there nevertheless

25
00:01:42.560 --> 00:01:47.319
there's some plasticity. And the open
question is, so if I could ask

26
00:01:47.959 --> 00:01:52.239
a broad question, is how much
that plasticity can be utilized? Sort of

27
00:01:52.359 --> 00:01:56.799
on the human side, there's some
plasticity in human brain, and on the

28
00:01:56.920 --> 00:02:02.840
machine side, we have networks machine
learning are ficial intelligence. It's able to

29
00:02:02.920 --> 00:02:07.800
adjust and figure out signals. So
there's a mysterious language that we don't perfectly

30
00:02:07.919 --> 00:02:13.759
understand that's within the human brain,
and then we're trying to understand that language

31
00:02:13.759 --> 00:02:17.479
to communicate both directions. So the
brain is adjusting a little bit. We

32
00:02:17.520 --> 00:02:22.719
don't know how much, and the
machine is adjusting. Where do you see

33
00:02:23.039 --> 00:02:25.960
as they try to sort of reach
together, almost like with an alien species,

34
00:02:27.080 --> 00:02:30.520
try to find a protocol, communication
protocol that works. Where do you

35
00:02:30.520 --> 00:02:36.719
see the biggest the biggest benefit arriving
from on the machine side or the human

36
00:02:36.759 --> 00:02:38.759
side? Do you see both of
them working together? I should think the

37
00:02:38.840 --> 00:02:46.039
machine side is far more malleable than
the biological side by huge amount. So

38
00:02:46.240 --> 00:02:50.639
it'll be the machine that adapts to
the brain. That's the only thing that's

39
00:02:50.639 --> 00:02:54.719
possible to brain can't adapt that well
to the machine. You can't have neurons

40
00:02:55.120 --> 00:03:00.199
start to regard an electrode as another
neuron, not just that this like the

41
00:03:00.240 --> 00:03:05.960
pulse and so something else is pulsing. So there is that, there is

42
00:03:06.000 --> 00:03:09.840
that that elsticity in the interface,
which we believe is something that can happen.

43
00:03:10.520 --> 00:03:14.800
But the vast majority of the malleability
will have to be on the machine

44
00:03:14.840 --> 00:03:19.759
side. But it's interesting when you
look at that synaptic plasticity at the interface

45
00:03:19.800 --> 00:03:24.080
side, there might be like an
emergent plasticity because it's a whole nother It's

46
00:03:24.080 --> 00:03:27.719
not like in the brain. It's
a whole nother extension of the brain.

47
00:03:28.639 --> 00:03:31.879
You know, we might have to
redefine what it means to be malleable for

48
00:03:31.960 --> 00:03:36.319
the brain, so maybe the brain
is able to adjust to external interfaces.

49
00:03:36.479 --> 00:03:38.919
There will be some adjustments to the
brain because there's gonna be something reading and

50
00:03:38.919 --> 00:03:47.360
simulating the brain, and so it
will adjust to that thing. But most

51
00:03:47.439 --> 00:03:53.159
the vast majority of the adjustment will
be on the machine side. This is

52
00:03:53.240 --> 00:03:54.960
just, this is just. It
has to be that otherwise it will not

53
00:03:55.039 --> 00:03:59.240
work. Ultimately, like we currently
operate on two layers. We have sort

54
00:03:59.240 --> 00:04:03.319
of lumbaco like prime primative brain layer, which is where all of our kind

55
00:04:03.319 --> 00:04:08.360
of impulses are coming from. It's
sort of like we've got we've got like

56
00:04:08.400 --> 00:04:12.479
a monkey brain where a computer stuck
on it. That's that's the human brain,

57
00:04:13.000 --> 00:04:15.719
and a lot of our impulses and
everything are driven by the monkey brain

58
00:04:15.600 --> 00:04:20.120
and the computer. The cortex is
constantly trying to make the monkey monkey brain

59
00:04:20.160 --> 00:04:25.439
happy. It's not the cortex that's
steering the monkey brains. The monkey brain

60
00:04:25.439 --> 00:04:30.240
is steering the cortex, you know. But the cortex is the part that

61
00:04:30.319 --> 00:04:33.879
tells the story of the whole thing. So we convince ourselves it's more interesting

62
00:04:33.920 --> 00:04:38.560
than just the monkey brain. The
cortex is like what we call like human

63
00:04:38.600 --> 00:04:41.959
intelligence, you know. So it's
like the that's like the advanced computer relative

64
00:04:42.000 --> 00:04:48.040
to other creatures. Other creatures do
not have either really they don't have the

65
00:04:48.079 --> 00:04:55.879
computer, or they have a very
weak computer relative to humans. But it's

66
00:04:56.199 --> 00:04:59.920
it's like it sort of seems like, surely the really smart thing should can

67
00:05:00.560 --> 00:05:04.519
the dumb thing, but actually,
don't think controls a small thing. So

68
00:05:04.560 --> 00:05:09.240
do you think some of the same
kind of machine learning methods or whether that's

69
00:05:09.319 --> 00:05:14.800
natural language processing applications are going to
be applied for the communication between the machine

70
00:05:15.639 --> 00:05:19.920
and the brain to learn how to
do certain things like movement of the body,

71
00:05:20.079 --> 00:05:26.160
how to process visual stimuli and so
on. Do you see the value

72
00:05:26.199 --> 00:05:30.720
of using machine learning to understand the
language of the two way communication with the

73
00:05:30.720 --> 00:05:35.000
brain? Sure? Yeah, absolutely. I mean we're neural net and that

74
00:05:35.720 --> 00:05:41.360
you know, AI is basically neural
net. So it's like digital neural net

75
00:05:41.399 --> 00:05:46.920
will interface with biological neural net and
hopefully bring us along for the ride.

76
00:05:46.639 --> 00:05:54.839
But the vast majority of our of
our intelligence will be digital. There's like

77
00:05:54.920 --> 00:06:00.879
like like think of like the difference
in intelligence between the cortex in your olympics

78
00:06:00.920 --> 00:06:06.279
system is gigantic. Your olympic system
really has no comprehension of what the hell

79
00:06:06.360 --> 00:06:12.759
the cortex is doing. You know, it's just literally hungry, you know,

80
00:06:13.040 --> 00:06:18.600
or tired or angry or taxi or
something. You know, it's just

81
00:06:19.120 --> 00:06:25.279
and then that in case, that's
that impulse to the cortex and tales the

82
00:06:25.279 --> 00:06:28.879
cortex to go satisfy that. So
then a lot of a great deal of

83
00:06:29.160 --> 00:06:33.240
like a massive amount of thinking,
like truly stupendous amount of thinking has gone

84
00:06:33.279 --> 00:06:43.199
into sex without purpose, without procation, without procreation, which which which is

85
00:06:43.240 --> 00:06:47.959
actually quite a silly action in the
absence of procreation. It's it's a bit

86
00:06:48.000 --> 00:06:53.120
silly. Why are you doing it? Because it makes the limbic system happy,

87
00:06:53.160 --> 00:06:58.759
that's why. That's why. But
it's pretty absurd, really, well,

88
00:06:58.839 --> 00:07:00.839
the whole of existence, that's pretty
absurd in some kind of sense.

89
00:07:01.160 --> 00:07:04.959
Yeah, But I mean, this
is a lot of computation has gone into

90
00:07:05.319 --> 00:07:10.879
how can I do more of that
with approcreation not even being a factor?

91
00:07:11.959 --> 00:07:17.160
This is I think a very important
era of research, finds FW an agency

92
00:07:17.199 --> 00:07:20.560
that should receive a lot of funding, especially after this conversation. If I

93
00:07:20.639 --> 00:07:30.000
propose the formation of a new agency, Oh boy, what is the most

94
00:07:30.040 --> 00:07:33.360
exciting or some of the most exciting
things that you see in the future impact

95
00:07:33.360 --> 00:07:40.680
of neuralink both on the science and
engineering and societal broad impact. So neuralink,

96
00:07:40.800 --> 00:07:44.600
I think at first we'll solve a
lot of brain related diseases. So

97
00:07:45.759 --> 00:07:48.879
it could be anything from like autism, schizophrenia, memory loss, like everyone

98
00:07:48.959 --> 00:07:54.800
experiences memory loss at certain points in
an age. Parents can't remember their kids'

99
00:07:54.839 --> 00:07:58.439
names and that kind of thing.
So there's a tremense amount of good that

100
00:07:59.040 --> 00:08:07.279
neuralink can do in solving critical critical
damage to the brain or the spinal cord.

101
00:08:07.800 --> 00:08:11.040
There's a lot that can be done
to improve quality of life of individuals,

102
00:08:11.480 --> 00:08:13.959
and that will be those will be
steps along the way, and then

103
00:08:15.040 --> 00:08:22.480
ultimately it's intended to address the risk
existential risk associated with a digital superintelligence,

104
00:08:24.560 --> 00:08:31.759
like we will not be able to
be smarter than a digital supercomputer. So

105
00:08:31.839 --> 00:08:35.960
therefore, if you cannot beat them, join them and release, we won't

106
00:08:35.000 --> 00:08:41.039
have that option. So you have
hope that your link will be able to

107
00:08:41.559 --> 00:08:46.000
be a kind of connection to allow
us to merge to ride the wave of

108
00:08:46.039 --> 00:08:52.000
the improving AI systems. I think
the chance is above zero percent, so

109
00:08:52.039 --> 00:08:56.440
it's non zero. There's a chance, and that's what have you seen,

110
00:08:56.480 --> 00:09:01.639
Doumin Dummer? Yes, yes,
so I'm saying there's a chance. He's

111
00:09:01.679 --> 00:09:03.360
saying one in a billion or one
in a million whatever. It was a

112
00:09:03.440 --> 00:09:07.200
dumb and dumber you know. It
went from maybe one in a million to

113
00:09:07.320 --> 00:09:09.559
improving. Maybe it'll be one in
a thousand, and then one one hundred,

114
00:09:09.679 --> 00:09:13.639
then one in ten. Depends on
the rate of improvement of neuralink and

115
00:09:13.679 --> 00:09:18.720
how fast we're able to do make
progress. You know. Well, I've

116
00:09:18.759 --> 00:09:22.399
talked to a few folks here that
quite brilliant engineers, so I'm excited.

117
00:09:22.720 --> 00:09:24.639
Yeah, I think it's like fundamentally
good, you know, you know,

118
00:09:24.679 --> 00:09:31.240
giving somebody back full motor control after
they've had a spinal cord injury, you

119
00:09:31.279 --> 00:09:37.480
know, restoring brain functionality after a
stroke, solving debilitating genetically orange brain diseases.

120
00:09:37.559 --> 00:09:41.200
These are all incredibly great, I
think. And in order to do

121
00:09:41.279 --> 00:09:46.000
these, you have to be able
to interface with neurons at detailed level and

122
00:09:46.039 --> 00:09:48.759
need to be build fire the right
neurons, read the right neurons, and

123
00:09:50.039 --> 00:09:58.240
and then effectively you can create a
circuit, replace what's broken with with silicon

124
00:09:58.679 --> 00:10:05.919
and such a fill in the the
missing functionality, and then over time we

125
00:10:05.039 --> 00:10:09.960
can have we develop a tertiary layer. So if like Olympic system is a

126
00:10:09.000 --> 00:10:13.320
primary layer, then the cortex is
like the second layer. And I said

127
00:10:13.320 --> 00:10:16.039
that, you know, obviously the
cortex is vastly more intelligent than the Olympic

128
00:10:16.120 --> 00:10:20.799
System. But people generally like the
fact that they have Olympic System and cortex.

129
00:10:20.159 --> 00:10:24.320
I haven't met anyone who wants to
lead either one of them. They're

130
00:10:24.360 --> 00:10:26.799
like, okay, I'll keep them
both. That's cool. The Olympic System

131
00:10:26.840 --> 00:10:30.639
is kind of fun. That's what
the fun is. Absolutely, And then

132
00:10:31.200 --> 00:10:35.639
people generally don't lose the cortex either, right, so they're like having the

133
00:10:35.679 --> 00:10:39.639
cortex and the Olympic System. Yeah, and then there's a tertiary layer,

134
00:10:39.679 --> 00:10:45.440
which will be digital superintelligence. And
I think there's room for optimism given that

135
00:10:45.600 --> 00:10:50.480
the cortex, the cortex is very
intelligent and the Olympic System is not.

136
00:10:50.720 --> 00:10:54.879
Yet they work together. Well,
perhaps they can be a tertiary layer where

137
00:10:56.039 --> 00:10:58.960
digital superintelligence lies, and that will
be vastly more intelligent than the cortex,

138
00:11:00.679 --> 00:11:05.960
but still coexist peacefully and in of
a nine manner with the cortex Olympic System.

139
00:11:05.720 --> 00:11:09.440
That's a super exciting future, both
in low level engineering that I saw

140
00:11:09.480 --> 00:11:16.080
as being done here and actual possibility
in the next few decades. It's important

141
00:11:16.080 --> 00:11:20.240
that neuralinks solved this problem sooner rather
than later, because the point at which

142
00:11:20.279 --> 00:11:24.879
we have digital superintelligence, that's when
we pass the singularity and things become just

143
00:11:24.000 --> 00:11:26.320
very uncertain. It doesn't mean that
they're necessarily bad or good. For the

144
00:11:26.360 --> 00:11:31.480
point of which we passed singularity,
things become extremely unstable. So we want

145
00:11:31.559 --> 00:11:35.559
to have a human brain interface before
the singularity, or at least not long

146
00:11:35.679 --> 00:11:41.039
after it, to minimize existential risk
for humanity and consciousness as we know it.

147
00:11:43.600 --> 00:11:46.200
But there's a lot of fascinating actual
engineering and low level problems here in

148
00:11:46.279 --> 00:11:52.600
your link that yeah, quite exciting. The problems that we face neuralink are

149
00:11:54.679 --> 00:12:01.240
material science, electrical engineering, software, mechanical engineering, micro vocation. It's

150
00:12:01.279 --> 00:12:05.840
a bunch of engineering disciplines essentially.
That's what it comes down to is you

151
00:12:05.919 --> 00:12:13.240
have to have a tiny electrode,
so it's so small it doesn't hurt neurons,

152
00:12:13.279 --> 00:12:16.000
but it's got to last for as
long as a person, so it's

153
00:12:16.000 --> 00:12:20.919
gonna last for decades. Uh.
And then you've got to take that signal

154
00:12:20.000 --> 00:12:26.240
you've got to process that single signal
locally at low power. So we need

155
00:12:26.279 --> 00:12:30.240
a lot of chip design engineers that, you know, because we've got to

156
00:12:31.039 --> 00:12:37.159
uh signal processing and do so in
a very power efficient way so that we

157
00:12:37.240 --> 00:12:41.519
don't heat your brain of because the
brain is very heat sensitive. And then

158
00:12:41.639 --> 00:12:43.600
and then we've got to take those
signals, we're going to do something with

159
00:12:43.679 --> 00:12:48.200
them. And then we've got to
stimulate and stimulate the back to to you

160
00:12:48.279 --> 00:12:54.879
know, so you could bi directional
communication. So if somebody's good at material

161
00:12:54.919 --> 00:13:00.919
science, software, mechanical engineering,
electrical engineering, trip design, microfabrication,

162
00:13:01.879 --> 00:13:05.120
that's what those are the things we
need to work on. We need to

163
00:13:05.159 --> 00:13:07.519
your good at material science so that
the we can have tiny electrodes that last

164
00:13:07.639 --> 00:13:11.279
long time. And it's a tough
thing with the science problems is a tough

165
00:13:11.320 --> 00:13:18.360
one because you're trying to read and
simulate electrically in an electric an electrically active

166
00:13:20.320 --> 00:13:24.559
area. Your brain is very electrically
active, electrochemically active. So how do

167
00:13:24.600 --> 00:13:30.799
you have say, a coding on
the electrode that doesn't dissolve over time and

168
00:13:31.200 --> 00:13:33.519
uh and is safe in the brain. This is a very hard problem.

169
00:13:35.399 --> 00:13:41.519
And then and then how do you
collect those signals in a way that is

170
00:13:41.559 --> 00:13:46.360
the most efficient because you really just
have very tiny amounts of power to process

171
00:13:46.440 --> 00:13:50.399
those signals, you know, and
then we need to automate the whole thing.

172
00:13:50.480 --> 00:13:54.519
So it's like laser, you know. So it's it's it's not if

173
00:13:54.559 --> 00:13:58.639
this is done by neurosurgeons. There's
no way it can scale to a large

174
00:13:58.679 --> 00:14:01.039
numbers of people, and it needs
to scales large numbers of people because I

175
00:14:01.039 --> 00:14:05.600
think ultimately we want the future repeated
to be determined by a large number of

176
00:14:07.039 --> 00:14:13.399
humans. Do you think that's this
has a chance to revolutionize surgery period,

177
00:14:13.720 --> 00:14:18.519
So neurosurgery, and yeah, for
sure, it's gotta be like laser.

178
00:14:18.679 --> 00:14:22.440
Like if laser had to be hand
done done by hand by a person,

179
00:14:22.080 --> 00:14:30.000
that wouldn't be great. You know, it's done by a robot and then

180
00:14:30.120 --> 00:14:33.399
arphomologist kind of just needs to make
sure your your heads in the right position

181
00:14:33.519 --> 00:14:35.759
and then they just press the button
and go