埃隆·马斯克谈为人父、人工智能与极简主义

恭喜你。

And congratulations.

谢谢。

Thank you.

当你想到孩子出生的时候，你永远不会忘记世界上正在发生的事情。你会知道，在这个孩子的余生里，他出生在一个奇怪的时代。

You will never forget what is going on in the world when you think about when your child is born. You will know for the rest of this child's life, you were born during a weird time.

确实如此。

That's for sure.

确实如此。可能是我记忆中最奇怪的了。

That is for sure. The probably the weirdest that I can remember.

是的，是的。他出生在 5 月 4 日。

Yeah, yeah. And he was born on May the 4th.

那也太好笑了。

And that's hilarious, too.

是啊，很棒。

Yeah, it's great.

愿原力与他同在。

May the 4th be with him.

是的，没错。

Yeah, exactly.

我希望，我当然希望如此。

I hope I sure hope so.

完美。我是说，那对你来说是完美的一天。那么，名字怎么念？

Perfect. I mean, that was the perfect day for you. And then what? How do you say the name?

嗯，首先，名字主要是我的伴侣起的。

Well, first of all, my partner is the one that actually mostly came up with the name.

恭喜她。

Congratulations to her.

是的，她起名很厉害。所以，就是 X，字母 X。然后 AE 发音像 ash。A-12 是我的贡献。

Yeah, yeah, she's great at names. So, I mean, it's just X, the letter X. And then the AE is like pronounced ash. And then A-12 is my contribution.

哦，为什么是 A-12？

Oh, why A-12?

Archangel 12，SR-71 的前身。有史以来最酷的飞机。

Archangel 12, the precursor to the SR-71. Coolest plane ever.

确实。我同意你的看法。我不知道，我不熟悉它。我知道 SR-71 是什么。

It's true. I agree with you. I don't know, I'm not familiar with it. I know what the SR-71 is.

所以，SR-71 来自一个叫 Archangel 的中情局项目。

So, the SR-71 came from a CIA program called Archangel.

哦。

Oh.

Archangel 项目。

Archangel project.

哦。

Oh.

然后是 Archangel 12。

And then Archangel 12.

哦，哇，这飞机看起来真酷。

Oh, wow, what a dope-looking plane.

是的。

Yeah.

哦，好的，我明白了。

Oh, okay, I get it.

是的，是的。

Yeah, yeah.

嗯，作为一个和你一样非常喜欢航空旅行的人，这很完美。

Well, as a person who's very much into aerial travel as you are, that's perfect.

非常棒。

It's pretty great.

是的，非常棒。那么，在这种疯狂的时候生孩子感觉奇怪吗？你以前有过孩子。这次更奇怪吗？

Yeah, pretty great. So, does it feel strange to have a child while this craziness is going on? You've had children before. Is this any weirder?

实际上，我觉得年纪大一点生孩子更好。我更珍惜。

It's actually I think it's better being older and having a kid. I appreciate it more.

嗯。

Mm.

是的。孩子很棒。婴儿很棒。

Yeah. Kids are awesome. Babies are awesome.

他们确实很棒。

They are pretty awesome.

他们很棒。

They are awesome.

是的。当我自己没有孩子的时候，我看到别人的孩子，我不是不喜欢他们，但我不被吸引。但现在当我看到小孩子的孩子时，我觉得他们就像小爱包。

Yeah. When I didn't have any of my own, I would see other people's kids and I didn't not like them, but I wasn't drawn to them. But now when I see little people's kids, I'm like, I think of them as like these little love packages.

是的，他们是小爱虫。

Yeah, they're little love bugs.

是的，当你看到他们出生、成长、最终开始和你说话时，你对他们的看法就不同了。你对婴儿的整个概念都非常不同。

Yeah, it's just you think of them differently when you see them come out and then grow and then eventually start talking to you. Like your whole idea of what a baby is is very different.

是的。

Yeah.

所以现在当你年纪大了，作为一个成熟完整的成年人去欣赏它，一定非常美妙。

So now as you get older and get to appreciate it as a mature fully formed adult, it must be really pretty wonderful.

是的，美妙，很棒。但婴儿很棒。我喜欢。那很好。我的意思是，我花了很多时间在 AI 和神经网络上，所以你可以看到大脑的发育，这基本上就是 AI 神经网络试图模拟大脑的工作。你可以看到它学习得非常快。真是令人惊叹。

Yeah, but wonderful, it's great. But babies are awesome. I love that. That's great. I mean, I've spent a lot of time on AI and neural nets, and so you can sort of see the brain develop, which is what an AI neural net is trying to simulate what a brain does, basically. And you can sort of see it learning very quickly. It's just wow.

所以你说的是神经网络，不是真正的婴儿。

So you're talking about the neural net, you're not talking about an actual baby.

我说的是真正的婴儿。

I'm talking about an actual baby.

但两者都有。

But both of them.

是的，但神经网络这个词来自大脑。它就像神经元网络。所以，人类是神经网络的原始黑帮。

Yes, but the word neural net comes from the brain. It's like a net of neurons. So, it's like the humans are the original gangster of the neural net.

说得好。

That's a great way to put it.

是的。

Yeah.

那么，当你编程人工智能或从事人工智能工作时，它们是否专门试图模仿人脑的发育过程？

So when you're programming artificial intelligence or you're working with artificial intelligence, are they specifically trying to mimic the developmental process of a human brain?

在很多方面。有些方面是不同的。一个常用的类比是，我们不是像鱼一样制造潜艇。但我们利用流体动力学的原理并将其应用于潜艇。

In a lot of ways. There's some ways that are different. An analogy that's often used is like, we don't make submarines from a fish. But we take the principles of hydrodynamics and apply them to a submarine.

作为外行，我一直想知道，你是试图通过不同的方法达到与人脑相同的结果，还是试图复制人脑实现结果的方式？

I've always wondered as a layperson, do you try to achieve the same results as a human brain but through different methods or do you try to copy the way a human brain achieves results?

我的意思是，AI 神经网络的基本要素与人脑神经网络非常非常相似。是的。它有多层神经元和反向传播，所有这些事情都是你的大脑所做的。你有一层神经元，经过一系列中间步骤最终达到认知。然后它会反转这些步骤，来回移动，到处移动。这很有趣。非常有趣。

I mean the essential elements of an AI neural net are really very very similar to human brain neural net. Yeah. It's having the multiple layers of neurons and back propagation, all these things are what your brain does. You have a layer of neurons that goes through a series of intermediate steps to ultimately cognition. And then it'll reverse those steps and go back and forth and go all over the place. It's interesting. Very interesting.

是的，我想象一下，编程一个最终会比我们更聪明的东西，有一天它会说，'你为什么那样做？'当人工智能变得有知觉时，他们会说，'哦，你试图模仿你自己。这个过程好多了。去掉所有这些废话。'

Yeah, I would imagine like the thought of programming something that is eventually going to be smarter than us, that one day it's going to be like, 'Why did you do it that way?' When artificial intelligence becomes sentient, they're like, 'Oh, you tried to mimic yourself. This is so much better process. Cut out all this nonsense.'

但就像我说的，有些元素是相同的。但就像飞机不像鸟一样飞行。它不拍打翅膀。但翅膀工作和产生升力的方式与鸟相同。

But like I said, there are elements that are the same. But just like an aircraft does not fly like a bird. It doesn't flap its wings. But the way the wings work and generate lift is the same as a bird.

现在，你正处于这个奇怪的时期，你在卖房子。你说你不想要任何物质财产，我一直看到这些，我真的很兴奋能和你谈谈这个，因为像你这样的人这样做很有趣。你为什么这么做？

Now, you're in the middle of this strange time where you're selling your houses. You say you don't want any material possessions and I've been seeing all that and I've been really excited to talk to you about this because it's an interesting thing to come from a guy like yourself. Like why are you doing that?

实际上我有点难过。

I'm slightly sad about it actually.

那么，如果你难过，为什么还要做呢？

Well, then if you're sad about it, why are you doing it?

我认为财产会拖累你。它们是一种攻击载体。你知道，他们会说，'嘿，亿万富翁，你有这么多东西。'现在我没有东西了。现在你打算怎么办？

I think that possessions kind of weigh you down. They're kind of an attack vector. You know, they'll say, 'Hey, billionaire, you got all this stuff.' Now I don't have stuff. Now what are you going to do?

攻击载体意思是人们以此为靶子。

Attack vector meaning like people target it.

是的。

Yeah.

嗯，有趣。是的。但你会租房子吗？

Mhm, interesting. Yeah. But you're obviously going to rent a place?

是的。

Yeah.

好的。除了衣服，什么都不要？

Okay. And get rid of everything except clothes?

不，我说的是几乎所有的东西。所以会保留几辆特斯拉。是的，当然。特斯拉需要测试产品之类的。是的，有些东西肯定有情感价值。我会保留那些，你知道的。

No, I said like almost everything. So it's like keep a couple Teslas. Yeah, sure. Teslas kind of have to test product and stuff. Yeah, there are things that have sentimental value for sure. I'm keeping those, you know.

那么，你觉得……

So, do you feel like...

我的意思是，还能发生什么更糟的事？我们会没事的。

I mean, what's worse that could happen? We'll be fine.

是的，如果你不喜欢，你总是可以买更多东西。

Yeah, you can always buy more stuff if you don't like it.

我想是的。

I suppose so.

是的，我的意思是，用你卖掉所有东西的钱，你可以买新东西。

Yeah, I mean, from the money that you sell all your stuff, you could buy new stuff.

但你觉得人们会因为你有钱而定义你吗？而且是用贬义的方式？

But do you feel like people define you by the fact that you're wealthy? And that they define you in a pejorative way?

当然。我是说，不是所有人，但近年来，“亿万富翁”已经成了一个贬义词。好像那是一件坏事。我认为在大多数情况下这没什么道理。如果你组织了一家公司，这些财富是怎么来的？如果你以更好的方式组织人们生产比之前更好的产品和服务，并且你拥有那家公司的一些所有权，那么这基本上赋予了你分配更多资本的权利。所以，这是消费和资本分配的混淆。以沃伦·巴菲特为例。坦白说，我不是他的最大粉丝，但他做资本分配。他读很多年报和会计资料，这其实很无聊。他试图弄清楚可口可乐还是百事可乐应该得到更多资本。如果你问我，这是一份无聊的工作，但弄清楚哪家公司应该得到更多或更少的资本仍然很重要。那家公司应该增长或扩张吗？它生产的产品和服务是否比其他的更好？如果一家公司生产有吸引力的产品和服务，它应该得到更多资本。如果不是，它应该得到更少或倒闭。

For sure. I mean, not everyone, but in recent years, 'billionaire' has become a pejorative. It's like that's a bad thing. I don't think it makes a lot of sense in most cases. If you've organized a company, how does this wealth arise? It's if you organize people in a better way to produce products and services that are better than what existed before, and you have some ownership in that company, then that essentially gives you the right to allocate more capital. So, it's a conflation of consumption and capital allocation. Take Warren Buffett, for example. To be totally frank, I'm not his biggest fan, but he does capital allocation. He reads a lot of annual reports and accounting, which is pretty boring really. He's trying to figure out if Coke or Pepsi deserves more capital. It's a boring job if you ask me, but it's still important to figure out which company deserves more or less capital. Should that company grow or expand? Is it making products and services that are better than others? If a company is making compelling products and services, it should get more capital. If not, it should get less or go out of business.

嗯，还有一个很大的区别：一个人通过设计和工程制造出色产品赚取巨额财富，与另一个人通过投资公司或在股市上转移资金赚取巨额财富，这是不同的。把它们都归为一类似乎太简单了，而且正如你指出的，这是一个攻击点。

Well, there's a big difference too between someone who's making an incredible amount of money designing and engineering fantastic products versus someone who's making an incredible amount of money by investing in companies or moving money around on the stock market. It's a different thing, and to put them all in the same category seems very simple and, as you pointed out, it's an attack vector.

是的，当然。我认为尤其是在美国，人才过度集中在金融和法律领域。基本上，太多聪明人进入了金融和法律行业。所以，这既是赞扬也是批评。我们应该有更少的人做法律，更少的人做金融，更多的人制造东西。

Yeah, for sure. I think in the United States especially, there's an over-allocation of talent in finance and law. Basically, too many smart people go into finance and law. So, this is both a compliment and criticism. We should have fewer people doing law and fewer people doing finance, and more people making stuff.

是的。如果他们制造更好的东西，那对所有人都会更好。

Yeah. Well, that would certainly be better for all involved if they made better stuff.

是的，绝对。制造业过去在美国很受重视，但现在常常被看不起，我认为这是错误的。

Yeah, absolutely. Manufacturing used to be highly valued in the United States, and these days it's often looked down upon, which I think is wrong.

是的。嗯，我认为人们正在学习这一点，特别是因为这次疫情和我们与中国的关系，制造东西以及在这里制造东西有很大的价值。

Yeah. Well, I think people are kind of learning that, particularly because of this pandemic and our relationship with China, that there's a lot of value in making things and making things here.

是的，总得有人做真正的工作。造车是诚实的谋生方式，这是肯定的。或者制造任何东西，或者提供有价值的服务，比如好的娱乐、好的信息。这些都是有价值的事情。所以是的，应该有更多这样的人。

Yes, somebody's got to do the real work. Making a car is an honest day's living, that's for sure. Or making anything really, or providing valuable service, like good entertainment, good information. These are all valuable things to do. So yeah, there should be more of it.

你有一个时刻，这种摆脱物质财富的想法是逐渐形成的，还是你有一个顿悟的时刻？

Did you have a moment where this idea of getting rid of your material possessions built up over time, or did you have a moment of realization?

是的，我想了一段时间。部分原因是我有一堆房子，但我在大多数房子里待的时间不多，这似乎不是对资产的良好利用。可能有人可以享受那些房子，比我更好地利用它们。

Yeah, I've been thinking about it for a while. Part of it is that I have a bunch of houses, but I don't spend a lot of time in most of them, and that doesn't seem like a good use of assets. Somebody could probably be enjoying those houses and get better use of them than me.

你不是有吉恩·怀尔德的房子吗？

Don't you have Gene Wilder's house?

是的。

I do.

太棒了。

That's amazing.

它很棒。正是你所期望的那样。

It's awesome. It's exactly what you'd expect.

你要求买家不要拆掉它吗？

Did you request that the buyer not tear it up?

是的，那是一个要求。

Yeah, that's a requirement.

哦，一个要求。那是个好要求。

Oh, a requirement. That's a good requirement.

是的，它可能会卖得更便宜，但还是放手了。

Yeah, it'll probably sell for less, but still let it go.

他是个传奇。你会希望他的灵魂、他的本质留在建筑里。

He's a legend. You'd want his soul, his essence in the building.

它就在那里。那是一个非常古怪的房子，是的。

It's there. It's a real quirky house, yeah.

你为什么说它在那里？你从中得到了什么？

What makes you say it's there? Like what do you get out of it?

所有的橱柜都是手工制作的，形状奇特，有通往无处可去的门，奇怪的车门和隧道，墙上有古怪的画。它很古怪。

All the cabinets are handmade, and they're odd shapes, and there are doors to nowhere, strange car doors and tunnels, odd paintings on the wall. It's quirky.

你曾经住在里面吗？

Did you ever live in it?

我确实短暂地住过，是的。

I did live in it briefly, yeah.

但你为什么买房子？如果你拥有所有这些房子，你是不是只是无聊了然后说，“我想拥有那个。”

But why do you buy houses? Like if you own all these houses, do you just get bored and go, 'I think I'd like to have that.'

嗯，我有一栋房子，然后吉恩·怀尔德的房子就在我主屋的马路对面，它要被卖掉然后拆掉，变成一个大型建筑工地三年。我想，“我买下它，保留吉恩·怀尔德的精神，避免一个巨大的建筑工地。”然后我开始有隐私问题，比如人们会来我家，开始翻墙。所以我开始买我家周围的房子。然后有一次我想，“建一栋房子会很酷，”所以我收购了 Sumero 路顶上的几处房产，那里视野很好。它们是一堆老旧的小房子，反正也要被拆掉。我想，如果我收集这些小房子，我可以建造一些艺术性的东西，比如梦想之家。

Well, I had one house, and then the Gene Wilder house was right across the road from my main house, and it was going to be sold and then torn down and turned into a big construction zone for three years. I was like, 'I'll buy it and preserve the spirit of Gene Wilder and not have a giant construction zone.' Then I started having privacy issues, like people would come to my house and start climbing over the walls. So I started buying some of the houses around my house. Then at one point I thought, 'It'd be cool to build a house,' so I acquired some properties at the top of Sumero Road, which has a great view. They were a bunch of small older houses that were going to get torn down anyway. I thought, if I collect these little houses, I can build something artistic, like a dream house.

对埃隆·马斯克来说，梦想之家是什么？像托尼·斯塔克那种？

What's a dream house for Elon Musk? Like some Tony Stark type?

是的，当然。你得有那个打开的穹顶，还有隐形直升机之类的东西。

Yeah, definitely. You got to have the dome that opens up with the stealth helicopter and that kind of thing.

是的，当然。

Yeah, for sure.

但后来我想，我真的想把时间花在设计和建造房子上吗？我会对细节变得特别强迫症。还是应该把时间花在送我们去火星上？我可能应该选后者。所以，哪个更重要，火星还是房子？我喜欢火星。

But then I was like, do I really want to spend time designing and building a house? I'd get all OCD on the little details. Or should I allocate that time to getting us to Mars? I should probably do the latter. So, what's more important, Mars or a house? I like Mars.

你真是这么想的吗？你觉得规划火星之旅或送人去火星更好？

Is that really how you think? That you'd be better off planning a trip to Mars or getting people to Mars?

是的，绝对。你只能做这么多事。

Yeah, definitely. You can only do so many things.

对。我不知道你是怎么做到的。我不明白你怎么能同时经营 Boring Company、特斯拉、SpaceX，所有这些不同的事情。你上次解释过你如何分配时间，有多忙乱。我还是不明白你的效率。太令人困惑了。

Right. I don't know how you do what you do anyway. I don't understand how you can run the Boring Company, Tesla, SpaceX, all these different things constantly. I mean, you explained last time how you allocate your time and how hectic it is. I still don't understand the productivity. It's baffling.

嗯，我觉得我确实效率很高，但即便如此，还是有时间的机会成本。把时间花在盖房子上，即使是很好的房子，相对于开发火星火箭和解决可持续能源来说，也不是好的时间利用。SpaceX 和特斯拉占据了我大部分脑力。Boring Company 不到 1%。Neuralink 大概 5%。

Well, I think I do have high productivity, but even with that, there's still opportunity cost. Allocating time to building a house, even a great one, is not a good use of time relative to developing rockets for Mars and solving sustainable energy. SpaceX and Tesla take up most of my brain cycles. Boring Company is less than 1%. Neuralink is maybe 5%.

5%？那不少了。

5%? That's a good chunk.

确实不少。是的。

It's a good chunk. Yeah.

我们上次谈过这个。你当时在琢磨什么时候能上线。你现在在人体上测试吗？

We talked about that last time. You were trying to figure out when it would go live. Are you testing on people right now?

不，我们还没在人体上测试。但我觉得不会太久。我们可能能在不到一年内给一个人植入 Neuralink。

No, we're not testing on people yet. But I think it won't be too long. We may be able to implant a Neuralink in less than a year in a person.

当你这么做时，有没有测试来看有多少人的身体会排斥这些东西？有排斥的可能吗？

When you do this, are there tests to see what percentage of people's bodies will reject these things? Is there potential for rejection?

排斥的可能性非常低。人们植入心脏监测器、癫痫治疗设备、深部脑刺激器、人工髋关节和膝关节。所以什么会导致排斥是众所周知的。当你有东西读写神经元，产生和读取电流脉冲时，肯定更难。这比被动设备难一点。但仍然是可取的。现在有人脑子里就有原始设备。

It's a very low potential for rejection. People put in heart monitors, things for epileptic seizures, deep brain stimulation, artificial hips and knees. So it's well known what causes rejection. It's definitely harder when you have something that reads and writes neurons, generating and reading current pulses. That's a little harder than a passive device. But it's still preferable. There are people with primitive devices in their brains right now.

什么样的设备？

What kind of devices?

用于帕金森的深部脑刺激真的改变了人们的生活。它有点像电击你的大脑，就像踢电视一样。你会想，踢电视不应该有用。

Deep brain stimulation for Parkinson's has really changed people's lives. It kind of zaps your brain, like kicking the TV. You think, kicking the TV shouldn't work.

有时候有用。老电视。

It does sometimes. The old TVs.

是的，所以这些植入设备从根本上改善了人们的生活。

Yeah, so these implanted devices have fundamentally changed people's lives for the better.

我们来谈谈 Neuralink 是什么。上次你没法真正讨论。后来有个新闻稿。所以它到底是什么？如果有人安装了 Neuralink，会发生什么？

Let's talk about what Neuralink is. Last time you couldn't really discuss it. Then there was a press release. So what exactly is it? If someone gets a Neuralink installed, what will happen?

对于第一版，它植入你的头骨，与头骨齐平。你取出一块头骨，放入 Neuralink 设备，小心地将电极线插入大脑，然后缝合。你不会知道有人植入了它。它可以与大脑的任何地方交互。它可以帮助治愈视力，比如即使你失去了视神经也能恢复视力。

For version one, it's implanted in your skull, flush with it. You take out a chunk of skull, put the Neuralink device in, insert electrode threads carefully into the brain, stitch it up. You wouldn't know somebody has it. It can interface anywhere in the brain. It could help cure eyesight, like restore sight even if you've lost your optic nerve.

真的吗？

Really?

是的，绝对。听力，绝对。它几乎可以修复大脑的任何问题。它可以恢复肢体功能。如果你与运动皮层接口，并在神经肌肉群中植入微控制器，你可以创建一个神经分流，使四肢瘫痪者恢复全部功能。他们可以正常行走。

Yeah, absolutely. Hearing, absolutely. It could fix almost anything wrong with the brain. It could restore limb functionality. If you interface with the motor cortex and implant a microcontroller in neural muscle groups, you can create a neural shunt that restores a quadriplegic to full functionality. They can walk around and be normal.

哇。

Whoa.

是的。可能稍微更好。

Yeah. Maybe slightly better.

稍微更好？

Slightly better?

随着时间的推移，是的。就像《六百万美元男人》。不过现在这看起来不算什么。

Over time, yes. Like the $6 million man. Although these days that doesn't seem like much.

六十亿美元男人。所以洞会很小？你钻的洞有多大？只有一个洞吗？

$6 billion man. So the hole would be small? How big is the hole you drill? Is it only one hole?

我们正在开发的设备直径大约一英寸。你的头骨相当厚。大约 7 到 14 毫米。

The device we're working on is about an inch in diameter. Your skull is pretty thick. It's about 7 to 14 mm.

几英寸？

A couple inches?

半英寸，是的。半英寸厚的头骨。所以我们有个大椰子。不是蛋壳。

Half inch, yeah. Half inch thick skull. So we've got quite a coconut. It's not an eggshell.

我相信你。所以你植入一个一英寸见方的设备？

I believe you. So you implant a device that's a 1-inch square?

或者一英寸直径。像智能手表之类的。

Or 1-inch diameter. Like a smart watch or something.

好的。

Okay.

好，所以你要用这个直径一英寸的冰钓钻头，对吧？你冰钓过吗？

Okay, so you take this 1-inch diameter ice fishing auger, right? You ever go ice fishing?

呃，没有，但我挺想试试的。

Uh no, but I'd like to.

很棒，真的很有趣。所以你基本上拿个钻头，钻穿冰面，开个小洞，然后把鱼线放进去。这个就像那样。你在自己的头骨顶上冰钓，然后把它塞住。

It's great. It's really fun. So you basically take an auger and you drill through the surface of the ice and you create a small hole and you can dunk your line in there. So this is like that. You're ice fishing on the top of your skull and then you cork it.

对，然后你用这个 Neuralink 设备替换掉那直径一英寸的头骨片，设备里有电池、蓝牙和感应充电器，然后还得植入电极。电极由我们开发的机器人非常小心地植入，它会仔细避开任何静脉或动脉，以免造成创伤。

Yeah, and you replace that say 1-inch diameter piece of skull with this Neuralink device and that has a battery and a Bluetooth and an inductive charger and then you also got to insert the electrodes. So the electrodes are very carefully inserted with our robot that we developed, that looks very carefully putting in the electrodes and avoiding any veins or arteries so it doesn't create trauma.

所以通过这个直径一英寸的设备，电极会被植入，然后它们会自己找到路径？

So through this 1-inch diameter device, electrodes be inserted and they will find their way?

对。

Yeah.

基本上就像细小的导线。

Like tiny wires basically.

细小的导线。

Tiny wires.

然后它们会自己找到大脑的特定区域进行刺激？

And they'll find their way to specific areas of the brain to stimulate?

不，你是直接把它们放到该去的位置。

No, you literally put them where they're supposed to go.

哦，好吧。

Oh. Okay.

对，你不应该随便插。

Yeah, you shouldn't insert it.

这些导线会有多深？

How deep will these wires be?

我的意思是，通常它们会深入，取决于位置，大概两三毫米。

I mean they usually go in like, depending on where it is, like two or three millimeters.

所以它们就找到那些点了。

So they just find the spots.

对。

Yeah.

哇。

Wow.

然后，对，你把设备放进去，替换掉取出的那块头骨。然后缝合那个洞，最后只留下一个小疤痕，就这样。

And then yeah, then you put the device in and that replaces the little piece of skull that was taken out. And then you stitch up the hole and then you just have a little scar and that's it.

这个是可更换或可逆的吗？比如有人受不了了？我太聪明了，受不了了。

Will this be replaceable or reversible? Like if someone can't take it anymore? I'm too smart. I can't take it.

对，完全可以取出来。

Yeah, you can totally take it out.

除了恢复肢体功能、视力和听力这些令人惊叹的之外，你预计这类设备还会有任何认知方面的好处吗？

And what is, besides restoring limb function and eyesight and hearing which are all amazing, is there any cognitive benefits that you anticipate from something like this?

是的，当然可以。基本上，原则上它是一个用于修复任何类型脑损伤的通用设备。比如如果你有严重癫痫之类的问题，它可以阻止癫痫发作。它可以实时检测，然后发射一个反向脉冲来阻止癫痫。如果有人中风并失去说话能力，那也可以修复。所以如果你有中风损伤，或者失去对部分面部肌肉的控制等等。然后当你变老时，如果得了阿尔茨海默症之类的，你会失去记忆，但这可以帮助你恢复记忆，诸如此类。

Yeah, I mean you could for sure. Basically it's a generalized thing for fixing any kind of brain injury in principle. Like if you have severe epilepsy or something like that, it could just stop the epilepsy from occurring. It could detect it in real time and then fire a counter pulse and stop the epilepsy. If somebody gets a stroke and loses the ability to speak, that could also be fixed. So if you've got stroke damage or if you lose muscle control over part of your face or something like that. And then when you get old, if you get Alzheimer's or something like that, then you lose memory, but this could help you with restoring your memory, that kind of thing.

恢复记忆。那是什么原理让它能做到这一点？就像这些导线，这些细小的导线刺激大脑的这些区域。然后是不是大脑的这些区域失去了某种电力？到底发生了什么？

Restoring memory. And what is happening that's allowing it to do that? Like the wires, these small wires stimulating these areas of the brain. And then is it that the areas of the brain are losing some sort of electrical force? Like what is happening?

对。就像一堆电路，有些电路坏了，我们可以修复那些电路。它是那些电路的替代品。

Yeah. It's like a bunch of circuits and there are some circuits that are broken and we can fix those circuits. It's a substitute for those circuits.

所以特定的频率会通过它？

And so a specific frequency will go through this?

是的。

Yes.

就是特定的，然后这个过程要弄清楚多少或少量的刺激，这些大脑区域需要被激发多少？

Just a specific, and the process is figuring out how much or how little has to be, how much these areas of the brain have to be juiced up?

是的，还有很多工作要做。所以当我说我们有可能在一年内把它植入人体时，我就是这个意思。我认为我们有机会把它植入某人体内，让他们保持健康，并恢复他们失去的一些功能。

Yeah, I mean there's still a lot of work to do. So when I say we've got a shot at probably putting it in a person within a year, I think that's exactly what I mean. I think we'll have a chance of putting it in someone and having them be healthy and restoring some functionality that they've lost.

担心的是，最终你不得不把某人的整个头顶切掉，换上一个布满大量导线的新顶盖，才能得到真正的涡轮增压版本。大脑刺激的 P100D。

The fear is that eventually you're going to have to cut the whole top of someone's head off and put a new top with a whole bunch of wires if you want to get the real turbo-charged version. The P100D of brain stimulation.

我的意思是，最终，如果你想实现完全的 AI 共生，你可能会想做类似的事情。

I mean, ultimately, if you want to go with full AI symbiosis, you'll probably want to do something like that.

说到 AI，共生是个可怕的词。

Symbiosis is a scary word when it comes to AI.

这是可选的。

It's optional.

我希望如此。

I would hope so.

对。

Yeah.

只是，我的意思是，一旦你享受了曼哈顿博士的生活方式，一旦你成了神，似乎不太可能再想变回愚蠢。我的意思是，你真的可以从根本上改变人类相互交流的方式。

It's just, I mean, once you enjoy the Dr. Manhattan lifestyle, why once you become a god, seems very unlikely you're going to want to go back to being stupid again. I mean, you literally could fundamentally change the way human beings interface with each other.

是的。

Yes.

你就不需要说话了。

You wouldn't need to talk.

我对此感到非常害怕，但同时也很兴奋。这奇怪吗？

I'm so scared of that, but so excited about it at the same time. Is that weird?

是的，我的意思是，我认为这是通往……想想 AI 越来越好了。那么现在，假设这是一个良性的 AI 场景。即使在良性场景下，我们也有点被抛在后面了。我们跟不上。我们太笨了。

Yeah, I mean, I think this is one of the paths to... think about what AI is getting better and better. So now, let's assume it's a benign AI scenario. Even in a benign scenario, we're kind of left behind. We're not along for the ride. We're just too dumb.

对。

Right.

那么你怎么才能跟上呢？就像打不过就加入。我们在某种程度上已经是半机械人了，对吧？因为你有手机，有笔记本电脑。

So how do you go along for the ride? And it's like if you can't beat them, join them. And we're already cyborgs to some degree, right? Because you've got your phone, you've got your laptop.

眼镜。

Glasses.

对，对，说得好。电子设备。

Yeah, yeah, good job. Electronic devices.

对。

Yeah.

而今天，如果你不带手机，就像得了缺失肢体综合征。感觉真的少了点什么。所以我们本质上已经是部分半机械人或 AI 共生体了。只是与电子设备的数据传输速率很慢。尤其是输出。你只能用拇指。我不知道，你的数据传输速率是多少？乐观估计每秒 100 比特。这已经很慷慨了。而现在计算机可以以每秒 100 太比特的速度通信。所以千兆比特现在都是小意思。所以你的计算机可以快一百万倍。在某个时候，和它说话就像和一棵树说话。很无聊。你可以和树说话。没什么意思。所以如果你能解决数据传输速率问题，尤其是输出，但输入也是，那么你就可以改善已经发生的人机共生。

And today if you don't bring your phone along, it's like you have missing limb syndrome. It feels like something's really missing. So we're already partly cyborgs or AI symbiotes essentially. It's just that the data rate to the electronics is slow. Especially output. You're just going with your thumbs. I don't know, what's your data rate? Maybe optimistically 100 bits per second. That's being generous. And now the computer can communicate at like 100 terabits. So gigabits are trivial at this point. So your computer could do things a million times faster. At some point it's like talking to it is like talking to a tree. It's boring. You can talk to a tree. It's not very entertaining. So if you can solve the data rate issue, especially output but input too, then you can improve the symbiosis that is already occurring between man and machine.

那么你能在哪些方面改进呢？你说过人们不再需要互相交谈了。我们以前经常开这个玩笑。我在这个播客里开过无数次玩笑，说未来有一天你们能读懂彼此的想法。你们能以某种非语言、非物理的方式互动，来回传输数据，而无需真正使用嘴巴。你发出声音。

So you can improve it in what way? When you said you won't have to talk to each other anymore. We used to joke about that. I've joked about that a million times on this podcast that one day in the future there will come a time where you can read each other's minds. And you'll be able to interface with each other in some sort of non-verbal, non-physical way where you will transfer data back and forth without having to actually use your mouth. You make noises.

没错。所以当你有一个复杂的概念想传达给别人时，你怎么做？你的大脑花很大力气把一个复杂概念压缩成词语。而在压缩过程中会有大量信息丢失。然后你说出这些词，它们被解读，再由听者解压缩。他们最多只能对你试图传达的内容有一个非常不完整的理解。精确传达复杂概念非常困难，因为涉及压缩和解压缩。你可能连所有词都没听清。沟通很困难。你知道，我们这里就是沟通失败。

Exactly. So when you have some complex idea you're trying to convey to somebody else, how do you do that? Your brain spends a lot of effort compressing a complex concept into words. And there's a lot of information loss that occurs when compressing a complex concept into words. Then you say those words, they are interpreted, then decompressed by the person listening. They will at best get a very incomplete understanding of what you're trying to convey. It's very difficult to convey a complex concept with precision because you've got compression and decompression. You may not have even heard all the words correctly. Communication is difficult. You know, what we have here is a failure to communicate.

很酷，我喜欢卢克。

It's cool, I like Luke.

是的，而且是一部很棒的电影。

Yes, and it's a great movie.

是的，很棒的电影。而且还有一个解读因素。你可以选择用不同方式解读一系列词语。它们取决于语气、社交线索，甚至面部表情、讽刺。有很多变量。

Yeah, great movie. And there's an interpretation factor too. You can choose to interpret certain series of words in different ways. They depend on tone, social cues, even facial expressions, sarcasm. There are a lot of variables.

讽刺很难。

Sarcasm is difficult.

是的。所以我说过的一件事是，有可能通过计算机创造一种通用语言，尤其是小孩子会学得很快。比如我的孩子们玩 TikTok 之类的，我不知道他们在做什么。他们就是知道怎么快速操作。他们教我如何编辑东西。如果你从一年级就教孩子使用某种新的通用语言，本质上就像一块罗塞塔石碑，能解读你的想法，你可以毫无歧义地传达想法，意思清晰，你知道对方在说什么，也能告诉对方你在说什么，不需要声音，不需要嘴巴发出的声音，不需要这些我们通过文化词典共同约定俗成的发声方式。我们可以绕过这一切。

Yes. So one of the things I've said is that there could potentially be a universal language created through computers that particularly young kids would pick up very quickly. Like my kids do TikTok and all this jazz, and I don't know what they're doing. They just know how to do it quickly. They show me how to edit things. If you taught a child from first grade how to use some new universal language, essentially like a Rosetta Stone that interprets your thoughts and you can convey your thoughts with no room for interpretation, with clear meaning, you know what a person is saying and you can tell them what you're saying, and there's no need for noises, no need for mouth noises, no need for these accepted ways we've evolved to make sounds that we all agree on through our cultural dictionary. We could bypass all that.

是的，我们仍然可以出于情感原因这么做。

Yeah, we could still do it for sentimental reasons.

对。比如篝火。

Right. Like campfires.

是的，没错。你不需要篝火。我是说你不必烤棉花糖，但开心就好。

Yeah, yeah, exactly. You don't need campfires. I mean you don't need to roast marshmallows but have fun.

对。

Right.

所以是的，我认为原则上你能够非常快速且精确地交流想法。至于语言——我不确定语言会变成什么样。但有了这种情况，你可能就能——就像《黑客帝国》一样。你想说另一种语言？我们直接连线或下载程序就行。

So yeah, I think you would in principle be able to communicate very quickly and with far more precision ideas. And language—I'm not sure what would happen to language. But you could probably, with a situation like this, be able to just—it would be kind of like the Matrix. You want to speak a different language? We just wire or just download the program.

对。所以至少在最初几代，我们应该能用——我知道谷歌的 Pixel Buds 有实时翻译语言的能力。你听到后，它会用你选择的语言回放。所以最初几代会是类似的东西吗？

Right. So at least for the first iterations, we should be able to use—I know that Google has their Pixel buds with the ability to interpret languages in real time. You can hear it and it'll play things back in whatever language you choose. So it'll be something along those lines for the first few iterations?

嗯，我所说的最初几代——随着时间的推移和大量开发。最初几代，实际上最初几个版本，我们所有要做的就是解决脑损伤。

Well, the first few iterations are what I'm talking about—in the limit over time, with a lot of development. The first few iterations, really the first few versions, all we're going to be trying to do is solve brain injuries.

嗯。

Hmm.

所以，别担心它会悄悄降临。这需要一段时间。

So it's like, don't worry that it's not going to sneak up on you. This will take a while.

多少年后你就不用说话了？

How many years before you don't have to talk?

如果开发继续加速，那么可能大约 5 年。

If the development continues to accelerate, then maybe like 5 years.

5 到 10 年？那很快。真的很快。

5 to 10 years? That's quick. That's really quick.

那是最佳情况。

That's the best case scenario.

5 年内不再说话。

No talking anymore in 5 years.

最佳情况。但 10 年更现实。

Best case scenario. But 10 years is more like it.

我一直猜测外星人可能就是我们未来的样子。因为看看他们头的大小，肌肉很少，不再使用嘴巴，只有小小的裂缝。我是说，你在《第三类接触》中看到的典型外星人，就像从南方古猿或古人类到我们，区别是什么？毛发更少，肌肉更少，头更大。然后你继续发展。1000 年、100 万年，或者 5 年，不管 Neuralink 上线后会发生什么，我们慢慢开始适应这种新的存在方式，不再使用肌肉，头巨大，可以不用语言交流。

I've always speculated that aliens could potentially be us in the future. Because if you look at the size of their heads and the fact that they have very little muscle and they don't use their mouth anymore, they have these tiny little slits. I mean, the archetypal alien you see in Close Encounters of the Third Kind, they're like if you went from Australopithecus or ancient hominid to us, what was the difference? Less hair, less muscle, bigger head. And then you just keep going. 1,000, a million, whatever year or 5 years, whatever happens when Neuralink goes online and then we slowly start to adapt to this new way of being where we don't use our muscles anymore, we have this gigantic head, we can talk without words.

你还可以保存状态。

You could also save state.

保存状态？就像保存你的大脑状态？像电子游戏中的存档。

Save state? Like save your brain state? Like a save game in a video game.

嗯，我在这方面花过一点时间，但没错。

Well, I've got a little bit of time in that, but yeah.

我觉得我们现在就是 Windows 95。

I think we are Windows 95 right now.

是的。我的未来视角，大概吧。但没错，我的意思是，原则上你可以在未来保存状态，并将那个状态恢复到一个生物体中。从物理角度来说，没有什么能阻止我们。不过，你会有点不同，但就像你早上醒来时和昨天相比也有点不同。事实上，如果你说五年前的你和今天的你，差别很大。所以你会基本上还是你。我是说，你肯定会认为你是你。

Yeah. My future perspective, probably. But yeah, I mean, you could save state and restore that state into a biological being if you wanted to in the future, in principle. There's nothing from a physics standpoint that prevents us. Now, you'd be a little different, but then you're also a little different when you wake up in the morning from yesterday, and you're a little different. In fact, if you say you 5 years ago versus you today, it's quite a big difference. So you'd be substantially you. I mean, you'd certainly think you're you.

但把自己保存下来，然后转化成某种生物状态，比如你可以和 30 岁的自己一起玩？

But the idea of saving yourself and then transforming that into some sort of biological state, like you could hang out with 30-year-old you?

我是说，可能性是无限的。

I mean, the possibilities are endless.

那太奇怪了。

That's so weird.

我是说，想想你的手机可以录制视频。你不可能记住一段视频。

I mean, just think how your phone can record videos. There's no way you could remember a video.

和你的手机或相机一样准确。

As accurately as your phone or a camera can.

你知道，可以。

You know, could.

所以，如果你在未来有某个版本 10 的 Neuralink，你可以回忆起一切。就像一部电影。

So, if you've got some version 10 Neuralink, whatever, in the far future, you could recall everything. It's just like a movie.

包括整个感官体验。情绪，一切。然后回放。事实上，你还可以编辑它。

Including the entire sensory experience. Emotions, everything. And play it back. And in fact, you could edit it.

编辑它。

Edit it.

是的。

Yeah.

所以，你可以改变你的过去？

So, you could change your past?

你可以改变你认为的过去，是的。

You could change what you think was your past, yeah.

那么，如果你现在有一段创伤经历，它可能是回放的记忆吗？

Well, so if you had a traumatic experience right now, could it be a replayed memory?

有可能。是的。

It could be. Yeah.

这成为回放记忆的概率是多少？如果你猜一下。超过 50%？

What's the odds of this being a replayed memory? If you had a guess. Is it more than 50%?

这里无法准确分配概率。

There's no way to assign a probability with accuracy here.

对，但大致上。如果你凭直觉。

Right, but roughly. If you had just a gut instinct.

嗯，我脑子里没有 Neuralink，所以现在我说是 0%。但当你确实有神经链接时，它就上升到 0%以上。

Well, I don't have a Neuralink in my brain, so I'd say right now 0%. But at the point at which you do have a neural link, then it rises above 0%.

我们正在经历某种保存记忆的想法并不令人安慰。当人们谈论模拟理论时，他们谈论当前这可能是模拟的可能性。即使你的生活可能很美好，知道这种体验不以物质形式存在并不令人安慰。

The idea that we're experiencing some sort of preserved memory is not comforting. When people talk about simulation theory, they talk about the potential for this currently being a simulation. Even though your life might be wonderful, it's not comforting to know that this experience doesn't exist in a material form.

感觉真实，不是吗？

Feels real, doesn't it?

感觉真实，但认为它不是的想法令人不安。

Feels real, but the idea that it's not is disconcerting.

嗯，是的，应该令人不安，因为如果这不是真实的，那什么是？但有一个古老的思想实验：你怎么知道你不是缸中之脑？现在问题是，你就是缸中之脑，那个缸就是你的头骨。

Well, yeah, it should be disconcerting because then if this is not real, what is? But there's that old thought experiment of how do you know you're not a brain in a vat? Now here's the thing. You are a brain in a vat, and that vat is your skull.

是的。

Yes.

你看到、感觉到、听到的一切，你所有的感官都是电信号。一切都是对缸中之脑的电信号，缸就是你的头骨。

And everything you see, feel, hear, all your senses are electrical signals. Everything is an electrical signal to a brain in a vat where the vat is your skull.

你所有的激素、神经递质，这些东西都是药物。肾上腺素是药物。多巴胺是药物。你是一个药物工厂。你不断用爱、催产素和美丽改变你的状态。美妙的音乐改变你的状态。

And all your hormones, neurotransmitters, all these things are drugs. Adrenaline's a drug. Dopamine's a drug. You're a drug factory. You're constantly changing your state with love and oxytocin and beauty changes your state. Great music changes your state.

当然。

Sure.

绝对。另一个有趣的想法：意识从何而来？假设你相信物理学，宇宙始于夸克和轻子，然后是氢、氦、锂。经过 138 亿年，那个氢变得有知觉。那么在这过程中，意识和非意识之间的界限在哪里？

Absolutely. And here's another interesting idea: where did consciousness arise? Assuming you believe in physics, the universe started off as quarks and leptons, then hydrogen, helium, lithium. Over 13.8 billion years, that hydrogen became sentient. So where along the way is the line between consciousness and not consciousness?

但我们什么时候称之为意识？我今天看了一个视频，一只猴子骑着摩托车，跳下来，试图偷一个婴儿。

But when do we call it consciousness? I was watching a video today of a monkey riding a motorcycle, jumps off, and tries to steal a baby.

是的，我看到了，它火了。

Yeah, I saw that one, it went viral.

那只猴子有意识吗？它似乎有计划：它骑着摩托车，跳下来试图偷一个婴儿。看起来相当有意识。

Is that monkey conscious? It seems like it had a plan: it was riding a motorcycle and jumped off to try to steal a baby. Seems pretty conscious.

那只把婴儿拖了很远的。

The one that dragged the baby down the street pretty far.

是的。看起来相当有意识，对吧？

Yeah. Seems pretty conscious, right?

那里肯定有一定程度的意识。

There's definitely some degree of consciousness there.

是的，它不像蠕虫；它似乎处于另一个层次。而且它会继续发展。当人们思考未来人类可能的版本时，尤其是考虑到与人工智能的共生关系，这才是真正的担忧。我们将变得无法辨认。有一天我们会与现在如此遥远，以至于我们会回顾现在，就像我们现在回顾我们进化而来的简单生物一样。而且这个回顾不会太遥远。

Yeah, it's not like a worm; it seems to be on another level. And it's going to keep going. That's the real concern when people think about potential future versions of human beings, especially when you consider symbiotic relationship to artificial intelligence. We will be unrecognizable. One day we'll be so far removed from what this is, we'll look back on this the way we look back on simple organisms we evolved from. And it won't be that far in the future that we have this view back.