Elon Musk: Neuralink, AI, Autopilot, and the Pale Blue Dot
生物与进化AI 与机器学习心理与人性太空与探索技术与编程
📋 章节目录
暂无章节信息
🔑 关键词
brainhumanconsciousnessablecortexdonneuralinkmachineintelligencegotlimbicteslaengineeringdigitalgoingdonepersonsafetycivilizationcars
💬 精彩语录
"Yeah, I think the work in Neuralink will definitely shed a lot of insight into how the brain, the mind"
是的，我认为 Neuralink 的工作肯定会给我们带来很多关于大脑、心灵如何运作的见解。
— Elon Musk （10:07.840）
"it can be when you have these companies effectively achieve regulatory capture of government. The bad"
当你让这些公司有效地实现对政府的监管时，情况就可能如此。不好的
— Elon Musk （08:58.320）
"from like autism, schizophrenia, memory loss, like everyone experiences memory loss at certain points"
比如自闭症、精神分裂症、记忆丧失，就像每个人都会在某些时候经历记忆丧失一样
— Elon Musk （18:05.600）
"from like autism, schizophrenia, memory loss, like everyone experiences memory loss at certain points"
比如自闭症、精神分裂症、记忆丧失，就像每个人都会在某些时候经历记忆丧失一样
— Elon Musk （18:16.480）
"vastly more intelligent than the cortex, but still coexist peacefully and in a benign manner with the"
比大脑皮层聪明得多，但仍然与大脑和平共处
— Elon Musk （21:18.560）
🎙️ 完整对话（349 条）
Lex Fridman (00:00.000)
The following is a conversation with Elon Musk, Part 2, the second time we spoke on the podcast,
以下是与埃隆·马斯克的对话，第二部分，我们第二次在播客上发言，
Lex Fridman (00:07.280)
with parallels, if not in quality, than an outfit, to the objectively speaking greatest
与客观地说最伟大的服装相比，即使不是在质量上，也有相似之处
Lex Fridman (00:13.120)
sequel of all time, Godfather Part 2. As many people know, Elon Musk is a leader of Tesla,
有史以来的续集《教父2》。正如很多人所知，埃隆·马斯克是特斯拉的领导者，
Lex Fridman (00:20.720)
SpaceX, Neuralink, and the Boring Company. What may be less known is that he's a world
SpaceX、Neuralink 和 Boring Company。可能不太为人所知的是，他是一个世界
Lex Fridman (00:26.560)
class engineer and designer, constantly emphasizing first principles thinking and taking on big
一流的工程师和设计师，不断强调第一性原则思考和承担大事
Elon Musk (00:32.480)
engineering problems that many before him will consider impossible. As scientists and engineers,
在他之前的许多人认为不可能的工程问题。作为科学家和工程师，
Elon Musk (00:39.600)
most of us don't question the way things are done, we simply follow the momentum of the crowd.
我们大多数人不会质疑做事的方式，我们只是跟随大众的势头。
Lex Fridman (00:44.880)
But revolutionary ideas that change the world on the small and large scales happen when you
但是，当你
Elon Musk (00:51.520)
return to the fundamentals and ask, is there a better way? This conversation focuses on the
回到根本上问，有没有更好的办法？这次谈话的重点是
Elon Musk (00:57.840)
incredible engineering and innovation done in brain computer interfaces at Neuralink.
Neuralink 在脑机接口领域完成了令人难以置信的工程和创新。
Elon Musk (01:04.160)
This work promises to help treat neurobiological diseases to help us further understand the
这项工作有望帮助治疗神经生物学疾病，帮助我们进一步了解
Elon Musk (01:09.440)
connection between the individual neuron to the high level function of the human brain.
单个神经元与人脑高级功能之间的联系。
Lex Fridman (01:14.400)
And finally, to one day expand the capacity of the brain through two way communication
最后，有一天，通过双向交流扩大大脑的容量
Elon Musk (01:20.240)
with computational devices, the internet, and artificial intelligence systems.
与计算设备、互联网和人工智能系统。
Elon Musk (01:25.440)
This is the Artificial Intelligence Podcast. If you enjoy it, subscribe by YouTube,
这是人工智能播客。如果您喜欢，请通过 YouTube 订阅，
Elon Musk (01:31.040)
Apple Podcasts, Spotify, support on Patreon, or simply connect with me on Twitter
Apple 播客、Spotify、Patreon 支持，或者直接在 Twitter 上与我联系
Elon Musk (01:36.320)
at Lex Friedman, spelled F R I D M A N. And now, as an anonymous YouTube commenter referred to
莱克斯·弗里德曼 (Lex Friedman)，拼写为 F R I D M A N。现在，正如一位匿名 YouTube 评论者所说
Elon Musk (01:43.520)
our previous conversation as the quote, historical first video of two robots conversing without
我们之前的对话作为引用，历史上第一个两个机器人在没有对话的情况下进行对话的视频
Lex Fridman (01:49.440)
supervision, here's the second time, the second conversation with Elon Musk.
监督，这是第二次，与埃隆·马斯克的第二次对话。
Elon Musk (01:57.840)
Let's start with an easy question about consciousness. In your view, is consciousness
让我们从一个关于意识的简单问题开始。在你看来，意识就是
Elon Musk (02:03.120)
something that's unique to humans or is it something that permeates all matter, almost like
Elon Musk (02:07.600)
a fundamental force of physics? I don't think consciousness permeates all matter. Panpsychists
Elon Musk (02:13.680)
believe that. Yeah. There's a philosophical. How would you tell? That's true. That's a good point.
Elon Musk (02:21.120)
I believe in scientific methods. I don't know about your mind or anything, but the scientific
Elon Musk (02:24.240)
method is like, if you cannot test the hypothesis, then you cannot reach meaningful conclusion that
Elon Musk (02:28.800)
it is true. Do you think consciousness, understanding consciousness is within the
Elon Musk (02:34.000)
reach of science of the scientific method? We can dramatically improve our understanding of
Elon Musk (02:40.160)
consciousness. You know, hot press to say that we understand anything with complete accuracy,
Lex Fridman (02:47.120)
but can we dramatically improve our understanding of consciousness? I believe the answer is yes.
Elon Musk (02:53.360)
Does an AI system in your view have to have consciousness in order to achieve human level
Elon Musk (02:58.480)
or superhuman level intelligence? Does it need to have some of these human qualities that
Elon Musk (03:03.360)
consciousness, maybe a body, maybe a fear of mortality, capacity to love those kinds of
Elon Musk (03:11.120)
silly human things? There's a different, you know, there's this, there's the scientific method,
Elon Musk (03:19.440)
which I very much believe in where something is true to the degree that it is testably. So
Lex Fridman (03:25.200)
and otherwise, you're really just talking about, you know, preferences or untestable beliefs or
Elon Musk (03:34.560)
that, you know, that kind of thing. So it ends up being somewhat of a semantic question, where
Elon Musk (03:42.320)
we were conflating a lot of things with the word intelligence. If we parse them out and say,
Lex Fridman (03:46.880)
you know, are we headed towards the future where an AI will be able to outthink us in every way?
Lex Fridman (03:57.520)
Then the answer is unequivocally yes.
Elon Musk (04:01.440)
In order for an AI system that needs to outthink us in every way, it also needs to have
Lex Fridman (04:07.360)
a capacity to have consciousness, self awareness, and understanding.
Elon Musk (04:12.320)
It will be self aware. Yes, that's different from consciousness. I mean, to me, in terms of what
Lex Fridman (04:18.400)
what consciousness feels like, it feels like consciousness is in a different dimension.
Lex Fridman (04:22.640)
But this is this could be just an illusion. You know, if you damage your brain in some way,
Elon Musk (04:30.480)
physically, you get you, you damage your consciousness, which implies that consciousness
Elon Musk (04:35.920)
is a physical phenomenon. And in my view, the thing is that that I think are really quite,
Elon Musk (04:42.720)
quite likely is that digital intelligence will be able to outthink us in every way, and it will
Elon Musk (04:48.880)
simply be able to simulate what we consider consciousness. So to the degree that you would
Elon Musk (04:54.080)
not be able to tell the difference. And from the from the aspect of the scientific method,
Elon Musk (04:58.160)
it's might as well be consciousness, if we can simulate it perfectly.
Elon Musk (05:01.440)
If you can't tell the difference, when this is sort of the Turing test, but think of a more
Elon Musk (05:06.800)
sort of advanced version of the Turing test. If you're if you're talking to a digital super
Elon Musk (05:13.600)
intelligence and can't tell if that is a computer or a human, like let's say you're just having
Elon Musk (05:19.440)
conversation over a phone or a video conference or something where you're you think you're talking
Elon Musk (05:26.480)
looks like a person makes all of the right inflections and movements and all the small
Elon Musk (05:33.360)
subtleties that constitute a human and talks like human makes mistakes like a human like
Lex Fridman (05:42.400)
and you literally just can't tell is this Are you video conferencing with a person or or an AI
Elon Musk (05:49.120)
might as well might as well be human. So on a darker topic, you've expressed serious concern
Elon Musk (05:54.960)
about existential threats of AI. It's perhaps one of the greatest challenges our civilization faces,
Lex Fridman (06:02.400)
but since I would say we're kind of an optimistic descendants of apes, perhaps we can find several
Elon Musk (06:08.480)
paths of escaping the harm of AI. So if I can give you an example of an example of an example
Elon Musk (06:16.480)
of escaping the harm of AI. So if I can give you three options, maybe you can comment which do you
Elon Musk (06:21.920)
think is the most promising. So one is scaling up efforts on AI safety and beneficial AI research
Elon Musk (06:29.040)
in hope of finding an algorithmic or maybe a policy solution. Two is becoming a multi planetary
Elon Musk (06:35.920)
species as quickly as possible. And three is merging with AI and riding the wave of that
Elon Musk (06:44.000)
increasing intelligence as it continuously improves. What do you think is most promising,
Lex Fridman (06:49.280)
most interesting, as a civilization that we should invest in?
Elon Musk (06:54.640)
I think there's a lot of tremendous amount of investment going on in AI, where there's a lack
Elon Musk (06:59.200)
of investment is in AI safety. And there should be in my view, a government agency that oversees
Elon Musk (07:07.760)
anything related to AI to confirm that it is does not represent a public safety risk,
Elon Musk (07:12.960)
just as there is a regulatory authority for the Food and Drug Administration is that's for
Elon Musk (07:20.320)
automotive safety, there's the FAA for aircraft safety, which I really come to the conclusion that
Elon Musk (07:25.920)
it is important to have a government referee or referee that is serving the public interest
Elon Musk (07:31.120)
in ensuring that things are safe when when there's a potential danger to the public.
Elon Musk (07:37.120)
I would argue that AI is unequivocally something that has potential to be dangerous to the public,
Lex Fridman (07:43.920)
and therefore should have a regulatory agency just as other things that are dangerous to the public
Elon Musk (07:48.480)
have a regulatory agency. But let me tell you, the problem with this is that the government
Elon Musk (07:54.240)
moves very slowly. And the rate of the rate, the usually way a regulatory agency comes into being
Elon Musk (08:01.920)
is that something terrible happens. There's a huge public outcry. And years after that,
Elon Musk (08:09.680)
there's a regulatory agency or a rule put in place, take something like, like seatbelts,
Elon Musk (08:15.120)
it was known for a decade or more that seatbelts would have a massive impact on safety and save so
Elon Musk (08:25.840)
many lives in serious injuries. And the car industry fought the requirement to put seatbelts in
Elon Musk (08:32.000)
tooth and nail. That's crazy. Yeah. And hundreds of 1000s of people probably died because of that.
Lex Fridman (08:41.040)
And they said people wouldn't buy cars if they had seatbelts, which is obviously absurd.
Elon Musk (08:45.680)
Yeah, or look at the tobacco industry and how long they fought any thing about smoking. That's part
Elon Musk (08:51.920)
of why I helped make that movie. Thank you for smoking. You can sort of see just how pernicious
Elon Musk (08:58.320)
it can be when you have these companies effectively achieve regulatory capture of government. The bad
Elon Musk (09:11.280)
people in the community refer to the advent of digital super intelligence as a singularity.
Elon Musk (09:17.040)
That is not to say that it is good or bad, but that it is very difficult to predict what will
Elon Musk (09:23.680)
happen after that point. And then there's some probability it will be bad, some probably it'll
Elon Musk (09:28.480)
be it will be good. We obviously want to affect that probability and have it be more good than bad.
Elon Musk (09:35.920)
Well, let me on the merger with AI question and the incredible work that's being done at Neuralink.
Elon Musk (09:40.960)
There's a lot of fascinating innovation here across different disciplines going on. So the flexible
Elon Musk (09:47.280)
wires, the robotic sewing machine, that responsive brain movement, everything around ensuring safety
Lex Fridman (09:52.960)
and so on. So we currently understand very little about the human brain. Do you also hope that the
Lex Fridman (10:02.560)
work at Neuralink will help us understand more about our about our human brain?
Elon Musk (10:07.840)
Yeah, I think the work in Neuralink will definitely shed a lot of insight into how the brain, the mind
Elon Musk (10:13.840)
works. Right now, just the data we have regarding how the brain works is very limited. You know,
Elon Musk (10:20.640)
we've got fMRI, which is that that's kind of like putting us, you know, stethoscope on the outside
Elon Musk (10:28.480)
of a factory wall and then putting it like all over the factory wall and you can sort of hear
Elon Musk (10:33.200)
the sounds, but you don't know what the machines are doing, really. It's hard. You can infer a few
Elon Musk (10:38.720)
things, but it's very broad brushstroke. In order to really know what's going on in the brain,
Elon Musk (10:43.200)
you really need you have to have high precision sensors. And then you want to have stimulus and
Elon Musk (10:47.600)
response. Like if you trigger a neuron, what, how do you feel? What do you see? How does it change
Elon Musk (10:53.280)
your perception of the world? You're speaking to physically just getting close to the brain,
Lex Fridman (10:57.200)
being able to measure signals, how do you know what's going on in the brain?
Elon Musk (11:00.400)
Physically, just getting close to the brain, being able to measure signals from the brain
Elon Musk (11:04.160)
will give us sort of open the door inside the factory.
Elon Musk (11:08.480)
Yes, exactly. Being able to have high precision sensors that tell you what individual neurons
Elon Musk (11:15.280)
are doing. And then being able to trigger a neuron and see what the response is in the brain.
Lex Fridman (11:22.000)
So you can see the consequences of if you fire this neuron, what happens? How do you feel? What
Elon Musk (11:28.720)
does it change? It'll be really profound to have this in people because people can articulate
Elon Musk (11:35.520)
their change. Like if there's a change in mood, or if they can tell you if they can see better,
Elon Musk (11:43.040)
or hear better, or be able to form sentences better or worse, or their memories are jogged,
Elon Musk (11:51.040)
or that kind of thing. So on the human side, there's this incredible general malleability,
Elon Musk (11:56.880)
plasticity of the human brain, the human brain adapts, adjusts, and so on.
Lex Fridman (12:01.040)
So that's not that plastic, to be totally frank.
Lex Fridman (12:03.200)
So there's a firm structure, but nevertheless, there's some plasticity. And the open question is,
Elon Musk (12:09.040)
sort of, if I could ask a broad question is how much that plasticity can be utilized. Sort of,
Elon Musk (12:15.120)
on the human side, there's some plasticity in the human brain. And on the machine side,
Elon Musk (12:20.560)
we have neural networks, machine learning, artificial intelligence, it's able to adjust
Lex Fridman (12:26.640)
and figure out signals. So there's a mysterious language that we don't perfectly understand
Elon Musk (12:31.760)
that's within the human brain. And then we're trying to understand that language to communicate
Elon Musk (12:37.120)
both directions. So the brain is adjusting a little bit, we don't know how much, and the
Elon Musk (12:42.160)
machine is adjusting. Where do you see, as they try to sort of reach together, almost like with
Elon Musk (12:48.080)
an alien species, try to find a protocol, communication protocol that works? Where do
Elon Musk (12:53.600)
you see the biggest, the biggest benefit arriving from on the machine side or the human side? Do you
Elon Musk (12:59.360)
see both of them working together? I think the machine side is far more malleable than the
Elon Musk (13:03.680)
biological side, by a huge amount. So it'll be the machine that adapts to the brain. That's the only
Elon Musk (13:12.480)
thing that's possible. The brain can't adapt that well to the machine. You can't have neurons start
Elon Musk (13:19.120)
to regard an electrode as another neuron, because neurons just, there's like the pulse. And so
Elon Musk (13:24.960)
something else is pulsing. So there is that elasticity in the interface, which we believe is
Elon Musk (13:32.320)
something that can happen. But the vast majority of the malleability will have to be on the machine
Elon Musk (13:37.520)
side. But it's interesting, when you look at that synaptic plasticity at the interface side,
Elon Musk (13:43.680)
there might be like an emergent plasticity. Because it's a whole nother, it's not like in the
Elon Musk (13:48.560)
brain, it's a whole nother extension of the brain. You know, we might have to redefine what it means
Elon Musk (13:53.840)
to be malleable for the brain. So maybe the brain is able to adjust to external interfaces. There
Elon Musk (13:59.440)
will be some adjustments to the brain, because there's going to be something reading and simulating
Elon Musk (14:03.680)
the brain. And so it will adjust to that thing. But most, the vast majority of the adjustment
Elon Musk (14:12.400)
will be on the machine side. This is just, this is just, it has to be that otherwise it will not
Elon Musk (14:18.720)
work. Ultimately, like, we currently operate on two layers, we have sort of a limbic, like prime
Elon Musk (14:23.440)
primitive brain layer, which is where all of our kind of impulses are coming from. It's sort of
Elon Musk (14:29.680)
like we've got, we've got like a monkey brain with a computer stuck on it. That's that's the
Elon Musk (14:34.720)
human brain. And a lot of our impulses and everything are driven by the monkey brain.
Lex Fridman (14:39.360)
And the computer, the cortex is constantly trying to make the monkey brain happy.
Elon Musk (14:44.720)
It's not the cortex that's steering the monkey brains, the monkey brain steering the cortex.
Elon Musk (14:51.040)
You know, the cortex is the part that tells the story of the whole thing. So we convince ourselves
Elon Musk (14:56.000)
it's, it's more interesting than just the monkey brain. The cortex is like what we call like human
Elon Musk (15:01.360)
intelligence. You know, it's just like, that's like the advanced computer relative to other
Elon Musk (15:05.280)
creatures. The other creatures do not have either. Really, they don't, they don't have the
Elon Musk (15:11.840)
computer, or they have a very weak computer relative to humans. But it's, it's like, it sort
Elon Musk (15:19.840)
of seems like surely the really smart thing should control the dumb thing. But actually,
Elon Musk (15:24.880)
the dumb thing controls the smart thing. So do you think some of the same kind of machine learning
Elon Musk (15:30.160)
methods, whether that's natural language processing applications are going to be applied for the
Elon Musk (15:35.920)
communication between the machine and the brain to learn how to do certain things like movement
Elon Musk (15:43.040)
of the body, how to process visual stimuli, and so on. Do you see the value of using machine
Elon Musk (15:50.320)
learning to understand the language of the two way communication with the brain? Sure. Yeah,
Elon Musk (15:55.440)
absolutely. I mean, we're neural net. And that, you know, AI is basically neural net.
Lex Fridman (16:02.800)
So it's like digital neural net will interface with biological neural net.
Lex Fridman (16:08.160)
And hopefully bring us along for the ride. Yeah. But the vast majority of our intelligence will be
Elon Musk (16:14.320)
digital. Like, so like, think of like, the difference in intelligence between your cortex
Lex Fridman (16:23.120)
and your limbic system is gigantic, your limbic system really has no comprehension of what the
Elon Musk (16:29.840)
hell the cortex is doing. It's just literally hungry, you know, or tired or angry or sexy or
Elon Musk (16:40.240)
something, you know, that's just and then that communicates that that impulse to the cortex and
Elon Musk (16:47.600)
tells the cortex to go satisfy that then love a great deal of like, a massive amount of thinking,
Elon Musk (16:54.480)
like truly stupendous amount of thinking has gone into sex without purpose, without procreation,
Elon Musk (17:00.960)
without procreation. Which is actually quite a silly action in the absence of procreation. It's
Elon Musk (17:11.440)
a bit silly. Why are you doing it? Because it makes the limbic system happy. That's why. That's why.
Lex Fridman (17:17.840)
But it's pretty absurd, really. Well, the whole of existence is pretty absurd in some kind of sense.
Elon Musk (17:24.880)
Yeah. But I mean, this is a lot of computation has gone into how can I do more of that with
Elon Musk (17:32.160)
procreation not even being a factor? This is, I think, a very important area of research by NSFW.
Elon Musk (17:40.160)
An agency that should receive a lot of funding, especially after this conversation.
Lex Fridman (17:44.160)
I propose the formation of a new agency. Oh, boy.
Lex Fridman (17:48.480)
What is the most exciting or some of the most exciting things that you see in the future impact
Lex Fridman (17:53.520)
of Neuralink, both in the science, the engineering and societal broad impact?
Elon Musk (17:59.120)
Neuralink, I think, at first will solve a lot of brain related diseases. So it could be anything
Elon Musk (18:05.600)
from like autism, schizophrenia, memory loss, like everyone experiences memory loss at certain points
Elon Musk (18:11.600)
in age. Parents can't remember their kids names and that kind of thing. So it could be anything
Elon Musk (18:16.480)
from like autism, schizophrenia, memory loss, like everyone experiences memory loss at certain points
Elon Musk (18:19.280)
in age. Parents can't remember their kids names and that kind of thing. So there's a tremendous
Elon Musk (18:24.400)
amount of good that Neuralink can do in solving critical damage to the brain or the spinal cord.
Elon Musk (18:34.480)
There's a lot that can be done to improve quality of life of individuals. And those will be steps
Elon Musk (18:40.720)
to address the existential risk associated with digital superintelligence. Like we will not be
Elon Musk (18:48.240)
able to be smarter than a digital supercomputer. So therefore, if you cannot beat them, join them.
Lex Fridman (18:58.240)
And at least we won't have that option.
Lex Fridman (19:01.520)
So you have hope that Neuralink will be able to be a kind of connection to allow us to merge,
Lex Fridman (19:09.200)
the wave of the improving AI systems. I think the chance is above zero percent.
Lex Fridman (19:15.600)
So it's non zero. There's a chance. And that's what I've seen. Dumb and Dumber.
Elon Musk (19:21.920)
Yes. So I'm saying there's a chance. He's saying one in a billion or one in a million,
Elon Musk (19:26.400)
whatever it was, a dumb and dumber. You know, it went from maybe one in a million to improving.
Elon Musk (19:31.120)
Maybe it'll be one in a thousand and then one in a hundred, then one in ten. Depends on the rate
Elon Musk (19:35.040)
of improvement of Neuralink and how fast we're able to do make progress.
Elon Musk (19:41.040)
Well, I've talked to a few folks here that are quite brilliant engineers, so I'm excited.
Elon Musk (19:45.440)
Yeah, I think it's like fundamentally good, you know,
Lex Fridman (19:48.400)
giving somebody back full motor control after they've had a spinal cord injury.
Elon Musk (19:53.840)
You know, restoring brain functionality after a stroke,
Elon Musk (19:57.920)
solving debilitating genetically oriented brain diseases. These are all incredibly
Elon Musk (20:02.160)
great, I think. And in order to do these, you have to be able to interface with neurons at
Elon Musk (20:07.440)
a detailed level and you need to be able to fire the right neurons, read the right neurons, and
Lex Fridman (20:13.200)
and then effectively you can create a circuit, replace what's broken with
Elon Musk (20:19.760)
with silicon and essentially fill in the missing functionality. And then over time,
Elon Musk (20:26.000)
we can develop a tertiary layer. So if like the limbic system is the primary layer, then the
Elon Musk (20:31.120)
cortex is like the second layer. And as I said, obviously the cortex is vastly more intelligent
Elon Musk (20:36.320)
than the limbic system, but people generally like the fact that they have a limbic system
Lex Fridman (20:40.080)
and a cortex. I haven't met anyone who wants to delete either one of them. They're like,
Elon Musk (20:44.480)
okay, I'll keep them both. That's cool. The limbic system is kind of fun.
Elon Musk (20:47.440)
That's where the fun is, absolutely. And then people generally don't want to lose their
Elon Musk (20:53.360)
cortex either. They're like having the cortex and the limbic system. And then there's a tertiary
Elon Musk (20:59.360)
layer, which will be digital superintelligence. And I think there's room for optimism given that
Elon Musk (21:05.520)
the cortex, the cortex is very intelligent and limbic system is not, and yet they work together
Elon Musk (21:11.760)
well. Perhaps there can be a tertiary layer where digital superintelligence lies, and that will be
Elon Musk (21:18.560)
vastly more intelligent than the cortex, but still coexist peacefully and in a benign manner with the
Elon Musk (21:24.880)
cortex and limbic system. That's a super exciting future, both in low level engineering that I saw
Elon Musk (21:30.320)
as being done here and the actual possibility in the next few decades. It's important that
Elon Musk (21:36.080)
Neuralink solve this problem sooner rather than later, because the point at which we have digital
Elon Musk (21:40.880)
superintelligence, that's when we pass the singularity and things become just very uncertain.
Elon Musk (21:45.440)
It doesn't mean that they're necessarily bad or good. For the point at which we pass singularity,
Elon Musk (21:48.640)
things become extremely unstable. So we want to have a human brain interface before the singularity,
Elon Musk (21:55.440)
or at least not long after it, to minimize existential risk for humanity and consciousness
Elon Musk (22:01.360)
as we know it. So there's a lot of fascinating actual engineering, low level problems here at
Elon Musk (22:07.200)
Neuralink that are quite exciting. The problems that we face in Neuralink are material science,
Elon Musk (22:15.600)
electrical engineering, software, mechanical engineering, microfabrication. It's a bunch of
Elon Musk (22:22.560)
engineering disciplines, essentially. That's what it comes down to, is you have to have a
Elon Musk (22:26.080)
tiny electrode, so small it doesn't hurt neurons, but it's got to last for as long as a person. So
Elon Musk (22:35.520)
it's going to last for decades. And then you've got to take that signal, you've got to process
Elon Musk (22:40.880)
that signal locally at low power. So we need a lot of chip design engineers, because we're going to
Elon Musk (22:48.800)
do signal processing, and do so in a very power efficient way, so that we don't heat your brain
Elon Musk (22:56.320)
up, because the brain is very heat sensitive. And then we've got to take those signals and
Elon Musk (23:01.040)
we're going to do something with them. And then we've got to stimulate the back to bidirectional
Elon Musk (23:10.080)
communication. So somebody's good at material science, software, and we've got to do a lot of
Elon Musk (23:15.360)
that. So somebody's good at material science, software, mechanical engineering, electrical
Elon Musk (23:20.880)
engineering, chip design, microfabrication. Those are the things we need to work on.
Elon Musk (23:27.520)
We need to be good at material science, so that we can have tiny electrodes that last a long time.
Lex Fridman (23:32.080)
And it's a tough thing with the material science problem, it's a tough one, because
Elon Musk (23:35.760)
you're trying to read and simulate electrically in an electrically active area. Your brain is
Elon Musk (23:43.680)
very electrically active and electrochemically active. So how do you have a coating on the
Elon Musk (23:49.520)
electrode that doesn't dissolve over time and is safe in the brain? This is a very hard problem.
Lex Fridman (23:59.040)
And then how do you collect those signals in a way that is most efficient? Because you really
Elon Musk (24:06.880)
just have very tiny amounts of power to process those signals. And then we need to automate the
Elon Musk (24:12.720)
whole thing so it's like LASIK. If this is done by neurosurgeons, there's no way it can scale to
Elon Musk (24:20.960)
a large number of people. And it needs to scale to a large number of people, because I think
Elon Musk (24:24.800)
ultimately we want the future to be determined by a large number of humans. Do you think that
Lex Fridman (24:32.720)
this has a chance to revolutionize surgery period? So neurosurgery and surgery all across?
Elon Musk (24:39.040)
Yeah, for sure. It's got to be like LASIK. If LASIK had to be done by hand by a person,
Elon Musk (24:45.680)
that wouldn't be great. It's done by a robot. And the ophthalmologist kind of just needs to make
Elon Musk (24:54.320)
sure your head's in the right position, and then they just press a button and go.
Elon Musk (25:00.000)
SmartSummon and soon Autopark takes on the full beautiful mess of parking lots and their human
Elon Musk (25:05.920)
to human nonverbal communication. I think it has actually the potential to have a profound impact
Elon Musk (25:13.680)
in changing how our civilization looks at AI and robotics, because this is the first time human
Elon Musk (25:19.440)
beings, people that don't own a Tesla may have never seen a Tesla or heard about a Tesla,
Elon Musk (25:24.080)
get to watch hundreds of thousands of cars without a driver. Do you see it this way, almost like an
Elon Musk (25:30.880)
education tool for the world about AI? Do you feel the burden of that, the excitement of that,
Elon Musk (25:36.080)
or do you just think it's a smart parking feature? I do think you are getting at something
Elon Musk (25:42.160)
important, which is most people have never really seen a robot. And what is the car that is
Elon Musk (25:47.680)
autonomous? It's a four wheeled robot. Yeah, it communicates a certain sort of message with
Elon Musk (25:53.200)
everything from safety to the possibility of what AI could bring to its current limitations,
Elon Musk (25:59.520)
its current challenges, it's what's possible. Do you feel the burden of that almost like a
Elon Musk (26:04.000)
communicator educator to the world about AI? We were just really trying to make people's
Elon Musk (26:09.600)
lives easier with autonomy. But now that you mentioned it, I think it will be an eye opener
Elon Musk (26:15.040)
to people about robotics, because they've really never seen most people never seen a robot. And
Elon Musk (26:20.960)
there are hundreds of thousands of Tesla's won't be long before there's a million of them that
Elon Musk (26:25.440)
have autonomous capability, and the drive without a person in it. And you can see the kind of
Elon Musk (26:31.760)
evolution of the car's personality and, and thinking with each iteration of autopilot,
Elon Musk (26:40.080)
you can see it's, it's uncertain about this, or it gets it, but now it's more certain. Now it's
Elon Musk (26:47.600)
moving in a slightly different way. Like, I can tell immediately if a car is on Tesla autopilot,
Elon Musk (26:53.200)
because it's got just little nuances of movement, it just moves in a slightly different way.
Elon Musk (26:58.720)
Cars on Tesla autopilot, for example, on the highway are far more precise about being in the
Elon Musk (27:02.960)
center of the lane than a person. If you drive down the highway and look at how at where cars
Elon Musk (27:08.960)
are, the human driven cars are within their lane, they're like bumper cars. They're like moving all
Elon Musk (27:13.840)
over the place. The car in autopilot, dead center. Yeah, so the incredible work that's going into
Elon Musk (27:20.720)
that neural network, it's learning fast. Autonomy is still very, very hard. We don't actually know
Lex Fridman (27:27.040)
how hard it is fully, of course. You look at the most problems you tackle, this one included,
Elon Musk (27:34.880)
with an exponential lens, but even with an exponential improvement, things can take longer
Lex Fridman (27:39.520)
than expected sometimes. So where does Tesla currently stand on its quest for full autonomy?
Lex Fridman (27:47.840)
What's your sense? When can we see successful deployment of full autonomy?
Elon Musk (27:55.840)
Well, on the highway already, the the probability of intervention is extremely low.
Elon Musk (28:00.160)
Yes. So for highway autonomy, with the latest release, especially the probability of needing
Elon Musk (28:08.480)
to intervene is really quite low. In fact, I'd say for stop and go traffic,
Elon Musk (28:13.200)
it's far safer than a person right now. The probability of an injury or impact is much,
Elon Musk (28:18.880)
much lower for autopilot than a person. And then with navigating autopilot, you can change lanes,
Elon Musk (28:25.360)
take highway interchanges, and then we're coming at it from the other direction, which is low speed,
Elon Musk (28:30.320)
full autonomy. And in a way, this is like, how does a person learn to drive? You learn to drive
Elon Musk (28:35.920)
in the parking lot. You know, the first time you learn to drive probably wasn't jumping on
Elon Musk (28:40.720)
August Street in San Francisco. That'd be crazy. You learn to drive in the parking lot, get things
Elon Musk (28:45.200)
get things right at low speed. And then the missing piece that we're working on is traffic
Elon Musk (28:52.400)
lights and stop streets. Stop streets, I would say actually also relatively easy, because, you know,
Elon Musk (28:59.200)
you kind of know where the stop street is, worst case in geocoders, and then use visualization to
Elon Musk (29:04.320)
see where the line is and stop at the line to eliminate the GPS error. So actually, I'd say it's
Elon Musk (29:10.720)
probably complex traffic lights and very windy roads are the two things that need to get solved.
Elon Musk (29:19.680)
What's harder, perception or control for these problems? So being able to perfectly perceive
Elon Musk (29:24.000)
everything, or figuring out a plan once you perceive everything, how to interact with all the
Elon Musk (29:29.600)
agents in the environment in your sense, from a learning perspective, is perception or action
Elon Musk (29:35.440)
harder? And that giant, beautiful multitask learning neural network, the hottest thing is
Elon Musk (29:42.240)
having accurate representation of the physical objects in vector space. So transfer taking the
Elon Musk (29:48.960)
visual input, primarily visual input, some sonar and radar, and and then creating the an accurate
Elon Musk (29:56.880)
vector space representation of the objects around you. Once you have an accurate vector space
Elon Musk (30:02.400)
representation, the planning and control is relatively easier. That is relatively easy.
Elon Musk (30:08.160)
Basically, once you have accurate vector space representation, then you're kind of like a video
Elon Musk (30:14.560)
game, like cars and like Grand Theft Auto or something like they work pretty well. They drive
Elon Musk (30:19.600)
down the road, they don't crash, you know, pretty much unless you crash into them. That's because
Elon Musk (30:24.160)
they've they've got an accurate vector space representation of where the cars are, and they're
Elon Musk (30:27.360)
just and then they're rendering that as the as the output. Do you have a sense, high level, that
Elon Musk (30:33.520)
Tesla's on track on being able to achieve full autonomy? So on the highway? Yeah, absolutely.
Lex Fridman (30:42.000)
And still no driver state, driver sensing? And we have driver sensing with torque on the wheel.
Elon Musk (30:48.320)
That's right. Yeah. By the way, just a quick comment on karaoke. Most people think it's fun,
Lex Fridman (30:55.120)
but I also think it is a driving feature. I've been saying for a long time, singing in the car
Lex Fridman (30:59.040)
is really good for attention management and vigilance management. That's right.
Elon Musk (31:02.720)
Tesla karaoke is great. It's one of the most fun features of the car. Do you think of a connection
Elon Musk (31:08.480)
between fun and safety sometimes? Yeah, you can do both at the same time. That's great.
Elon Musk (31:12.640)
I just met with Andrew and wife of Carl Sagan, directed Cosmos. I'm generally a big fan of Carl
Elon Musk (31:19.760)
Sagan. He's super cool. And had a great way of putting things. All of our consciousness,
Elon Musk (31:25.280)
all civilization, everything we've ever known and done is on this tiny blue dot.
Elon Musk (31:29.920)
People also get they get too trapped in there. This is like squabbles amongst humans.
Elon Musk (31:34.720)
Let's not think of the big picture. They take civilization and our continued existence for
Elon Musk (31:39.680)
granted. I shouldn't do that. Look at the history of civilizations. They rise and they fall. And now
Elon Musk (31:47.760)
civilization is all it's globalized. And so civilization, I think now rises and falls together.
Elon Musk (31:56.480)
There's no there's not geographic isolation. This is a big risk. Things don't always go up. That
Elon Musk (32:05.120)
should be that's an important lesson of history. In 1990, at the request of Carl Sagan, the Voyager
Elon Musk (32:12.720)
One spacecraft, which is a spacecraft that's reaching out farther than anything human made
Elon Musk (32:18.560)
into space, turned around to take a picture of Earth from 3.6 billion years ago. And that's
Elon Musk (32:24.720)
a picture of Earth from 3.7 billion miles away. And as you're talking about the pale blue dot,
Elon Musk (32:31.520)
that picture there takes up less than a single pixel in that image. Yes. Appearing as a tiny
Elon Musk (32:37.600)
blue dot, as a pale blue dot, as Carl Sagan called it. So he spoke about this dot of ours in 1994.
Lex Fridman (32:46.640)
And if you could humor me, I was wondering if in the last two minutes you could read the words
Elon Musk (32:54.160)
that he wrote describing this pale blue dot. Sure. Yes, it's funny. The universe appears to be 13.8
Elon Musk (33:01.520)
billion years old. Earth is like four and a half billion years old.
Elon Musk (33:07.920)
In another half billion years or so, the sun will expand and probably evaporate the oceans and make
Elon Musk (33:14.320)
life impossible on Earth, which means that if it had taken consciousness 10% longer to evolve,
Elon Musk (33:19.200)
it would never have evolved at all. It's 10% longer. And I wonder how many dead one planet
Elon Musk (33:29.680)
civilizations there are out there in the cosmos.
Elon Musk (33:31.520)
That never made it to the other planet and ultimately extinguished themselves or were destroyed
Elon Musk (33:35.200)
by external factors. Probably a few. It's only just possible to travel to Mars. Just barely.
Lex Fridman (33:46.640)
If G was 10% more, it wouldn't work really.
Elon Musk (33:50.080)
If G was 10% lower, it would be easy. Like you can go single stage from the surface of Mars all the
Elon Musk (34:00.240)
way to the surface of the Earth. Because Mars is 37% Earth's gravity. We need a giant booster
Elon Musk (34:08.240)
to get off the Earth. Channeling Carl Sagan. Look again at that dot. That's here. That's home. That's us.
Elon Musk (34:25.360)
On it, everyone you love, everyone you know, everyone you've ever heard of, every human being
Elon Musk (34:30.960)
who ever was, lived out their lives. The aggregate of our joy and suffering, thousands of confident
Elon Musk (34:37.600)
religions, ideologies and economic doctrines. Every hunter and farger, every hero and coward,
Elon Musk (34:42.960)
every creator and destroyer of civilization, every king and peasant, every young couple in love,
Elon Musk (34:49.840)
every mother and father, hopeful child, inventor and explorer, every teacher of morals, every
Elon Musk (34:57.760)
corrupt politician, every superstar, every supreme leader, every saint and sinner in the history of
Elon Musk (35:06.400)
our species lived there on a mode of dust suspended in a sunbeam. Our planet is a lonely speck in the
Elon Musk (35:13.840)
great enveloping cosmic dark. In our obscurity, in all this vastness, there is no hint that help
Elon Musk (35:20.640)
will come from elsewhere to save us from ourselves. The Earth is the only world known so far to harbor
Elon Musk (35:25.680)
life. There is nowhere else, at least in the near future, to which our species could migrate. This
Elon Musk (35:32.080)
is not true. This is false. Mars. And I think Carl Sagan would agree with that. He couldn't even
Elon Musk (35:39.840)
imagine it at that time. So thank you for making the world dream. And thank you for talking today.
Lex Fridman (35:45.760)
I really appreciate it. Thank you.
🔗 相关节目