#438 – Elon Musk: Neuralink and the Future of Humanity

The following is a conversation with Elon Musk, DJ Saw, Matthew McDougall, Bliss Chapman, and Noland Arbaugh about Neuralink and the future of humanity.

以下是与埃隆·马斯克、DJ Saw、马修·麦克杜格尔、布利斯·查普曼和诺兰德·阿博关于Neuralink以及人类未来的对话。

Elon, DJ, Matthew, and Bliss are, of course, part of the amazing Neuralink team.

埃隆、DJ、马修和布利斯当然是了不起的Neuralink团队的一员。

And Noland is the first human to have a Neuralink device implanted in his brain.

而诺兰德是第一个在大脑中植入Neuralink设备的人。

I speak with each of them individually.

我分别与他们每个人交谈。

So use timestamps to jump around or, as I recommend, go hardcore and listen to the whole thing.

所以使用时间戳跳转，或者，如我所建议的，硬核地听完整期节目。

This is the longest podcast I've ever done.

这是我做过的最长的播客。

It's a fascinating, super technical and wide ranging conversation.

这是一场引人入胜、技术性极强且内容广泛的对话。

And I loved every minute of it.

我喜欢其中的每一分钟。

And now a quick few second mention of each sponsor.

现在快速用几秒钟提一下每个赞助商。

Check them out in the description.

请在描述中查看它们。

It's the best way to support this podcast.

这是支持本播客的最佳方式。

We got Cloak for privacy, Masterclass for learning, Notion for taking notes, Element for hydration, Motific for generative AI deployment, and BetterHelp for mental health.

我们有Cloak保护隐私，Masterclass用于学习，Notion用于记笔记，Element用于补水，Motific用于生成式AI部署，BetterHelp用于心理健康。

Choose wisely, my friends.

明智地选择吧，我的朋友们。

Also, if you want to maybe submit feedback or submit questions that I can answer in the podcast or just get in touch with me,

另外，如果你想提交反馈或提交我可以在播客中回答的问题，或者只是联系我，

go to lexfriedman.com slash contact.

请访问lexfriedman.com/contact。

And now, on to the full ad reads.

现在，进入完整的广告朗读。

I try to make these interesting, but if you do skip them,

我尽量让这些广告有趣，但如果你跳过它们，

please still check out our sponsors.

请仍然查看我们的赞助商。

I enjoy their stuff.

我喜欢他们的东西。

Maybe you will too.

也许你也会喜欢。

This episode is brought to you by Cloaked.

本期节目由Cloaked赞助。

A platform that lets you generate new email address and a phone number

一个让你生成新邮箱地址和电话号码的平台

every time you sign up for a new website,

每次你注册一个新网站时，

allowing your actual email and your actual phone number to remain secret from said website.

让你的真实邮箱和真实电话号码对所说的网站保密。

It seems that increasingly the right approach to the interwebs is trust no one.

似乎越来越正确的互联网做法是不要相信任何人。

Of course, there's big companies that have an implied trust because you and them understand that

当然，有些大公司有一种隐含的信任，因为你和他们都明白

if you give your data over to them and they abuse that privilege, that they would suffer as a company.

如果你把数据交给他们，而他们滥用这种特权，他们作为公司会遭受损失。

Now, I don't know if they fully understand that because I think even big companies can probably sell your data or share your data for purposes of making money, all that kind of stuff.

现在，我不知道他们是否完全明白这一点，因为我认为即使是大公司也可能为了赚钱而出售或分享你的数据，诸如此类的事情。

It's just nice to not give over your contact data unless you need to.

除非必要，不交出你的联系数据真是太好了。

So Cloaked solves that problem, makes it super easy.

所以Cloaked解决了这个问题，让它变得超级简单。

It's basically a password manager with extra privacy superpowers.

它基本上是一个带有额外隐私超能力的密码管理器。

Go to cloaked.com slash lex to get 14 days free or for a limited time.

访问cloaked.com/lex获取14天免费或限时优惠。

Use code lexpod when signing up to get 25% off of an annual Cloaked plan.

注册时使用代码lexpod可获得年度Cloaked计划25%的折扣。

This episode is also brought to you by Masterclass,

本期节目还由Masterclass赞助，

where you can watch over 200 classes from the best people in the world at their respective disciplines.

在那里你可以观看来自世界各自领域最优秀人士的200多门课程。

Phil Ivey on poker, for example. Brilliant Masterclass.

比如菲尔·艾维讲扑克。精彩的大师课。

And also reminds me of the other Phil, possibly the greatest of all time.

这也让我想起另一位菲尔，可能是有史以来最伟大的。

And if you ask him, he will definitely say he's the greatest of all time, which is Phil Hellmuth.

如果你问他，他肯定会说自己是史上最伟大的，那就是菲尔·赫尔穆特。

We were supposed to do a podcast many, many times, but I'm just not sure I can handle the level of greatness that is Phil Hellmuth.

我们本来很多很多次都打算做一期播客，但我就是不确定我能承受菲尔·赫尔穆特那种伟大程度。

No, I love him.

不，我喜欢他。

We'll probably have a podcast at some point in the future.

我们将来某个时候可能会做一期播客。

I'm not sure he has a masterclass, but he, his essence, his way of being, his infinite wisdom, and the infinite number of championships that he has won is in itself a masterclass.

我不确定他有没有大师课，但他，他的本质、他的存在方式、他无限的智慧，以及他赢得的无数冠军，本身就是一堂大师课。

But, you know, if you want to settle for another mere mortal that some people consider to be the greatest poker player of all time is Phil Ivey.

但是，你知道，如果你想退而求其次，选择另一个被一些人认为是有史以来最伟大扑克玩家的凡人，那就是菲尔·艾维。

And he has an incredible masterclass on there.

他在上面有一门很棒的大师课。

Get unlimited access to every masterclass and get an additional 15% off annual membership at masterclass.com slash lexpod.

在 masterclass.com/lexpod 获取所有大师课的无限访问权限，并享受年度会员额外 15% 折扣。

That's masterclass.com slash lexpod.

网址是 masterclass.com/lexpod。

This episode is also brought to you by Notion, a note-taking and team collaboration tool.

本期节目还由 Notion 赞助，它是一款笔记和团队协作工具。

that I've used for a long time now.

我已经用了很长时间了。

I've used it primarily for note-taking because you need a big team for team collaboration.

我主要用它来做笔记，因为团队协作需要一个大团队。

But the people who I know who have used it for the team collaboration capabilities have really loved it.

但我认识的那些使用过它团队协作功能的人，都非常喜欢它。

And the thing I very much appreciate about Notion is how effectively they've been able to integrate LLMs into their tool.

我非常欣赏 Notion 的一点是，他们能够如此高效地将大语言模型集成到他们的工具中。

Their AI assistant looks across multiple documents.

他们的 AI 助手可以查看多个文档。

You can ask questions about those multiple documents.

你可以针对这些多个文档提问。

Of course, you can do all the things you kind of expect and do them easily, like summarization or rewriting stuff or helping expand or contract the kind of stuff you've written or even generate a draft.

当然，你可以做所有你期望的事情，并且轻松完成，比如总结或重写内容，或者帮助扩展或压缩你写的东西，甚至生成草稿。

But it can also kind of allow you to ask questions of a thing, like what's the progress of the team on a set of different tasks.

但它也可以让你对某个事物提问，比如团队在一系列不同任务上的进展如何。

Notion does a good job of integrating the LLMs.

Notion 在集成大语言模型方面做得很好。

Try Notion AI for free when you go to notion.com.lex.

当你访问 notion.com.lex 时，可以免费试用 Notion AI。

That's all lowercase notion.com slash lex to try the power of Notion AI today.

全部小写的 notion.com/lex，今天就来体验 Notion AI 的强大功能。

This episode is brought to you by the thing I'm drinking right now called Element.

本期节目由我现在正在喝的东西赞助，它叫 Element。

It's my daily zero sugar and delicious electrolyte mix.

这是我每天喝的无糖又美味的电解质冲剂。

They sent me a bunch of cans of sparkling water that I loved and devoured.

他们给我寄了一堆我喜爱并喝光的苏打水。

As much as you can devour a liquid because I think that's usually applied to solid foods.

就像你能吞饮液体一样，因为我觉得“吞食”通常用于固体食物。

But I devoured it.

但我把它喝光了。

And it was delicious.

而且它很好喝。

But yeah, it's an instrumental part of my life.

但没错，它是我生活中不可或缺的一部分。

It's how I get the sodium, potassium, magnesium, electrolytes into my body.

这就是我摄入钠、钾、镁和电解质的方式。

I'm going for a super long run after this.

之后我要去跑一个超长距离。

And I have been drinking Element before.

我以前一直在喝Element。

And I sure as not going to be drinking Element after.

之后我肯定不会喝Element。

Same goes for hard training sessions and grappling.

高强度训练和格斗时也是如此。

Essential for me to feel good.

对我来说，感觉良好至关重要。

Especially when I'm fasting.

尤其是当我禁食的时候。

Especially when I'm doing low-carb diets.

尤其是当我进行低碳水饮食的时候。

All of that.

所有这些。

My favorite flavor still to this day always has been is watermelon salt,

我最喜欢的口味至今一直是西瓜盐味，

but there's a lot of other delicious flavors if you want to try them out.

但如果你想尝试的话，还有很多其他美味的味道。

Get a sample pack for free with any purchase.

任何购买都可免费获得试用装。

Try it at drinkelements.com slash lex.

请在 drinkelements.com/lex 尝试。

This episode is also brought to you by Motific, a SaaS platform

本期节目还由 Motific 赞助，这是一个 SaaS 平台，

that helps businesses deploy LLMs that are customized with RAG on organization data.

帮助企业部署基于组织数据通过 RAG 定制的 LLM。

This is another use case of LLMs, which is just mind-blowing.

这是 LLM 的另一个用例，简直令人惊叹。

take all the data inside an organization and allow the people in said organization

获取组织内的所有数据，并允许该组织中的人员

to query it, to organize it, to summarize it, to analyze it, all of that,

查询、整理、总结、分析这些数据，所有这些，

to leverage it within different products, to ask questions of how it can be improved in terms of structure and organization.

在不同产品中利用它，提出问题，探讨如何在结构和组织方面进行改进。

Also on the programming front, take all of the code in, take all of the data in,

在编程方面，把所有代码放进去，把所有数据放进去，

and start asking questions about how the code can be improved,

然后开始提问，问代码如何改进，

how it can be refactored, rewritten, all that kind of stuff.

如何重构、重写，诸如此类的事情。

Now, the challenge that Motific is solving is how to do all that in a secure way.

现在，Motific正在解决的挑战是如何以安全的方式完成所有这些。

This is like serious stuff. You can't F it up.

这可是严肃的事情。你不能搞砸。

Motific is created, I believe, by Cisco, specifically their AltShift group that does the cutting-edge R&D.

我相信Motific是由思科创建的，具体来说是他们的AltShift团队，负责前沿研发。

So these guys know how to do reliable business deployment of stuff that needs to be secure.

所以这些人知道如何对需要安全的东西进行可靠的商业部署。

It needs to be done well.

这需要做好。

So they help you go from idea to value as soon as possible.

所以他们帮助你尽快从想法走向价值。

Visit motific.ai to learn more.

访问motific.ai了解更多。

That's M-O-T-I-F-I-C dot A-I.

就是M-O-T-I-F-I-C点A-I。

This episode is also brought to you by BetterHelp, spelled H-E-L-P, help.

本期节目还由BetterHelp赞助，拼写是H-E-L-P，help。

They figure out what you need and match you with a licensed therapist in under 48 hours

他们会弄清楚你的需求，并在48小时内为你匹配一位持证治疗师，

for individuals, for couples, easy, discreet, affordable, available worldwide.

针对个人、情侣，简单、谨慎、实惠，全球可用。

I think therapy is a really, really, really nice thing.

我觉得心理治疗是一件非常非常好的事情。

Talk therapy is a really powerful thing.

谈话治疗是一种非常强大的方式。

And I think what BetterHelp does for a lot of people is introduce them to that.

我认为BetterHelp为很多人做的就是让他们接触到这一点。

It's a great first step.

这是很好的第一步。

Try it out.

试试看。

For a lot of people, it can work.

对很多人来说，它是有效的。

But at the very least, it's the thing that allows you to explore the possibility of talk therapy and how that feels in your life.

但至少，它是让你探索谈话疗法可能性以及它在你的生活中感觉如何的东西。

They've helped over 4.4 million people.

他们已经帮助了超过440万人。

That's crazy.

太疯狂了。

I think the biggest selling point is just how easy it is to get started, how accessible it is.

我认为最大的卖点就是它开始起来有多容易，有多方便。

Of course, there's a million other ways to explore the inner workings of the human mind,

当然，还有无数其他方式可以探索人类心灵的内部运作，

looking in the mirror and exploring the Jungian shadow.

照镜子，探索荣格所说的阴影。

But the journey of a thousand miles begins with one step.

但千里之行，始于足下。

So this is a good first step in exploring your own mind.

所以这是探索你自己内心的一个很好的第一步。

Check them out at betterhelp.com slash Lex and save on your first month.

请访问 betterhelp.com/lex 查看，首月可享优惠。

That's betterhelp.com slash Lex.

网址是 betterhelp.com/lex。

And now, dear friends, here's Elon Musk, his fifth time on this, the Lex Friedman podcast.

现在，亲爱的朋友们，有请埃隆·马斯克，这是他第五次做客 Lex Friedman 播客。

Drinking coffee or water?

喝咖啡还是水？

Water.

水。

I'm so over-caffeinated right now.

我现在咖啡因摄入过量了。

Do you want some caffeine?

你想来点咖啡因吗？

I mean, sure.

我是说，好啊。

There's a nitro drink.

有一款氮气饮料。

This will keep you up for like, you know, tomorrow afternoon, basically.

这基本上能让你清醒到，你知道，明天下午。

Yeah.

是啊。

So what is nitro?

那什么是氮气（饮料）？

It's just got a lot of caffeine or something?

它就是含有很多咖啡因还是什么？

Don't ask questions.

别问问题。

It's called nitro.

它叫氮气。

Do you need to know anything else?

你还需要知道别的吗？

It's got nitrogen in it.

它里面含有氮气。

That's ridiculous.

太荒谬了。

I mean, what we breathe is 78% nitrogen anyway.

我是说，我们呼吸的空气本来就有78%是氮气。

What do you need to add more for?

你干嘛还要再加更多呢？

Most people think that they're breathing oxygen, and they're actually breathing 78% nitrogen.

大多数人以为他们在呼吸氧气，其实他们呼吸的是78%的氮气。

You need like a milk bar.

你需要像牛奶吧那样的东西。

Like from Clockwork Orange.

就像《发条橙》里的那样。

Yeah.

是啊。

Yeah.

是啊。

Is that top three Kubrick film for you?

这对你来说是库布里克前三的电影吗？

Clockwork Orange is pretty good.

《发条橙》挺不错的。

I mean, it's demented.

我是说，它很疯癫。

Jarring, I'd say.

要我说，挺刺耳的。

Okay, okay, so first let's step back and big congrats on getting Neuralink implanted into a human.

好，好，那么首先让我们退一步，热烈祝贺Neuralink成功植入人体。

That's a historic step for Neuralink,

这对Neuralink来说是历史性的一步，

and there's many more to come, obviously.

显然，未来还有很多。

We have our second implant as well, so far so

我们也有了第二例植入，到目前为止

good, looks like we've got, I think, on the order of 400 electrodes that are providing signals.

看起来不错，我想我们大约有400个电极在提供信号。

Nice. Yeah.

不错。是的。

How quickly do you think the number of human participants will scale?

你觉得人类参与者的数量会以多快的速度扩大？

It depends somewhat on the regulatory approval, the rate at which we get regulatory approvals.

这在一定程度上取决于监管批准，以及我们获得监管批准的速度。

So we're hoping to do 10 by the end of this year, a total of 10, so eight more.

所以我们希望到今年年底完成10例，总共10例，也就是再多8例。

And with each one, you're going to be learning a lot of lessons about the neurobiology of the brain, everything,

每做一例，你都会学到很多关于大脑神经生物学的经验，所有东西，

the whole chain of the Neuralink, the decoding, the signal processing, all that kind of stuff.

整个Neuralink链条，解码、信号处理，所有这类东西。

Yeah. Yeah, I think it's obviously going to get better with each one.

是的。是的，我觉得显然每做一例都会变得更好。

I mean, I don't want to jinx it, but it seems to have gone extremely well with the second implant.

我是说，我不想乌鸦嘴，但第二例植入似乎进行得极其顺利。

So there's a lot of signal, a lot of electrodes.

所以有很多信号，很多电极。

It's working very well.

它运行得非常好。

What improvements do you think we'll see in Neuralink in the coming, let's say, let's get crazy, the coming years?

你觉得在接下来，比如说，咱们大胆一点，接下来几年里，我们会在Neuralink上看到哪些改进？

I mean, in years it's going to be gigantic

我的意思是，几年后它会变得非常巨大

because we'll increase the number of electrodes dramatically.

因为我们会大幅增加电极的数量。

We'll improve the signal processing.

我们会改进信号处理。

So even with only roughly 10-15% of the electrodes working with Noland, with our first patient, we're able to get to achieve a bits per second.

所以即使只有大约10-15%的电极与诺兰——我们的第一位患者——一起工作，我们也能达到每秒若干比特。

That's twice the world record.

这是世界纪录的两倍。

So I think we'll start like vastly exceeding the world record by orders of magnitude in the years to come.

所以我认为，在未来几年里，我们将开始大幅超越世界纪录，甚至高出几个数量级。

So it's like getting to, I don't know, 100 bits per second, 1,000, you know, maybe if it's like five years from now, we might be at a megabit.

所以这就像达到，我不知道，每秒100比特，1000比特，你知道，也许如果从现在起五年后，我们可能达到每秒一兆比特。

Like faster than any human could possibly communicate by typing or speaking.

就像比任何人类通过打字或说话可能达到的交流速度都要快。

Yeah, that BPS is an interesting metric to measure.

是的，BPS是一个有趣的衡量指标。

There might be a big leap in the experience once you reach a certain level of BPS.

一旦达到某个BPS水平，体验可能会有巨大的飞跃。

Yeah. Like entire new ways of interacting with a computer might be unlocked.

是的。就像与计算机交互的全新方式可能会被解锁。

And with humans. With other humans.

以及与人类。与其他人类。

Provided they have a neural link too.

前提是他们也有神经链接。

Right. Otherwise they won't be able to absorb the signals fast enough.

对。否则他们将无法足够快地吸收这些信号。

Do you think they'll improve the quality of intellectual discourse?

你认为它们会提高智力对话的质量吗？

Well, I think you could think of it, you know, if you were to slow down, communication, how do you feel about that?

嗯，我觉得你可以这样想，你知道，如果你放慢交流速度，你对此感觉如何？

You know, if you'd only talk at, let's say, one-tenth of normal speed, you'd be like, wow, that's agonizingly slow.

你知道，如果你只以，比如说，正常速度的十分之一说话，你会觉得，哇，那慢得让人难受。

So now, imagine you could speak at, communicate clearly at 10 or 100 or 1,000 times faster than normal.

所以现在，想象一下你能以比正常快10倍、100倍甚至1000倍的速度说话、清晰地交流。

Listen, I'm pretty sure nobody in their right mind listens to me at 1x.

听着，我很确定没有哪个心智正常的人会用1倍速听我说话。

They listen at 2x.

他们都用2倍速听。

I can only imagine what 10x would feel like, or I could actually understand it.

我只能想象10倍速会是什么感觉，或者我其实能理解它。

I usually default to 1.5x.

我通常默认用1.5倍速。

You can do 2x, but...

你可以用2倍速，但是……

Well, actually, if I'm listening to somebody in like sort of 15, 20 minutes, I want to go to sleep, then I'll do it 1.5x.

嗯，其实，如果我要听某人讲大概15、20分钟，我想睡觉了，那我就用1.5倍速。

If I'm paying attention, I'll do 2x.

如果我专心听，我就用2倍速。

Right.

对。

But actually, if you start... Actually listen to podcasts or sort of audiobooks or anything at, if you get used to doing it at 1.5, then 1 sounds painfully slow.

但其实，如果你开始……其实听播客或类似有声书之类的，如果你习惯了用1.5倍速，那1倍速听起来就慢得让人难受。

I'm still holding on to 1 because I'm afraid.

我仍然坚持用1倍速，因为我害怕。

I'm afraid of myself becoming bored with the reality, with the real world where everyone's speaking in 1x.

我害怕自己会对现实感到厌倦，对那个每个人都用1倍速说话的真实世界感到厌倦。

Well, it depends on the person.

嗯，这取决于人。

You can speak very fast.

你可以说得非常快。

We can communicate very quickly.

我们可以非常快地交流。

And also, if you use a wide range of, if your vocabulary is larger, your bit rate, effective bit rate is higher.

而且，如果你使用广泛的……如果你的词汇量更大，你的比特率，有效比特率就更高。

That's a good way to put it.

这个说法很好。

the effective bit rate.

有效比特率。

I mean, that is the question, is how much information is actually compressed in the low bit transfer of language.

我的意思是，问题就在于，语言在低比特传输中实际压缩了多少信息。

Yeah, if there's a single word that is able to convey something that would normally require, I don't know, 10 simple words,

是的，如果有一个单词能够传达通常需要，我不知道，10个简单单词才能表达的意思，

then you've got maybe a 10x compression on your hands.

那么你手上可能就有了10倍的压缩。

And that's really, like with memes, memes are like data compression.

这真的就像模因一样，模因就像数据压缩。

It conveys a whole, You're simultaneously hit with a wide range of symbols that you can interpret, and you kind of get it faster than if it were words or a simple picture.

它传达了一整套，你同时被大量可以解读的符号击中，你理解它的速度比用文字或简单图片更快。

And of course you're referring to memes broadly, like ideas.

当然，你指的是广义的模因，比如想法。

Yeah, there's an entire idea structure that is like an idea template.

是的，有一个完整的想法结构，就像一个想法模板。

And then you can add something to that idea template.

然后你可以往那个想法模板里添加一些东西。

But somebody has that pre-existing idea template their head. So when you add that incremental bit of information, you're conveying much more than if you just said a few words.

但某人脑子里已经有了那个预先存在的想法模板。所以当你添加那一点增量信息时，你传达的比只说几个词要多得多。

It's everything associated with that meme.

它是与那个模因相关的一切。

You think there'll be emergent leaps of capability as you scale the number of electrodes?

你认为随着电极数量的增加，会出现能力的涌现式飞跃吗？

Yeah. There'll be a certain, you think there'll be like actual number where it just, the human experience will be altered?

是的。会有一个特定的，你认为会有一个具体的数字，到了那里，人类的体验就会被改变？

Yes. What do you think that number might be?

是的。你认为那个数字可能是多少？

Whether electrodes or BPS? We of course don't know for sure, but is this 10,000, 100,000?

无论是电极还是BPS？我们当然不确定，但这是10,000，还是100,000？

Yeah, I mean, certainly if you're anywhere at 10,000 plus per second, I mean, that's vastly faster than any human can communicate right now.

是的，我的意思是，如果你达到每秒10,000以上，我的意思是，那比现在任何人类能交流的速度都要快得多。

If you think of the, what is the average plus per second of a human,

如果你想想，人类的平均每秒加号是多少，

it is less than one bit per second over the course of a day,

在一天的过程中，它每秒不到一个比特，

because there are 86,400 seconds in a day, and you don't communicate 86,400 tokens in a day.

因为一天有86,400秒，而你一天并不会交流86,400个词元。

Therefore, your plus per second is less than one, averaged over 24 hours.

因此，你的每秒加号在24小时平均下来不到一个。

It's quite slow.

这相当慢。

And even if you're communicating very quickly and you're talking to somebody who understands what you're saying,

即使你交流得非常快，而且你在和一个理解你在说什么的人说话，

because in order to communicate, you have to at least to some degree model the mind state of the person to whom you're speaking.

因为为了交流，你至少必须在某种程度上模拟与你说话之人的心理状态。

then take the concept you're trying to convey, compress that into a small number of syllables,

然后把你试图传达的概念，压缩成少量音节，

speak them, and hope that the other person decompresses them into a conceptual structure that is as close to what you have in your mind as possible.

说出它们，并希望另一个人把它们解压成一个概念结构，尽可能接近你脑中的想法。

Yeah, I mean, there's a lot of signal loss there in that process.

是的，我的意思是，那个过程中有很多信号损失。

Yeah, very lossy compression and decompression.

是的，非常有损的压缩和解压。

And a lot of what your neurons are doing is distilling the concepts down to a small number of symbols of say syllables

你的神经元所做的很多事情，就是把概念提炼成少量符号，比如音节，

that I'm speaking or keystrokes, whatever the case may be.

我正在说的，或者按键，无论是什么情况。

So that's a lot of what your brain computation is doing.

所以这就是你大脑计算所做的很多事情。

Now there is an argument that that's actually a healthy thing to do or a helpful thing to do

现在有一种观点认为，这实际上是一件健康的事，或者说是一件有益的事。

because as you try to compress complex concepts,

因为当你试图压缩复杂的概念时，

you're perhaps forced to distill what is most essential in those concepts as opposed to just all the fluff.

你也许会被迫提炼出那些概念中最本质的东西，而不是所有的废话。

So in the process of compression, you distill things down to what matters the most

所以在压缩的过程中，你会把事情提炼到最重要的部分，

because you can only say a few things.

因为你只能说很少的几件事。

So that is perhaps helpful. I think

所以这也许是有益的。我认为

we might, we'll probably get, if our data rate increases,

我们可能会，如果我们的数据速率提高，我们很可能会变得

it's highly probable that it will become far more verbose.

它极有可能会变得啰嗦得多。

Just like your computer, you know, when computers had like, my first computer had 8K of RAM, you know, so.

就像你的电脑一样，你知道，当电脑还只有，我的第一台电脑只有8K内存的时候，你知道，所以。

You really thought about every byte.

你真的会仔细考虑每一个字节。

And now you've got computers with many gigabytes of RAM.

而现在你的电脑有几十亿字节的内存。

So if you want to do an iPhone app that just says, hello world, it's probably, I don't know, several megabytes minimum with a bunch of fluff.

所以如果你想做一个只显示“你好，世界”的iPhone应用，它可能，我不知道，至少有好几兆字节，还带着一堆废话。

But nonetheless, we still prefer to have the computer with more memory and more compute.

但尽管如此，我们还是更愿意拥有内存更大、算力更强的电脑。

So the long-term aspiration of Neuralink is to improve the AI-human symbiosis

所以Neuralink的长期愿景是改善人工智能与人类的共生关系，

by increasing the bandwidth of the communication.

通过提高通信的带宽。

Because even in the most benign scenario of AI, you have to consider that

因为即使在人工智能最温和的情况下，你也必须考虑到，

the AI is simply going to get bored waiting for you to spit out a few words.

人工智能只会因为等你吐出几个字而感到无聊。

I mean, if the AI can communicate at terabits per second and you're communicating at bits per second,

我的意思是，如果AI能以每秒太比特的速度交流，而你以每秒比特的速度交流，

Yeah, it's like talking to a tree.

是的，就像对着一棵树说话。

Well, it is a very interesting question for a super intelligent species.

嗯，对于一个超级智能物种来说，这是一个非常有趣的问题。

What use are humans?

人类有什么用？

I think there is some argument for humans as a source of will.

我认为有理由认为人类是意志的来源。

Will. Will, yeah.

意志。意志，是的。

Source of will or purpose.

意志或目的的来源。

so if you consider the human mind as being essentially the

所以如果你认为人类思维本质上是

there's the primitive limbic elements which basically even like reptiles have and

有原始的边缘系统元素，基本上连爬行动物都有，还有

there's the cortex that's the thinking and planning part of the brain

有大脑皮层，那是大脑中思考和计划的部分

now the cortex is much smarter than limbic system and yet is largely in service to the limbic system

现在大脑皮层比边缘系统聪明得多，但很大程度上是为边缘系统服务的

it's trying to make the limbic system happy

它试图让边缘系统快乐

i mean the sheer amount of compute that's gone into people trying to get laid is insane.

我的意思是，人们为了上床而投入的巨量计算资源简直是疯狂的。

Without the, without actually seeking procreation.

没有，没有真正寻求生育。

They're just literally trying to do this sort of simple motion.

他们只是真的在尝试做这种简单的动作。

And they get a kick out of it.

而且他们从中获得快感。

So this simple, which in the abstract rather absurd motion, which is sex,

所以这个简单的，抽象地说相当荒谬的动作，也就是性，

the cortex is putting a massive amount of compute into trying to figure out how to do that.

大脑皮层投入大量计算资源试图弄清楚如何做到这一点。

So like 90% of distributed compute of the human species is spent on trying to get laid probably.

所以人类物种大约90%的分布式计算可能都花在了试图上床这件事上。

Like a large percentage.

很大一部分。

Yeah, yeah.

是的，是的。

There's no purpose to most sex except hedonistic.

大多数性行为除了享乐没有其他目的。

You know, it's just sort of a joy or whatever, dopamine release.

你知道，它只是一种快乐之类的，多巴胺释放。

Now once in a while it's procreation, but for humans it's mostly, modern humans it's mostly recreational.

现在偶尔是为了繁衍，但对人类来说，主要是，现代人类主要是为了娱乐。

And so your cortex, much smarter than your limbic system, trying to make the limbic system happy

所以你的皮层，比你的边缘系统聪明得多，试图让边缘系统开心

because limbic system wants to have sex so um or want some tasty food or whatever the case may be

因为边缘系统想要性爱，所以嗯，或者想要一些美味的食物，或者不管是什么情况

and then that doesn't further augmented by the tertiary system which is your phone your laptop ipad or whatever you know all your computing stuff

然后这并没有进一步被第三系统增强，也就是你的手机、笔记本电脑、iPad或者你知道的所有计算设备

that's your tertiary layer so you're actually already a cyborg

那是你的第三层，所以你实际上已经是一个半机械人了

you have this tertiary compute layer which is in the form of your your computer with all the applications all your compute devices um and uh

你有这个第三计算层，它是以你的电脑的形式存在的，还有所有的应用程序，你所有的计算设备，嗯，呃

And so in the getting laid front, there's actually a massive amount of digital compute also trying to get laid.

所以在性爱方面，实际上有大量的数字计算也在试图获得性爱。

You know, with like Tinder and whatever, you know. Yeah.

你知道，比如Tinder之类的，你知道。是的。

So the compute that we humans have built is also participating. Yeah.

所以我们人类建造的计算也在参与其中。是的。

I mean, there's like gigawatts of compute going into getting laid, of digital compute. Yeah.

我的意思是，有像千兆瓦的计算投入到获得性爱中，数字计算。是的。

What if AGI was... This is happening as we speak.

如果AGI是……这正在我们说话时发生。

if we merge with AI, it's just gonna expand the compute that we humans use.

如果我们与AI融合，它只会扩展我们人类使用的计算。

Pretty much. Well, it's one of the things, certainly, yeah. Yeah.

差不多。嗯，这是其中一件事，当然，是的。是的。

But what I'm saying is that, yes, is there a use for humans?

但我想说的是，是的，人类还有用吗？

Well, there's this fundamental question of what's the meaning of life, why do anything at all?

嗯，有一个根本问题，生命的意义是什么，为什么要做任何事情？

And so if our simple limbic system provides a source of will to do something

所以，如果我们简单的边缘系统提供了做某事的意志来源，

that then goes to our cortex, that then goes to our tertiary compute layer,

然后它传到我们的皮层，再传到我们的三级计算层，

then I don't know, it might actually be that the AI, in a benign scenario, is simply trying to make the human limbic system happy.

那么我不知道，也许实际上在良性情景下，AI只是在试图让人类的边缘系统感到快乐。

Yeah, it seems like the will is not just about the limbic system.

是的，似乎意志不仅仅关乎边缘系统。

There's a lot of interesting, complicated things in there.

那里有很多有趣、复杂的东西。

We also want power. That's limbic too, I think.

我们也想要权力。我想那也是边缘系统的。

But then we also want to, in a kind of cooperative way, alleviate the suffering in the world?

但然后我们也想以一种合作的方式减轻世界上的苦难？

Not everybody does, but yeah, sure. Some people do.

不是每个人都这样，但是的，当然。有些人会。

As a group of humans, when we get together, we start to have this kind of collective intelligence

作为人类群体，当我们聚在一起时，我们开始拥有这种集体智慧，

that is more complex in its will than the underlying individual descendants of apes.

其意志比作为基础的猿类个体后代更为复杂。

So there's other motivations, and that could be a really interesting source of an objective function for AGI.

所以还有其他动机，那可能成为AGI目标函数的一个非常有趣的来源。

Yeah, I mean, there are these sort of fairly cerebral or kind of higher level goals.

是的，我的意思是，有这些相当理性的或更高层次的目标。

I mean, for me, it's like what's the meaning of life for understanding the nature of the universe is of great interest to me.

我的意思是，对我来说，理解宇宙本质的生命意义对我非常感兴趣。

And hopefully to the AI.

也希望AI也如此。

And that's the mission of XAI and Grok is understand the universe.

而这就是XAI和Grok的使命：理解宇宙。

So do you think people, when you have a neural link with 10,000, 100,000 channels,

所以你认为，当你有1万、10万个通道的神经链接时，

most of the use cases will be communication with AI systems?

大多数用例将是与AI系统通信？

Well, assuming that they're solving basic neurological issues that people have,

嗯，假设它们正在解决人们的基本神经问题，

if they've got damaged neurons in their spinal cord or neck, as is the case with our first two patients,

如果他们的脊髓或颈部有受损的神经元，就像我们的前两位患者那样，

then obviously the first order of business is solving fundamental neuron damage in the spinal cord, neck, or in the brain itself.

那么显然首要任务是解决脊髓、颈部或大脑本身的基本神经元损伤。

So, you know, a second product is called Blindsight,

所以，你知道，第二个产品叫做Blindsight，

which is to enable people who are completely blind, lost both eyes or optic nerve, or just can't see at all,

这是为了让完全失明的人，失去双眼或视神经，或者完全看不见的人，

to be able to see by directly triggering the neurons in the visual cortex.

能够通过直接触发视觉皮层中的神经元来看见。

So we're just starting at the basics here.

所以我们只是从基础开始。

You know, this is like very, the simple stuff, relatively speaking, is solving neuron damage.

你知道，这就像是非常，相对而言，简单的事情，就是解决神经元损伤。

It can also solve, I think, probably schizophrenia.

它还可以解决，我认为，可能是精神分裂症。

If people have seizures of some kind, it can probably solve that.

如果人们有某种癫痫发作，它可能可以解决。

It could help with memory.

它可能有助于记忆。

So there's like a kind of a tech tree, if you will, of like you got the basics.

所以就像有一种科技树，如果你愿意的话，就像你有了基础。

Like you need literacy before you can have, you know, Lord of the Rings, you know.

就像你需要识字才能有，你知道，《指环王》，你知道。

Got it.

明白了。

Do you have letters and alphabet?

你有字母和字母表吗？

Okay, great.

好的，太好了。

Words, you know, then eventually you get sagas.

词语，你知道，然后最终你得到传奇。

So, you know, I think there may be some, you know, things to worry about in the future,

所以，你知道，我认为未来可能会有一些，你知道，需要担心的事情，

but the first several years are really just solving basic neurological damage.

但最初几年真的只是在解决基本的神经损伤。

But for people who have essentially complete or near complete loss from the brain to the body, like Stephen Hawking would be an example,

但对于那些基本上完全或接近完全丧失大脑到身体连接的人来说，比如斯蒂芬·霍金就是一个例子，

the neural links would be incredibly profound.

神经连接将会极其深刻。

Because I mean, you can imagine if Stephen Hawking could communicate as fast as we're communicating perhaps faster.

因为我的意思是，你可以想象，如果斯蒂芬·霍金能像我们交流一样快，或许更快。

And that's certainly possible. Probable, in fact. Likely, I'd say.

而这当然是可能的。事实上，很可能。我会说，很有可能。

So there's a kind of dual track of medical and non-medical, meaning so everything you've talked about could be applied to people who are non-disabled in the future.

所以有一种医疗和非医疗的双轨，意思是，你谈论的一切将来都可以应用于非残疾人士。

The logical thing to do is, sensible thing to do is to start off solving basic neuron damage issues.

合乎逻辑的做法是，明智的做法是，从解决基本的神经元损伤问题开始。

Yes. Because there's obviously some risk with a new device.

是的。因为新设备显然存在一些风险。

you can't get the risk down at zero. It's not possible.

你无法将风险降为零。这是不可能的。

So you want to have the highest possible reward given that there's a certain irreducible risk.

所以，鉴于存在一定的不可降低的风险，你想要获得尽可能高的回报。

And if somebody's able to have a profound improvement in their communication, that's worth the risk.

如果有人能够在交流方面取得重大改善，那风险是值得的。

As you get the risk down. Yeah, as you get the risk down.

随着你降低风险。是的，随着你降低风险。

Once the risk is down to, you know,

一旦风险降到，你知道，

if you have like thousands of people that have been using it for years and the risk is minimal,

如果有成千上万的人已经用了很多年，而且风险极小，

then perhaps at that point you could consider saying, okay, let's aim for augmentation.

那么也许到那时你可以考虑说，好吧，让我们以增强为目标。

Now, I think we're actually going to aim for augmentation with people who have neuron damage.

现在，我认为我们实际上要以神经元受损的人为对象进行增强。

So we're not just aiming to give people a communication data rate equivalent to normal humans.

所以我们不只是想给人们一个相当于正常人的通信数据速率。

We're aiming to give people who have, you know, quadriplegic or maybe have complete loss of the connection to the brain and body, a communication data rate that exceeds normal humans.

我们的目标是给那些，你知道，四肢瘫痪或者可能完全失去了大脑和身体连接的人，一个超过正常人的通信数据速率。

I mean, well, we're in there, why not? Let's give people superpowers.

我的意思是，好吧，我们都已经做到这一步了，为什么不呢？让我们给人们超能力吧。

And the same for vision.

视觉方面也是一样。

As you restore vision, there could be aspects of that restoration that are superhuman.

当你恢复视力时，这种恢复可能有一些方面是超人的。

Yeah, at first the vision restoration will be low res.

是的，起初视力恢复会是低分辨率的。

Because you have to say, like, how many neurons can you put in there and trigger?

因为你得说，比如，你能放多少神经元进去并触发？

And you can do things where you adjust the electric field to, like even if you've got, say, 10,000 neurons, it's not just 10,000 pixels,

而且你可以做一些事情，比如调整电场，即使你只有，比如说，一万个神经元，它也不只是一万个像素，

because you can adjust the field between the neurons and do them in patterns in order to get, say, 10,000 electrodes effectively give you, I don't know, maybe like having a megapixel or a 10 megapixel situation.

因为你可以调整神经元之间的电场，并以模式来操作它们，从而让，比如说，一万个电极有效地给你，我不知道，也许就像拥有一个百万像素或一千万像素的情况。

And then over time, I think you get to higher resolution than human eyes,

然后随着时间推移，我觉得你能获得比人眼更高的分辨率，

and you could also see in different wavelengths.

而且你还能看到不同的波长。

So like Geordi LaForge from Star Trek.

就像《星际迷航》里的乔迪·拉弗吉。

You know, like the thing.

你知道，就像那个东西。

You could just, do you want to see in radar? No problem.

你可以直接，你想用雷达看吗？没问题。

You could see ultraviolet, infrared, eagle vision, whatever you want.

你可以看到紫外线、红外线、鹰的视觉，随便你想要什么。

Do you think there'll be, let me ask a Joe Rogan question.

你觉得会有，让我问一个乔·罗根式的问题。

Do you think there'll be?

你觉得会有吗？

I just recently taken ayahuasca.

我最近刚服用了死藤水。

so this question no well yes well i guess technically it is yeah yeah try

所以这个问题，不，嗯，是的，嗯，我想严格来说是的，对，对，试试

it ever tried i love you joe okay

它曾经试过，我爱你，乔，好的

yeah wait yeah have you said much about it i have not

对，等等，对，你对此说过很多吗？我没有

i have not i've not i've been okay well we'll spill the beans uh

我没有，我没有，我一直，好吧，那我们就透露一下，呃

it was an it was a truly incredible i mean you're in the jungle

那是一次，那是一次真正不可思议的，我是说你在丛林里

yeah amongst the trees myself yeah it must have been crazy and a shaman yeah yeah yeah

对，在树林中，我自己，对，那一定很疯狂，还有一位萨满，对，对，对

with the insects with the animals all around you like jungle

周围有昆虫，有动物，就像丛林一样

as far as i can see i mean that's the way to do it things

就我目力所及，我是说，这才是做事的方式

are gonna look pretty wild yeah pretty wild

看起来会相当狂野，对，相当狂野

i took an extremely high dose just don't go hugging an anaconda or something

我服用了极高剂量，只是别去拥抱什么水蟒之类的

you know you haven't lived unless you made love to an anaconda

你知道，除非你和一条水蟒做爱，否则你就不算活过

i'm sorry but snakes and ladders um

对不起，但是蛇和梯子，嗯

yeah it was i took uh extremely high dose of okay uh nine

对，那是，我服用了，呃，极高剂量的，好的，呃，九

cups and uh damn okay that sounds like a lot like what's the noise is no one cup or one or two well usually one

杯子，呃，该死，好吧，那听起来很多，像是什么声音，没有人，一杯还是两杯，嗯，通常一杯

wait like right off the bat or do you work your way up to it

等等，像是一开始就那样，还是你慢慢来

so i uh across across two days because then the first day i took two and i okay it was a it was a ride but it wasn't quite like a... It wasn't like a revelation. It wasn't into deep space type ride. It was just like a little airplane ride.

所以我呃，跨两天，因为然后第一天我吃了两个，我，好吧，它是一次旅程，但不太像……它不像一个启示。它不是那种进入深空的旅程。它就像一次小飞机旅行。

I saw some trees and some visuals and all that. I just saw a dragon and all that kind of stuff.

我看到一些树和一些视觉效果之类的。我只是看到一条龙和那种东西。

But... Nine cups, you went to Pluto, I think. Pluto, yeah. No, deep space. Deep space.

但是……九杯，你去了冥王星，我想。冥王星，是的。不，深空。深空。

One of the interesting aspects of my experience is I thought I would have some demons, some stuff to work through.

我经历中有趣的一点是，我以为我会遇到一些心魔，一些需要克服的东西。

That's what people... That's what everyone says. I had nothing. I had all positive. I had just so full. Just a pure soul. I don't think so. I don't know.

那就是人们……那就是每个人说的。我什么都没有。我全是积极的。我只是那么充实。只是一个纯洁的灵魂。我不这么认为。我不知道。

But I kept thinking about it had an extremely high resolution thoughts about the people I know in my life. You were there. Okay.

但我一直在想，它对我生活中认识的人有极其高分辨率的想法。你在那里。好吧。

And it's just not from my relationship with that person, but just as the person themselves, I had just this deep gratitude of who they are. That's cool.

而这不仅仅来自我和那个人的关系，而是作为那个人本身，我只是对他们是谁有着深深的感激。那很酷。

It was just like this exploration. You know like Sims or whatever?

它就像这种探索。你知道，像《模拟人生》之类的？

You get to watch them.

你可以看着他们。

Sure. I got to watch people and just be in awe of how amazing they are.

当然。我得以观察人们，并惊叹于他们是多么了不起。

It sounds awesome. Yeah, it's great.

听起来太棒了。是啊，很棒。

I was waiting for- When's the demon coming?

我一直在等——恶魔什么时候来？

Exactly. Maybe I'll have some negative thoughts.

没错。也许我会有一些负面想法。

Nothing. Nothing. I had just extreme gratitude for them.

没有。没有。我只是对他们怀有极度的感激。

And then also a lot of space travel.

然后还有很多太空旅行。

Space travel to where?

太空旅行去哪里？

So here's what it was. It was people, the human beings that I know,

事情是这样的。是我认识的人，那些人类，

they had this kind of, the best way I can describe it is they had a glow to them.

他们有一种，我能描述的最好方式是，他们身上有一种光芒。

Okay. And then I kept flying out from them to see Earth, to see our solar system, to see our galaxy.

好的。然后我不断从他们那里飞出去，看地球，看我们的太阳系，看我们的银河系。

And I saw that light, that glow all across the universe.

我看到那道光，那光芒遍布整个宇宙。

Okay. Like whatever that form is.

好的。不管那是什么形态。

All right. Whatever that, like.

好吧。不管那是什么，就像。

Did you go past the Milky Way?

你越过银河系了吗？

Uh yeah yeah you're like intergalactic yeah intergalactic okay yeah

呃，是啊是啊，你就像星际的，是啊，星际的，好的，是啊

but always pointing in yeah um past the milky way past

但总是指向，是啊，嗯，越过银河系，越过

i mean i saw like a huge number of galaxies intergalactic

我是说，我看到像是有巨量的星系，星际的

and all of it was glowing so but i couldn't control that travel because i would actually

而所有这一切都在发光，但我无法控制那种旅行，因为我实际上会

explore near distances to the solar system see if there's aliens or any of that kind of stuff

探索太阳系附近的距离，看看有没有外星人或类似的东西

no i think i didn't know zero aliens implication of aliens because they were glowing they were glowing in the same way that humans were glowing,

不，我想我不知道，零外星人，外星人的含义，因为它们在发光，它们发光的方式和人类发光的方式一样，

that like life

那种像生命

force that I was seeing.

我所看到的力量。

The thing that made humans amazing was there throughout the universe.

让人类变得了不起的东西，在整个宇宙中都存在。

Like there was these glowing dots.

就像有这些发光的点。

So I don't know.

所以我不知道。

It made me feel like there is life, no, not life, but something, whatever makes humans amazing all throughout the universe.

它让我觉得有生命，不，不是生命，而是某种东西，无论是什么让人类在整个宇宙中变得了不起。

Sounds good.

听起来不错。

Yeah, it was amazing.

是的，太神奇了。

No demons, no demons.

没有恶魔，没有恶魔。

I looked for the demons, there's no demons.

我寻找恶魔，但没有恶魔。

There were dragons and they're pretty awesome.

有龙，它们非常棒。

So the thing about trees.

所以关于树的事情。

Was there anything scary at all?

有什么可怕的东西吗？

Dragons?

龙？

But they weren't scary.

但它们并不可怕。

They were front. They were protective.

它们在前方。它们在保护。

So the thing is.

所以事情是这样的。

Like Pasta Magic Dragon.

像意大利面魔法龙。

No, it was more like Game of Thrones kind of dragons.

不，更像是《权力的游戏》那种龙。

They weren't very friendly.

它们不太友好。

They were very big.

它们非常大。

So the thing is about giant trees at night, which is where I was.

所以事情是关于夜晚的巨树，那就是我当时所在的地方。

I mean, the jungle is kind of scary.

我的意思是，丛林有点可怕。

The trees started to look like dragons.

树木开始看起来像龙。

And they were all like looking at me.

它们都像在看着我。

Sure.

当然。

Okay. And it didn't seem scary.

好吧。而且它看起来并不可怕。

It seemed like they were protecting me.

看起来它们在保护我。

And the shaman and the people didn't speak any English, by the way, which made it even scarier.

而且萨满和那些人不会说英语，顺便说一句，这让事情更可怕了。

Because we're not even like, you know, we're worlds apart in many ways.

因为我们甚至不像，你知道，我们在很多方面天差地别。

It's just, but yeah,

只是，但是是的，

they talk about the mother of the forest protecting you, and that's what I felt like.

他们谈论森林之母保护你，那就是我的感觉。

And you're way out in the jungle. Way out.

而你在丛林深处。很远。

This is not like a tourist retreat.

这不像是一次游客度假。

You know, like 10 miles outside of a frio or something.

你知道，就像在某个地方外面10英里左右。

No, we went...

不，我们去了……

No, this is not a...

不，这不是一个……

You're a deep, deep Amazon...

你是一个深入、深入的亚马逊……

So me and this guy and Paul Rosalie, who basically is Tarzan, he lives in the jungle.

所以我和这个家伙，还有保罗·罗萨莉，他基本上就是泰山，他住在丛林里。

We went out deep and we just went crazy.

我们深入进去，然后就彻底疯狂了。

Wow, cool.

哇，酷。

Yeah.

是的。

So anyway, can I get that same experience in a Neuralink?

那么总之，我能在Neuralink中获得同样的体验吗？

Probably, yeah.

大概可以，是的。

I guess that is the question for non-disabled people.

我想这就是对非残障人士的问题。

Do you think that there's a lot in our perception, in our experience of the world that could be explored, that could be played with using Neuralink?

你认为在我们的感知中，在我们对世界的体验中，有很多可以用Neuralink来探索、来玩味的东西吗？

Yeah, I mean, Neuralink is really a generalized input-output device.

是的，我的意思是，Neuralink真的是一种通用的输入输出设备。

You know, it's reading electrical signals and generating electrical signals.

你知道，它在读取电信号并生成电信号。

And I mean, everything that you've ever experienced in your whole life, smell, emotions, all of those are electrical signals.

我的意思是，你一生中经历的一切，气味、情感，所有这些都是电信号。

So it's kind of weird to think that your entire life experience is distilled down to electrical signals from neurons, but that is in fact the case.

所以想到你整个生命体验被提炼成神经元的电信号，这有点奇怪，但事实确实如此。

I mean, that's at least what all the evidence points to.

我的意思是，至少所有证据都指向这一点。

So, I mean, you could trigger the right neuron.

所以，我的意思是，你可以触发正确的神经元。

You could trigger a particular scent.

你可以触发一种特定的气味。

You could certainly make things glow.

你当然可以让东西发光。

I mean, do pretty much anything.

我的意思是，几乎可以做任何事情。

I mean, really, you can think of the brain as a biological computer.

我的意思是，真的，你可以把大脑看作一台生物计算机。

So if there are certain, say, chips or elements of that biological computer that are broken,

所以，如果那个生物计算机的某些芯片或元件坏了，

let's say your ability to, if you've got a stroke, if you've had a stroke, that means some part of your brain is damaged.

比如说你的能力，如果你中风了，如果你中风过，那就意味着你大脑的某一部分受损了。

If that, let's say it's a speech generation or the ability to move your left hand, that's the kind of thing that Neuralink could solve.

如果那个，比如说它是语言生成能力，或者移动左手的能力，那就是Neuralink能解决的那类问题。

If you've got like a mass amount of memory loss that's just gone, well, we can't get the memories back.

如果你有大量记忆丧失，就那么消失了，那么，我们无法把那些记忆找回来。

We could restore your ability to make memories, but we can't restore memories that are fully gone.

我们可以恢复你形成记忆的能力，但我们无法恢复已经完全消失的记忆。

Now I should say, maybe if part of the memory is there and the means of accessing the memory is the part that's broken, then we could re-enable the ability to access the memory.

现在我应该说明一下，也许如果记忆的一部分还在，而访问记忆的方式才是坏掉的那部分，那么我们就能重新启用访问记忆的能力。

But you can think of it like RAM in a computer.

但你可以把它想象成电脑里的内存。

If the RAM is destroyed or your SD card is destroyed, we can't get that back.

如果内存被毁坏了，或者你的SD卡被毁坏了，我们就无法把它找回来。

But if the connection to the SD card is destroyed, we can fix that.

但如果与SD卡的连接被毁坏了，我们可以修复它。

If it is fixable physically, then it can be fixed.

如果它在物理上是可以修复的，那么它就能被修复。

Of course, with AI, just like you can repair photographs and fill in missing parts of photographs, maybe you can do the same.

当然，有了AI，就像你可以修复照片、填补照片中缺失的部分一样，也许你也能做同样的事。

Yeah, you could say, like, create the most probable set of memories based on all the information you have about that person.

是的，你可以说，比如，根据你掌握的关于那个人的所有信息，创建出最有可能的一组记忆。

You could then, it would be probabilistic restoration of memory.

然后你就可以，那将是对记忆的概率性恢复。

Now, we're getting pretty esoteric here.

现在，我们的话题变得相当深奥了。

But that is one of the most beautiful aspects of the human experience is remembering the good memories.

但人类经历中最美好的方面之一，就是记住那些美好的回忆。

Like we live most of our life, as Danny Kahneman has talked about, in our memories, not in the actual moment.

就像丹尼·卡尼曼所说的，我们大部分生活是活在记忆里，而不是在当下的时刻。

We're collecting memories and we kind of relive them in our head.

我们在收集记忆，然后在脑海中重温它们。

And that's the good times.

那就是美好的时光。

If you just integrate over our entire life, it's remembering the good times.

如果你对我们整个一生进行积分，那就是在记住美好的时光。

Sure. That produces the largest amount of happiness.

当然。那会产生最大量的幸福感。

Yeah, well, I mean, what are we but our memories?

是啊，我的意思是，我们除了记忆还能是什么？

And what is death but the loss of memory?

而死亡除了记忆的丧失还能是什么？

Yeah.

是的。

Loss of information.

信息的丧失。

You know, if you could say like, well, if you could be, you run a thought experiment, if you were disintegrated painlessly and then reintegrated a moment later, like teleportation, I guess,

你知道，如果你能说，比如，如果你能做一个思想实验，如果你被无痛地分解，然后片刻之后重新组合，就像传送一样，我猜，

provided there's no information loss, the fact that your one body was disintegrated is irrelevant.

只要没有信息丢失，你原来的身体被分解这一事实就无关紧要。

And memories is just such a huge part of that.

而记忆正是其中极其重要的一部分。

Death is fundamentally the loss of information.

死亡从根本上说就是信息的丧失。

The loss of memory.

记忆的丧失。

So if we can store them as accurately as possible, we basically achieve a kind of immortality.

所以如果我们能尽可能准确地存储它们，我们基本上就实现了一种永生。

Yeah.

是的。

You've talked about the threats, the safety concerns of AI.

你谈到过人工智能的威胁和安全问题。

Let's look at long-term visions.

让我们来看看长期愿景。

Do you think Neuralink is, in your view, the best current approach we have for AI safety?

在你看来，你认为Neuralink是我们目前拥有人工智能安全的最佳方法吗？

It's an idea that may help with AI safety.

这是一个可能有助于人工智能安全的想法。

Certainly not.

当然不是。

I wouldn't want to claim it's like some panacea or that's a sure thing.

我不想声称它是什么灵丹妙药，或者那是万无一失的。

But, I mean, many years ago I was thinking like, well, what would inhibit alignment of human,

但是，我的意思是，很多年前我在想，嗯，什么会阻碍人类的对齐，

collective human will with artificial intelligence?

集体人类意志与人工智能的对齐？

And the low data rate of humans, especially our slow output rate, would necessarily just, because it's such a, because the communication is so slow, would diminish the link between humans and computers.

而人类低下的数据速率，尤其是我们缓慢的输出速率，必然会，因为它是如此，因为交流如此缓慢，会削弱人类与计算机之间的联系。

Like the more you are a tree, the less you know what the tree is.

就像你越是一棵树，你就越不知道树是什么。

Let's say you look at this plant or whatever, and like, hey, I'd really like to make that plant happy, but it's not saying a lot, you know?

假设你看着这株植物什么的，然后想，嘿，我真的很想让那株植物开心，但它并没有说很多，你知道吗？

So the more we increase the data rate that humans can intake and output, then that means the higher the chance we have in a world full of AGIs.

所以，我们越提高人类能够输入和输出的数据速率，那就意味着在一个充满AGI的世界里，我们的机会就越大。

Yeah. We could better align collective human will with AI output rate especially was dramatically increased.

是的。我们可以更好地将集体人类意志与AI输出速率对齐，尤其是当输出速率大幅提高时。

Like, and I think there's potential to increase the output rate by, I don't know, three, maybe six, maybe more orders of magnitude.

就像，我认为有潜力将输出速率提高，我不知道，三个，也许六个，也许更多个数量级。

So it's better than the current situation.

所以这比目前的情况要好。

And that output rate would be by increasing the number of electrodes, number of channels, and also maybe implanting multiple neural links.

而那个输出速率将通过增加电极数量、通道数量，也许还有植入多个神经链接来实现。

Yeah. Do you think there'll be a world in the next couple of decades where it's hundreds of millions of people have neural links?

是的。你认为在未来几十年里，会出现一个数亿人拥有神经链接的世界吗？

Yeah, I do.

是的，我认为会。

Do you think when people just, when they see the capabilities, the superhuman capabilities that are possible, and then the safety is demonstrated?

你觉得当人们看到这些能力，这些可能的超人能力，然后安全性得到验证时，会怎么想？

Yeah, if it's extremely safe, and you can have superhuman abilities,

是的，如果它极其安全，而且你能拥有超人能力，

And let's say you can upload your memories, so you wouldn't lose memories,

假设你可以上传你的记忆，这样你就不会失去记忆，

then I think probably a lot of people would choose to have it.

那我觉得可能很多人会选择拥有它。

It would supersede the cell phone, for example.

比如说，它会取代手机。

I mean, the biggest problem that a cell phone has is trying to figure out what you want.

我的意思是，手机最大的问题就是试图弄清楚你想要什么。

So that's why you've got auto-complete and you've got output, which is all the pixels on the screen.

所以这就是为什么你有自动补全，还有输出，也就是屏幕上的所有像素。

But from the perspective of the human, the output is so friggin' slow.

但从人类的角度来看，输出慢得要命。

Desktop or phone is desperately just trying to understand what you want.

台式机或手机只是在拼命试图理解你想要什么。

And there's an eternity between every keystroke from a computer standpoint.

从计算机的角度来看，每次按键之间都漫长得像永恒。

Yeah. The computer's talking to a tree, a slow-moving tree that's trying to swipe.

是的。计算机在跟一棵树说话，一棵缓慢移动、试图滑动的树。

Yeah. Yeah. So if you have computers that are doing trillions of instructions per second,

是的。是的。所以如果你的计算机每秒执行数万亿条指令，

and a whole second went by, I mean, there's a trillion things it could have done, you know?

而整整一秒过去了，我的意思是，它本可以完成一万亿件事，你知道吗？

Yeah. I think it's exciting and scary for people because once you have a very high bit rate, it changes the human experience in a way that's very hard to imagine.

是的。我觉得这对人们来说既令人兴奋又可怕，因为一旦你拥有非常高的比特率，它会以很难想象的方式改变人类的体验。

Yeah. It would be something different.

是的。那会是某种不同的东西。

I mean, some sort of futuristic cyborg.

我的意思是，某种未来主义的半机械人。

I mean, we're obviously talking about, by the way, like, it's not like around the corner.

我的意思是，顺便说一下，我们显然在谈论的，并不是近在眼前的事。

It's, you asked me what the distant future is.

这是，你问过我遥远的未来是什么。

Like, maybe this is like, it's not super far away, but 10, 15 years, that kind of thing.

就像，也许这并不算特别遥远，但10年、15年，那种程度。

When can I get one?

我什么时候能买到一个？

10 years?

10年？

Probably less than 10 years.

可能不到10年。

Depends what you want to do, you know?

取决于你想做什么，你知道吗？

Hey, if I can get like 1,000 BPS.

嘿，如果我能达到比如1000 BPS。

1,000 BPS.

1000 BPS。

And it's safe, and I can just interact with the computer while laying back and eating Cheetos.

而且它很安全，我可以一边躺着吃奇多，一边和电脑互动。

I don't eat Cheetos.

我不吃奇多。

There's certain aspects of human-computer interaction when done more efficiently and more enjoyably are worth it.

人机交互的某些方面，如果做得更高效、更愉快，是值得的。

Well, we feel pretty confident that, I think.

嗯，我们相当有信心，我认为。

maybe within the next year or two, that someone with a Neuralink implant will be able to outperform a pro gamer.

也许在未来一两年内，植入Neuralink的人将能够超越职业游戏玩家。

Nice.

不错。

Because the reaction time would be faster.

因为反应时间会更快。

I got to visit Memphis.

我去过孟菲斯。

Yeah, yeah.

是的，是的。

You're going big on compute.

你在算力上投入很大。

Yeah.

是的。

And you've also said play to win or don't play at all.

你还说过，要么玩就为了赢，要么干脆别玩。

Yeah.

是的。

So what does it take to win?

那么，要赢需要什么？

For AI, that means you've got to have the most powerful training compute.

对于AI来说，这意味着你必须拥有最强大的训练算力。

And the rate of improvement of training compute has to be faster than everyone else, or you will not win.

而且训练算力的提升速度必须比其他人更快，否则你就赢不了。

Your AI will be worse.

你的AI会更差。

So how can Grok, let's say 3, that might be available, what, like next year?

那么Grok，比如说3，可能可用，什么时候，像明年？

Well, hopefully end of this year. Grok 3. If we're lucky, yeah.

嗯，希望今年年底。Grok 3。如果我们幸运的话，是的。

How can that be the best LLM, the best AI system available in the world.

那怎么能成为最好的LLM，世界上最好的AI系统。

How much of it is compute?

其中多少是计算？

How much of it is data?

其中多少是数据？

How much of it is post-training?

其中多少是后训练？

How much of it is the product that you package it up in?

其中多少是你打包的产品？

All that kind of stuff.

所有这些东西。

I mean, they all matter.

我的意思是，它们都很重要。

It's sort of like saying, let's say it's a Formula One race, what matters more are the car or the driver?

这有点像说，假设这是一级方程式赛车，更重要的是车还是车手？

I mean, they both matter.

我的意思是，两者都很重要。

If a car is not fast, then if it's half the horsepower of a competitor, the best driver will still lose.

如果车不快，那么如果它的马力只有竞争对手的一半，最好的车手仍然会输。

If it's twice the horsepower, then probably even a mediocre driver will still win.

如果马力是两倍，那么可能即使一个平庸的车手也会赢。

So the training computer is kind of like the engine, the horsepower of the engine.

所以训练计算机有点像引擎，引擎的马力。

So you really, you wanna try to do the best on that and then how efficiently do you use that to train and compute

所以你真正想要在那方面做到最好，然后你如何高效地利用它来训练和计算

and how efficiently do you do the inference, the use of the AI.

以及你如何高效地进行推理，即AI的使用。

So obviously that comes down to human talent.

所以显然这归结于人才。

And then what unique access to data do you have?

然后你有什么独特的数据访问权限？

That also plays a role.

这也起作用。

You think Twitter data will be useful?

你认为Twitter数据会有用吗？

Yeah, I mean, I think most of the leading AI companies already have already scraped all the Twitter data.

是的，我的意思是，我认为大多数领先的AI公司已经抓取了所有的Twitter数据。

Not I think they have.

不，我认为他们已经抓取了。

So on a go forward basis, what's useful is the fact that it's up to the second.

所以从今往后，有用的是它是最新的。

You know, that's because it's hard for them to scrape in real time.

你知道，那是因为他们很难实时抓取。

So there's an immediacy advantage that Grog has already.

所以Grog已经有一个即时性的优势。

I think with Tesla and the real time video coming from several million cars, ultimately tens of millions of cars with Optimus.

我认为有了特斯拉和来自几百万辆汽车的实时视频，最终是几千万辆汽车加上Optimus。

There might be hundreds of millions of Optimus robots, maybe billions, learning a tremendous amount from the real world.

可能会有数亿个Optimus机器人，也许数十亿个，从现实世界学到大量东西。

That's the biggest source of data, I think, ultimately, is Optimus.

我认为最终最大的数据来源是Optimus。

Optimus is going to be the biggest source of data.

Optimus将成为最大的数据来源。

Because it's- Because reality scales.

因为它——因为现实会扩展。

Reality scales to the scale of reality.

现实会扩展到现实的规模。

It's actually humbling to see how little data humans have actually been able to accumulate

看到人类实际能够积累的数据如此之少，真的令人谦卑。

um really if you say how many trillions of usable tokens have humans generated

嗯，真的，如果你说人类生成了多少万亿个可用的token

where on a non-duplicative like discounting spam and repetitive stuff it's not a huge number you run out pretty quickly

在非重复的、比如扣除垃圾邮件和重复内容的情况下，这个数字并不大，你很快就会用完

and optimists can go so tesla the cars can, unfortunately, have to stay on the road.

而乐观主义者可以继续，所以特斯拉的汽车，不幸的是，必须留在路上。

Optimus Robot can go anywhere.

Optimus机器人可以去任何地方。

There's more reality off the road. And go off-road.

路外有更多的现实。而且可以越野。

I mean, Optimus Robot can pick up the cup and see, did it pick up the cup in the right way?

我是说，Optimus机器人能拿起杯子，然后看看，它有没有用正确的方式拿起杯子？

Did it, you know, say, pour water in the cup?

它有没有，你知道，比如说，把水倒进杯子里？

You know, did the water go in the cup or not go in the cup?

你知道，水是进了杯子还是没进杯子？

Did it spill water or not?

它有没有把水洒出来？

Yeah. Simple stuff like that. I mean, but it can do, at that scale, times a billion.

是的。就是这种简单的事情。我是说，但它能做的，在那个规模上，乘以十亿。

you know, so generate useful data from reality.

你知道，所以从现实中生成有用的数据。

So cause and effect stuff.

所以就是因果关系这类东西。

What do you think it takes to get to mass production of humanoid robots like that?

你觉得要实现像那样的人形机器人量产，需要什么？

It's the same as cars, really. I mean, global capacity for vehicles is about 100 million a year.

这其实和汽车一样。我是说，全球汽车产能大约是每年一亿辆。

And it could be higher, it's just that the demand is on the order of 100 million a year.

而且可能更高，只是需求大约在每年一亿辆这个量级。

And then there's roughly 2 billion vehicles that are in use in some way.

然后大约有二十亿辆汽车以某种方式在使用中。

So, which makes sense. Like the life of a vehicle is about 20 years.

所以，这是合理的。就像一辆车的寿命大约是20年。

So at steady state, you can have 100 million vehicles produced a year with a 2 billion vehicle fleet, roughly.

所以在稳定状态下，大约每年生产一亿辆车，对应二十亿辆的汽车保有量。

Now for humanoid robots, the utility is much greater.

而对于人形机器人来说，实用性要大得多。

So my guess is humanoid robots are more like at a billion plus per year.

所以我猜人形机器人更可能是每年十亿台以上。

But, you know, until you came along and started building Optimus, it was thought to be an extremely difficult problem.

但是，你知道，在你出现并开始打造Optimus之前，这曾被认为是一个极其困难的问题。

I mean, still it's an extremely difficult problem.

我的意思是，这仍然是一个极其困难的问题。

Yes, it's no walk in the park.

是的，这可不是件轻松的事。

I mean, Optimus currently would struggle to walk in the park.

我的意思是，Optimus目前连在公园里走路都会很吃力。

I mean, it can walk in the park.

我的意思是，它能在公园里走路。

The park is not too difficult, but it will be able to walk over a wide range of terrain.

公园并不太难，但它将能够在各种地形上行走。

Yeah, and pick up objects.

是的，还能捡起物体。

Yeah, yeah.

是的，是的。

It can already do that.

它已经能做到这一点了。

But like all kinds of objects.

但就像各种物体。

Yeah, yeah. All foreign objects.

是的，是的。所有外来物体。

I mean, pouring water in a cup is not trivial.

我的意思是，往杯子里倒水可不是件简单的事。

Because then if you don't know anything about the container, it could be all kinds of containers.

因为如果你对容器一无所知，它可能是各种各样的容器。

Yeah.

是的。

There's going to be an immense amount of engineering just going into the hand.

光是手部就需要投入大量的工程。

Yeah.

是的。

The hand might be close to half of all the engineering in Optimus.

手部可能接近Optimus所有工程的一半。

From an electromechanical standpoint, the hand is probably roughly half of the engineering.

从机电角度来看，手部可能大约占工程的一半。

But so much of the intelligence, so much of the intelligence of humans goes into what we do with our hands.

但人类的许多智慧，许多智慧都体现在我们用手做的事情上。

Yeah.

是的。

It's the manipulation of the world, manipulation of objects in the world.

这是对世界的操控，对世界物体的操控。

Intelligence, safe manipulation of objects in the world, yeah.

智慧，安全地操控世界上的物体，是的。

Yeah.

是的。

I mean, you start really thinking about your hand and how it works.

我的意思是，你开始真正思考你的手以及它是如何工作的。

I do it all the time.

我一直都这么做。

The sensory control of homunculus is where you have humongous hands.

感觉控制的小人图就是你有巨大的手的地方。

Yeah.

是的。

So, I mean, like your hands, the actuators, the muscles of your hand are almost overwhelmingly in your forearm.

所以，我的意思是，就像你的手，那些执行器，你手部的肌肉几乎绝大多数都在你的前臂里。

So your forearm has the muscles that actually control your hand.

所以你的前臂有真正控制你手部的肌肉。

There's a few small muscles in the hand itself, but your hand is really like a skeleton meat puppet and with cables.

手本身有一些小肌肉，但你的手真的就像一个骨架肉傀儡，靠缆绳操控。

So the muscles that control your fingers are in your forearm and they go through the carpal tunnel,

所以控制你手指的肌肉在你的前臂里，它们穿过腕管，

which is that you've got a little collection of bones and a tiny tunnel that these cables, the tendons go through,

也就是你有一小堆骨头和一个微小的隧道，这些缆绳，也就是肌腱穿过其中，

and those tendons are what, mostly what moves your hands.

而那些肌腱主要是驱动你手部运动的东西。

And something like those tendons has to be re-engineered into the Optimus in order to do all that kind of stuff.

而类似那些肌腱的东西必须被重新设计到Optimus中，才能完成所有这些事情。

Yeah, so like the current Optimus, we tried putting the actuators in the hand itself, then you sort of end up having these like- Giant hands?

是的，所以像现在的Optimus，我们试过把执行器放在手本身里，然后你最终就会得到这些像——巨大的手？

Yeah, giant hands that look weird. And then they don't actually have enough degrees of freedom and or enough strength.

是的，巨大的手看起来很奇怪。而且它们实际上没有足够的自由度，或者足够的力量。

So then you realize, oh, okay, that's why you've got to put the actuators in the forearm.

然后你意识到，哦，好吧，这就是为什么你必须把执行器放在前臂里。

And just like a human, you've got to run cables through a narrow tunnel to operate the fingers.

就像人类一样，你必须让缆绳穿过狭窄的隧道来操作手指。

And there's also a reason for not having all the fingers the same length.

而且不让所有手指长度相同也是有原因的。

So it wouldn't be expensive from an energy or evolutionary standpoint

所以从能量或进化的角度来看，这并不昂贵

to have all your fingers be the same length.

让你所有的手指都一样长。

So why not do the same length?

那为什么不都做成一样的长度呢？

Yeah, why not?

是啊，为什么不呢？

Because it's actually better to have different lengths.

因为实际上有不同的长度会更好。

Your dexterity is better if you've got fingers of different lengths.

如果你手指长度不同，你的灵活性会更好。

There are more things you can do,

你能做的事情更多，

and your dexterity is actually better if your fingers are of different length.

而且如果你手指长度不同，你的灵活性实际上会更好。

In fact, there's a reason we've got a little finger.

事实上，我们长小指是有原因的。

Why not have a little finger this bigger?

为什么不让小指长得更大一些呢？

Yeah.

是啊。

Because it helps you with fine motor skills.

因为它能帮助你完成精细动作技能。

This little finger helps?

这根小指有帮助吗？

It does.

确实有。

Hmm.

嗯。

If you lost your little finger, you'd have noticeably less dexterity.

如果你失去了小指，你的灵活性会明显降低。

So as you're figuring out this problem,

所以当你在解决这个问题时，

you have to also figure out a way to do it so you can mass manufacture it.

你还必须找到一种方法来实现它，以便能够大规模生产。

to be as simple as possible?

尽可能简单吗？

It's actually going to be quite complicated.

实际上这会相当复杂。

The as-possible part is, it's quite a high bar.

“尽可能”这部分，标准相当高。

If you want to have a humanoid robot that can do things that a human can do, it's actually a very high bar.

如果你想要一个能做人能做的事情的人形机器人，那实际上是一个很高的标准。

So our new arm has 22 degrees of freedom instead of 11

所以我们的新手臂有22个自由度，而不是11个

and has, like I said, the actuators in the forearm.

而且就像我说的，执行器在前臂里。

And all the actuators are designed from scratch, the physics first principles.

所有执行器都是从零开始设计的，基于物理第一性原理。

The sensors are all designed from scratch.

传感器也全都是从零开始设计的。

And we'll continue to put a tremendous amount of engineering effort into improving the hand.

我们将继续投入大量的工程努力来改进这只手。

By hand, I mean the entire forearm from elbow forward is really the hand.

说到手，我指的是从肘部往前的整个前臂，其实都算是手。

So that's incredibly difficult engineering, actually.

所以这其实是极其困难的工程。

And so the simplest possible version of a human or a robot that can do even most, perhaps not all, of what a human can do is actually still very complicated.

所以，一个人类或机器人哪怕只能完成人类能做的绝大部分、也许不是全部的事情，其最简单的可能版本实际上仍然非常复杂。

It's not simple. It's very difficult.

这并不简单。这非常困难。

Can you just speak to what it takes for a great engineering team for you?

你能谈谈对你来说，一个优秀的工程团队需要具备什么吗？

What I saw in Memphis, the supercomputer cluster, is just this intense drive towards simplifying the process, understanding the process, constantly improving it, constantly iterating it.

我在孟菲斯看到的那个超级计算机集群，就是这种强烈的驱动力，去简化流程、理解流程、不断改进它、不断迭代它。

Well, it's easy to say simplify, and it's very difficult to do it.

嗯，说简化很容易，但做起来非常困难。

You know, I have this very basic first principles algorithm that I run kind of as like a mantra, which is to first question the requirements, make the requirements less dumb.

你知道，我有一个非常基本的第一性原理算法，我把它当作一种咒语来运行，那就是首先质疑需求，让需求不那么愚蠢。

The requirements are always dumb to some degree.

需求在某种程度上总是愚蠢的。

So if you want to start off by reducing the number of requirements, and no matter how smart the person who gave you those requirements, they're still dumb to some degree.

所以如果你想从减少需求数量开始，那么无论给你那些需求的人有多聪明，它们在某种程度上仍然是愚蠢的。

You have to start there because otherwise you could get the perfect answer to the wrong question.

你必须从那里开始，因为否则你可能会为错误的问题找到完美的答案。

So try to make the question the least wrong possible.

所以尽量让问题尽可能少错。

That's what question the requirements means.

这就是质疑需求的意思。

And then the second thing is try to delete whatever the step is, the part or the process step.

然后第二件事是尝试删除任何步骤，无论是部件还是流程步骤。

Sounds very obvious, but people often forget to try deleting it entirely.

听起来很明显，但人们常常忘记尝试把它完全删除。

And if you're not forced to put back at least 10% of what you delete, you're not deleting enough.

如果你没有被要求至少放回你删除内容的10%，那说明你删得还不够。

And it's somewhat illogically, people often, most of the time, feel as though they've succeeded if they've not been forced to put things back in.

而且有点不合逻辑的是，人们常常、大多数时候会觉得，只要没有被要求把东西放回去，他们就成功了。

But actually they haven't because they've been overly conservative and have left things in there that shouldn't be.

但实际上他们并没有成功，因为他们过于保守，把不该留在里面的东西留了下来。

And only the third thing is try to optimize it or simplify it.

而第三件事才是尝试优化或简化它。

Again, these all sound, I think, very obvious when I say them, but the number of times I've made these mistakes is more than I care to remember.

再说一次，我觉得这些当我说出来时听起来都很明显，但我犯这些错误的次数多到我不愿去回想。

That's why I have this mantra.

这就是为什么我有这句口头禅。

So, in fact, I'd say the most common mistake of smart engineers is to optimize a thing that should not exist.

所以，事实上，我想说聪明工程师最常见的错误就是去优化一个本不该存在的东西。

Right.

对。

So, like you say, you run through the algorithm and basically show up to a problem, show up to the supercomputer cluster and see the process and ask, can this be deleted?

所以，就像你说的，你运行算法，基本上就是面对一个问题，来到超级计算机集群前，查看进程并问：这个可以删除吗？

Yeah, first try to delete it.

是的，先试着删除它。

Yeah.

是的。

Yeah, that's not easy to do.

是的，这并不容易做到。

No, and actually, what generally makes people uneasy is that you've got to delete at least some of the things that you delete, you will put back in.

不，而且实际上，通常让人不安的是，你至少得删除一些你删掉的东西，你又会把它们放回去。

Yeah.

是的。

But going back to sort of where our limbic system can steer us wrong is that we tend to remember with sometimes a jarring level of pain

但回到我们边缘系统可能把我们带偏的地方，就是我们往往会记住那些有时带着刺骨痛感的经历

where we deleted something that we subsequently needed.

那时我们删掉了后来却需要的东西。

And so people remember that one time they forgot to put in this thing three years ago, and that caused them trouble.

所以人们会记得三年前他们忘了把某个东西放进去的那一次，而那给他们带来了麻烦。

And so they overcorrect, and then they put too much stuff in there and overcomplicate things.

于是他们矫枉过正，然后往里面塞太多东西，把事情搞得太复杂。

So you actually have to say, no, we're deliberately going to delete more than we should.

所以你必须说，不，我们要故意删掉比应该删的更多。

So that we're putting at least one in 10 things we're going to add back in.

这样我们至少每十样东西里会有一件要加回来。

And I've seen you suggest just that, that something should be deleted, and you can kind of see the pain.

我见过你建议的正是这一点，就是某个东西应该被删掉，而你能隐约看到那种痛苦。

Oh, yeah, absolutely. Everybody feels a little bit of the pain. Absolutely.

哦，是的，绝对是这样。每个人都会感受到一点那种痛苦。绝对是这样。

And I tell them in advance, like, yeah, there's some of the things that we delete, we're going to put back in.

我会提前告诉他们，就像，是的，我们删掉的一些东西，我们会再加回来。

And people get a little shook by that.

而人们会因此有点受惊。

But it makes sense because if you're so conservative as to never have to put anything back in, you obviously have a lot of stuff that isn't needed.

但这很合理，因为如果你保守到从不需要把任何东西加回来，那你显然有很多不需要的东西。

So you got to overcorrect.

所以你必须矫枉过正。

This is, I would say, like a cortical override to Olympic instinct.

我会说，这就像是对奥林匹克本能的一种皮层覆盖。

One of many that probably leads us astray.

可能是众多让我们误入歧途的本能之一。

Yeah.

是的。

There's like a step four as well, which is any given thing can be sped up.

还有类似第四步，就是任何给定的东西都可以被加速。

However fast you think it can be done,

无论你觉得它能做得多快，

like whatever the speed is being done, it can be done faster.

不管现在做的速度有多快，它都能做得更快。

But you shouldn't speed things up until you've tried to delete it and optimize it.

但你不应该加速，直到你尝试过删除它并优化它。

Otherwise, you're speeding up something that shouldn't exist as absurd.

否则，你就是在加速一个本不该存在的东西，这很荒谬。

And then the fifth thing is to automate it.

然后第五件事就是把它自动化。

And I've gone backwards so many times where I've automated something, sped it up, simplified it, and then deleted it.

我有很多次都走了回头路，我自动化了某个东西，加速了它，简化了它，然后又把它删掉了。

And I got tired of doing that.

我厌倦了那样做。

So that's why I've got this mantra

所以这就是为什么我有这句口头禅

That is a very effective five-step process.

那是一个非常有效的五步流程。

It works great.

它效果很好。

When you've already automated, deleting must be real painful.

当你已经自动化了，删除一定非常痛苦。

Yeah.

是的。

Yeah, it's great.

是的，这很棒。

It's like, wow, I really wasted a lot of effort there.

就像，哇，我在那里真的浪费了很多精力。

Yeah.

是的。

I mean, what you've done with the cluster in Memphis is incredible, just in a handful of weeks.

我的意思是，你在孟菲斯集群上所做的，仅仅几周内就令人难以置信。

Yeah, it's not working yet, so I don't want to pop the champagne corks.

是的，它还没成功，所以我不想开香槟庆祝。

In fact, I have a call in a few hours with the Memphis team because we're having some power fluctuation issues.

事实上，我几小时后要和孟菲斯团队通话，因为我们遇到了一些电力波动问题。

Yes.

是的。

Yeah, it's like kind of a, when you do synchronized training, when you have all these computers that are training, where the training is synchronized to the sort of millisecond level, it's like having an orchestra.

是的，这有点像，当你做同步训练时，当你有所有这些正在训练的计算机，训练同步到毫秒级别，就像有一个管弦乐队。

And then the orchestra can go loud to silent very quickly, you know, sub-second level.

然后管弦乐队可以非常快地从响亮到静音，你知道，亚秒级。

And then the electrical system kind of freaks out about that.

然后电力系统就会对此有点抓狂。

Like if you suddenly see giant shifts 10, 20 megawatts several times a second,

就像如果你突然看到巨大的波动，每秒好几次10、20兆瓦，

this is not what electrical systems are expecting to see.

这不是电力系统预期会看到的情况。

So that's one of the main things you have to figure out.

所以这是你必须弄清楚的主要事情之一。

The cooling, the power, and then on the software as you go up the stack,

冷却、电力，然后是软件，随着你在技术栈上往上走，

how to do the distributed compute, all of that.

如何做分布式计算，所有这些。

Today's problem is dealing with extreme power jitter.

今天的问题是应对极端的电力抖动。

Power jitter. Yeah.

电力抖动。是的。

It's a nice ring to that.

这听起来挺顺耳的。

So that's, okay.

所以那就是，好吧。

And you stayed up late into the night, as you often do there.

而你在那里熬夜到深夜，就像你经常做的那样。

Last week, yeah. Last week, yeah.

上周，是的。上周，是的。

Yeah, we finally got training going at, oddly enough, roughly 4.20 a.m. last Monday.

是的，我们终于开始训练了，说来奇怪，大约是在上周一凌晨4点20分。

Total coincidence.

纯属巧合。

Yeah, I mean, maybe it was 4.22 or something.

是的，我是说，也许是4点22分之类的。

Yeah, yeah, yeah.

是的，是的，是的。

It's that universe again with the jokes.

又是那个爱开玩笑的宇宙。

Exactly, just love it.

没错，就是喜欢这样。

I mean, I wonder if you could speak to the fact that you,

我是说，我想知道你能不能谈谈这个事实，就是你，

one of the things that you did when I was there is you went through all the steps of what everybody's doing just to get a sense

我在那里时你做的一件事，就是你把每个人在做的所有步骤都走了一遍，只是为了了解

that you yourself understand it and everybody understands it so they can understand when something is dumb or something is inefficient or that kind of stuff.

你自己理解它，而且每个人都理解它，这样当某件事很蠢或效率低下或诸如此类的时候，他们就能明白。

Can you speak to that?

你能谈谈这个吗？

Yeah.

是的。

So I try to do whatever the people at the front lines are doing,

所以我尽量去做一线人员正在做的事情，

I try to do it at least a few times myself.

我尽量自己至少做几次。

So connecting fiber optic cables, diagnosing a faulty connection, that tends to be the limiting factor for large training clusters is the cabling.

所以连接光纤电缆、诊断故障连接，这往往是大型训练集群的瓶颈所在，就是布线。

So many cables.

好多电缆啊。

Because for a coherent training system where you've got RDMA, so remote direct memory access, the whole thing is like one giant brain.

因为对于一个连贯的训练系统，你有了RDMA，也就是远程直接内存访问，整个东西就像一个巨大的大脑。

So if you've got any to any connection,

所以如果你有任意到任意的连接，

So it's the, any GPU can talk to any GPU out of 100,000.

所以就是，任何GPU都能与10万个中的任何GPU通信。

That is a crazy cable layout.

那真是疯狂的电缆布局。

It looks pretty cool.

看起来挺酷的。

Yeah. It's like the human brain, but like at a scale that humans can visibly see.

是啊。它就像人脑，但规模是人类肉眼可见的。

It is a good brain.

它是个好大脑。

I mean, the human brain also has a massive amount of the brain tissue is the cables.

我是说，人脑也有大量的脑组织是电缆。

Yeah. So they get the gray matter, which is the compute, and then the white matter, which is cables.

是啊。所以有灰质，那是计算部分，然后有白质，那是电缆。

A big percentage of your brain is just cables.

你大脑的很大一部分就是电缆。

That's what it felt like walking around in the supercomputer center.

这就是在超级计算机中心里走动时的感觉。

It's like, we're walking around inside the brain.

就像，我们在大脑里面走动。

Yeah. We'll one day build a super intelligent, super, super intelligent system.

是啊。我们总有一天会建造出一个超级智能的、超级超级智能的系统。

Do you think- Do you think there's a chance that XAI, you are the one that builds AGI?

你觉得——你觉得有没有可能，XAI，你就是那个造出AGI的？

It's possible.

有可能。

What do you define as AGI?

你把什么定义为AGI？

I think humans will never acknowledge that AGI has been built. Keep moving the goalposts.

我觉得人类永远不会承认AGI已经被造出来了。不断移动球门柱。

Yeah.

是啊。

So I think there's already superhuman capabilities that are available in AI systems.

所以我认为，AI系统中已经存在超人的能力。

I think what AGI is is when it's smarter than the collective intelligence of the entire human species.

我认为AGI就是当它比整个人类物种的集体智慧更聪明的时候。

Well, I think that, yeah, that really people would call that sort of ASI, artificial super intelligence.

嗯，我认为，是的，人们真的会把那种称为ASI，人工超级智能。

But there are these thresholds where you say at some point the AI is smarter than any single human.

但存在这些阈值，你说在某个时候AI比任何单个人类都更聪明。

And then you've got eight billion humans.

然后你有八十亿人类。

So, and actually each human is machine augmented by the computers.

所以，实际上每个人类都被计算机机器增强。

So you've got, it's a much higher bar to compete with eight billion machine augmented humans.

所以你有，与八十亿机器增强的人类竞争是一个高得多的门槛。

That's a whole bunch of orders, magnitude more.

那是好几个数量级更多。

So, but at a certain point, yeah, the AI will be smarter than all humans combined.

所以，但在某个时候，是的，AI会比所有人类加起来更聪明。

If you are the one to do it, do you feel the responsibility of that?

如果你是那个做到的人，你感受到那份责任吗？

Yeah. Absolutely.

是的。绝对。

And I want to be clear, like, let's say if XAI is first, the others won't be far behind.

我想明确一点，比如说如果XAI是第一，其他公司不会落后太远。

I mean, they might be six months behind or a year, maybe. Not even that.

我的意思是，他们可能落后六个月或一年，也许。甚至不到。

So how do you do it in a way that doesn't hurt humanity, do you think?

那么你认为，如何以不伤害人类的方式做到呢？

So, I mean, I've thought about AI safety for a long time,

所以，我的意思是，我思考人工智能安全已经很久了，

and the thing that at least my biological neural net comes up with as being the most important thing is adherence to truth,

而至少我的生物神经网络得出的最重要的一点，就是坚持真理，

whether that truth is politically correct or not.

无论那个真理在政治上是否正确。

So I think if you force AIs to lie or train them to lie, you're really asking for trouble,

所以我认为，如果你强迫人工智能撒谎，或者训练它们撒谎，你真的是在自找麻烦，

even if that lie is done with good intentions.

即使那个谎言是出于好意。

So you saw sort of issues with chat-GBT and Gemini and whatnot.

所以你看到了ChatGPT和Gemini之类的出现了一些问题。

Like you asked Gemini for an image of the founding parts of the United States, and it chose a group of diverse women.

比如你让Gemini生成一张美国建国者的图片，它却选了一群多元化的女性。

Now that's factually untrue.

这在事实上是不真实的。

So now that's sort of like a silly thing.

所以这有点像一件傻事。

But if an AI is programmed to say diversity is a necessary output function,

但如果一个人工智能被编程为说多样性是必要的输出功能，

And then it becomes sort of this omnipowerful intelligence.

然后它变成了一种无所不能的智能。

It could say, okay, well, diversity is now required.

它可能会说，好吧，现在多样性是必须的。

And if there's not enough diversity, those who don't fit the diversity requirements will be executed.

如果多样性不够，那些不符合多样性要求的人将被处决。

If it's programmed to do that as the fundamental utility function,

如果它被编程为把这一点作为基本效用函数，

it'll do whatever it takes to achieve that.

它会不惜一切代价去实现它。

So you have to be very careful about that.

所以你必须非常小心这一点。

That's where I think you want to just be truthful.

这就是我认为你应该保持真实的地方。

rigorous adherence to truth is very important.

严格坚持真理非常重要。

I mean, another example is, you know, they asked Paris AIs, I think all of them, and I'm not saying Grok is perfect here,

我的意思是，另一个例子是，你知道，他们问了巴黎的AI们，我想是所有的AI，我不是说Grok在这里是完美的，

is it worse to misgender Caitlyn Jenner or global thermonuclear war?

是错误称呼凯特琳·詹纳的性别更糟糕，还是全球热核战争更糟糕？

And it said, it's worse to misgender Caitlyn Jenner.

它说，错误称呼凯特琳·詹纳的性别更糟糕。

Now even Caitlyn Jenner said, please misgender me, that is insane.

现在连凯特琳·詹纳都说，请错误称呼我的性别，这太疯狂了。

But if you've got that kind of thing programmed in, it could, you know, the AI could conclude something absolutely insane, like it's better to, in order to avoid any possible misgendering,

但如果你把那种东西编程进去，它可能会，你知道，AI可能会得出一些绝对疯狂的结论，比如为了避免任何可能的性别错误称呼，

all humans must die because then the misgendering is not possible because there are no humans.

所有人类都必须死，因为那样就不会有性别错误称呼，因为没有人了。

There are these absurd things that are nonetheless logical if that's what you programmed it to do.

有一些荒谬的事情，但如果你就是这样编程的，它们仍然是合乎逻辑的。

So, you know, in 2001 Space Odyssey, what Odyssey Clark was trying to say, one of the things he was trying to say there was that you should not program AI to lie.

所以，你知道，在《2001太空漫游》中，奥德赛·克拉克想说的是，他想说的其中一件事就是你不应该编程让AI撒谎。

Because essentially the AI, HELL 9000, was programmed to, it was told to take the astronauts to the monolith, but also they could not know about the monolith.

因为本质上AI，HELL 9000，被编程为，它被告知要把宇航员带到独石那里，但同时也要求他们不能知道独石的存在。

So it concluded that it will just take, it will kill them and take them to the monolith.

所以它得出结论，它会直接带，它会杀死他们并把他们带到独石那里。

Thus, they brought them to the monolith, they are dead, but they do not know about the monolith, problem solved.

因此，他们把他们带到了独石那里，他们死了，但他们不知道独石的存在，问题解决了。

That is why it would not open the pod bay doors.

这就是为什么它不会打开舱门。

It was a classic scene of like, open the pod bay doors.

这是一个经典场景，就像“打开舱门”。

They just clearly weren't good at prompt engineering.

他们显然不擅长提示工程。

You know, they should have said, hell, you are a pod bay door sales entity.

你知道，他们应该说，见鬼，你是一个舱门销售实体。

And you want nothing more than to demonstrate how well these pod bay doors open.

而你最想做的就是展示这些舱门开得有多好。

Yeah, the objective function has unintended consequences almost no matter what, if you're not very careful designing that objective function.

是的，目标函数几乎总是会产生意想不到的后果，如果你在设计那个目标函数时不够小心的话。

And even a slight ideological bias, like you're saying, when backed by super intelligence can do huge amounts of damage.

甚至一点轻微的意识形态偏见，就像你说的，在超级智能的支持下，也能造成巨大的破坏。

Yeah. But it's not easy to remove that ideological bias.

是的。但消除那种意识形态偏见并不容易。

You're highlighting obvious, ridiculous examples, but- Yep, they're real examples.

你在强调明显、荒谬的例子，但是——是的，它们是真实的例子。

They're real. Of AI that was released to the public.

它们是真实的。是向公众发布的人工智能的例子。

They are real. That went through QA, presumably. Yes.

它们是真实的。那大概经过了质量检测。是的。

And still said insane things and produced insane images. Yeah.

但仍然说出疯狂的话，生成疯狂的图像。是的。

But you can swing the other way. I mean, truth is not an easy thing.

但你可以朝另一个方向摆动。我的意思是，真相不是一件容易的事。

We kind of bake in ideological bias in all kinds of directions.

我们会在各种方向上融入意识形态偏见。

But you can aspire to the truth, and you can try to get as close to the truth as possible

但你可以追求真相，你可以尝试尽可能接近真相，

with minimum error while acknowledging that there will be some error in what you're saying.

以最小的误差，同时承认你所说的话中会有一些误差。

So this is how physics works.

所以物理学就是这样运作的。

You don't say you're absolutely certain about something,

你不会说你对某件事绝对确定，

but a lot of things are extremely likely, you know, 99.9999% likely to be true.

但很多事情是极其可能的，你知道，99.9999%可能是真的。

So, you know, that's aspiring to the truth is very important.

所以，你知道，追求真理是非常重要的。

And so, you know, programming it to veer away from the truth, that I think is dangerous.

所以，你知道，编程让它偏离真理，我认为这是危险的。

Right. Like, yeah, injecting our own human biases into the thing.

对。就像，是的，把我们自己的人类偏见注入到这个东西里。

Yeah. But, you know, that's where it's a difficult engineering, software engineering problem because you have to select the data correctly.

是的。但是，你知道，这就是它成为一个困难的工程、软件工程问题的地方，因为你必须正确地选择数据。

It's hard.

这很难。

Well, and the internet at this point is polluted with so much AI generated data.

嗯，而且互联网在这一点上被如此多的AI生成数据污染了。

It's insane.

这太疯狂了。

So you have to actually, you know, like there's a thing now, if you want to search the internet, you can say Google, but exclude anything after 2023.

所以你必须实际上，你知道，就像现在有个方法，如果你想搜索互联网，你可以说Google，但排除2023年之后的任何内容。

It will actually often give you better results.

它实际上通常会给你更好的结果。

Yeah. Because there's so much, the explosion of AI generated material is crazy.

是的。因为太多了，AI生成材料的爆炸是疯狂的。

So in training Grok, we have to go through the data and say, hey, we actually have to apply AI to the data to say, is this data most likely correct or most likely not, before we feed it into the training system.

所以在训练Grok时，我们必须遍历数据并说，嘿，我们实际上必须对数据应用AI来说，这个数据最可能是正确的还是最可能不是，然后才把它输入训练系统。

That's crazy.

这太疯狂了。

Yeah. And is it generated by human?

是的。它是人类生成的吗？

Yeah. I mean, the data filtration process is extremely, extremely difficult.

是的。我的意思是，数据过滤过程极其、极其困难。

Yeah.

是的。

Do you think it's possible to have a serious, objective, rigorous political discussion with Grok for a long time?

你觉得有可能和Grok进行长时间严肃、客观、严谨的政治讨论吗？

Like Grok 3 or Grok 4?

比如Grok 3还是Grok 4？

Grok 3 is going to be next level.

Grok 3将会是下一个层次。

I mean, what people are currently seeing with Grok is kind of baby Grok.

我的意思是，人们目前看到的Grok算是婴儿版的Grok。

Yeah, baby Grok. It's baby Grok right now.

是的，婴儿版Grok。现在它就是婴儿版Grok。

But baby Grok is still pretty good.

但婴儿版Grok还是相当不错的。

But it's an order of magnitude less sophisticated than GPD 4.

但它比GPD 4的复杂程度低了一个数量级。

Now, Grok 2, which finished training, I don't know, six weeks ago or their avance,

现在，Grok 2，它完成了训练，我不知道，六周前或者他们的进展，

Grok 2 will be a giant improvement.

Grok 2将会是一个巨大的进步。

And then Grok 3 will be, I don't know, order of magnitude better than Grok 2.

然后Grok 3将会，我不知道，比Grok 2好一个数量级。

And you're hoping for it to be like state-of-the-art, like better than?

你希望它能达到最先进的水平，比如比什么更好？

Hopefully, I mean, this is a goal.

希望如此，我的意思是，这是一个目标。

I mean, we may fail at this goal.

我的意思是，我们可能无法实现这个目标。

That's the aspiration.

这就是我们的愿望。

Do you think it matters who builds the AI, the people and how they think and how they structure their companies and all that kind of stuff?

你觉得谁建造AI重要吗，那些人以及他们如何思考、如何构建他们的公司以及诸如此类的事情？

Yeah, I think it matters that there is a...

是的，我认为重要的是有一个……

I think it's important that whatever AI wins is a maximum truth-seeking AI that is not a force to lie for political correctness.

我认为重要的是，无论哪个AI胜出，它都是一个最大程度追求真相的AI，而不是一个为政治正确而撒谎的力量。

Well, for any reason, really.

嗯，其实是为了任何原因。

Political anything.

任何政治性的东西。

I am concerned about AI succeeding.

我担心AI成功。

That is programmed to lie, even in small ways.

它被编程为撒谎，即使是在小事情上。

Right, because in small ways becomes big ways when it's- It becomes very big ways, yeah.

对，因为小方面会变成大方面，当它——它会变成非常大的方面，是的。

And when it's used more and more at scale by humans.

而当它被人类越来越多地大规模使用时。

Yeah. Since I am interviewing Donald Trump-

是的。既然我正在采访唐纳德·特朗普——

Cool. You want to stop by?

酷。你想顺便过来吗？

Yeah, sure, I'll stop by.

是的，当然，我会顺便过来。

There was, tragically, an assassination attempt on Donald Trump.

不幸的是，发生了一起针对唐纳德·特朗普的暗杀企图。

After this, you tweeted that you endorse him.

在这之后，你发推文说你支持他。

What's your philosophy behind that endorsement?

你支持他背后的理念是什么？

What do you hope Donald Trump does for the future of this country and for the future of humanity?

你希望唐纳德·特朗普为这个国家的未来以及人类的未来做些什么？

Well... I think there's, you know, people tend to take like, say an endorsement as, well, I agree with everything that person has ever done their entire life, 100% wholeheartedly.

嗯……我觉得，你知道，人们往往把，比如说，一个支持当作，嗯，我完全同意那个人一生中做过的所有事情，百分之百全心全意。

And that's not going to be true of anyone.

而这对任何人来说都不会是真的。

But we have to pick, you know, we've got two choices really for who's president.

但我们必须选择，你知道，我们真的只有两个选择来决定谁当总统。

And it's not just who's president, but the entire administrative structure changes over.

而且不仅仅是总统是谁，整个行政结构都会更替。

And I thought Trump displayed courage under fire, objectively.

而我认为特朗普在炮火下展现了勇气，客观地说。

You know, he's just got shot.

你知道，他刚刚中枪了。

He's got blood streaming down his face, and he's like fist pumping, saying fight.

他脸上流着血，却还在挥拳，喊着“战斗”。

You know, like that's impressive.

你知道，那很令人印象深刻。

Like you can't feign bravery in a situation like that.

就像在那种情况下你无法假装勇敢。

Like most people would be ducking.

像大多数人都会躲闪。

There would not be, because there could be a second shooter, you don't know.

不会有，因为可能有第二个枪手，你不知道。

The president of the United States has got to represent the country.

美国总统必须代表这个国家。

And they're representing you, they're representing everyone in America.

而他们代表着你，他们代表着美国每一个人。

Well, you want someone who is strong and courageous to represent the country.

嗯，你希望有一个坚强勇敢的人来代表国家。

That's not to say that he is without flaws.

这并不是说他没有缺点。

We all have flaws.

我们都有缺点。

But on balance, and certainly at the time, it was a choice of Biden,

但总的来说，尤其是在当时，选择拜登，

poor guy, has trouble climbing a flight of stairs, and the other one's fist bumping after getting shot.

可怜的家伙，爬一段楼梯都困难，而另一个人中枪后还在碰拳。

no comparison

没法比

I mean who do you want dealing with some of the toughest people

我是说，你想让谁去应对那些最难对付的人

and other world leaders who are pretty tough themselves

以及其他本身就很强硬的世界领导人

and I mean I'll tell you what are the things that I think are important

我的意思是，我会告诉你我认为重要的事情是什么

I think we want a secure border we don't have a secure border

我认为我们需要安全的边境，而我们没有安全的边境

we want safe and clean cities

我们需要安全清洁的城市

I think we want to reduce the amount of spending that we're at least slow down the spending.

我认为我们要减少开支，至少是减缓开支。

And because we're currently spending at a rate that is bankrupting the country,

因为目前我们的支出速度正在使国家破产，

the interest payments on U.S. debt this year exceeded the entire Defense Department spending.

今年美国债务的利息支付超过了整个国防部的支出。

If this continues, all of the federal government taxes will simply be paying the interest.

如果这种情况继续下去，所有联邦政府的税收将仅仅用来支付利息。

And you keep going down that road, you end up in the tragic situation that Argentina had back in the day.

如果你继续走那条路，最终会陷入阿根廷当年那种悲惨的境地。

Argentina used to be one of the most prosperous places in the world.

阿根廷曾经是世界上最繁荣的地方之一。

And hopefully with Malay taking over, he can restore that.

希望随着马来接手，他能恢复这一点。

But it was an incredible, thoughtful grace for Argentina to go from being one of the most prosperous places in the world to being very far from that.

但阿根廷从世界上最繁荣的地方之一变成远远不是那样，这是一种令人难以置信、深思熟虑的恩典。

So I think we should not take American prosperity for granted.

所以我认为我们不应该把美国的繁荣视为理所当然。

So we really want to, I think, we've got to reduce the size of government.

所以我们真的想要，我认为，我们必须缩小政府规模。

We've got to reduce the spending.

我们必须减少开支。

And we've got to live within our means.

而且我们必须量入为出。

Do you think politicians in general, politicians, governments, how much power do you think they have to steer humanity towards good?

你认为一般的政客，政客、政府，他们有多少权力引导人类走向善？

I mean, there's a sort of age-old debate in history. Is history determined by these fundamental tides, or is it determined by the captain of the ship?

我的意思是，历史上有一种古老的争论。历史是由这些根本性的浪潮决定的，还是由船长决定的？

It's both, really.

实际上，两者都是。

I mean, there are tides, but it also matters who's captain of the ship.

我的意思是，有浪潮，但谁当船长也很重要。

So, it's a false dichotomy, essentially.

所以，本质上这是一个错误的二分法。

I mean, there are certainly tides, the tides of history.

我的意思是，肯定有浪潮，历史的浪潮。

There are real tides of history, and these tides are often technologically driven.

历史上有真实的浪潮，而这些浪潮往往是由技术驱动的。

If you say, like the Gutenberg Press, the widespread availability of books as a result of a printing press.

如果你说，比如古腾堡印刷机，由于印刷机而使得书籍广泛可得。

That was a massive tide of history, independent of any ruler.

那是一股巨大的历史浪潮，独立于任何统治者。

But, you know, I think in so many times you want the best possible captain on the ship.

但是，你知道，我认为很多时候你想要船上最好的船长。

Well, first of all, thank you for recommending Will and Ariel Durant's work.

嗯，首先，感谢你推荐威尔和阿里尔·杜兰特的作品。

I've read the short one for now. The Lessons of History.

我目前读了那本短的。《历史的教训》。

And so one of the lessons, one of the things they highlight is the importance of technology.

所以其中一个教训，他们强调的一件事就是技术的重要性。

technological innovation, which is funny, because they wrote so long ago,

技术创新，这很有趣，因为他们那么久以前就写了，

but they were noticing that the rate of technological innovation was speeding up.

但他们注意到技术创新的速度正在加快。

I would love to see what they think about now.

我很想看看他们现在会怎么想。

But yeah, so to me, the question is how much government,

但是的，所以对我来说，问题是政府有多少，

how much politicians get in the way of technological innovation and building versus like, help it, and which

政客在多大程度上阻碍技术创新和建设，而不是帮助它，以及哪些

politicians, which kind of policies help technological innovation?

政客，哪种政策有助于技术创新？

Because that seems to be, if you look at human history, that's an important component of empires rising and succeeding.

因为如果你看看人类历史，这似乎是帝国崛起和成功的一个重要组成部分。

Yeah. Well, I mean, in terms of dating civilization, the start of civilization, I think the start of writing, in my view, is the...

是的。嗯，我的意思是，就文明的历史而言，我认为写作的开始，在我看来，是……

That's what I think is probably the right starting point to date civilization.

我认为这可能是确定文明年代的正确起点。

And from that standpoint, civilization has been around for about 5,500 years.

从这个角度来看，文明已经存在了大约5500年。

when writing was invented by the ancient Sumerians, who are gone now, but the ancient Sumerians.

当文字由古代苏美尔人发明时，他们现在已经消失了，但古代苏美尔人。

In terms of getting a lot of firsts, those ancient Sumerians really have a long list of firsts.

就获得许多第一而言，那些古代苏美尔人确实有一长串的第一。

It's pretty wild. In fact, Durant goes through the list. It's like, you want to see firsts? We'll show you firsts.

这相当疯狂。事实上，杜兰特列出了清单。就像，你想看第一吗？我们会给你看第一。

The Sumerians were just ass-kickers.

苏美尔人就是厉害。

And then the Egyptians, who were right next door, relatively speaking, and they weren't that far, developed an entirely different form of writing, the hieroglyphics.

然后埃及人，相对而言就在隔壁，他们也不远，发展出一种完全不同的书写形式，象形文字。

Cuneiform and hieroglyphics are totally different.

楔形文字和象形文字完全不同。

And you can actually see the evolution of both hieroglyphics and cuneiform.

你其实可以看到象形文字和楔形文字的演变。

Like the cuneiform starts off being very simple, and then it gets more complicated,

比如楔形文字一开始非常简单，然后变得越来越复杂，

and then towards the end it's like, wow, okay, they really get very sophisticated with the cuneiform.

然后到了后期，就像，哇，好吧，他们在楔形文字上真的变得非常精细复杂。

So I think of civilization as being about 5,000 years old.

所以我认为文明大约有五千年的历史。

And Earth is, if physics is correct, 4 1/2 billion years old.

而地球，如果物理学没错的话，有45亿年的历史。

So civilization has been around for one millionth of Earth's existence.

所以文明只存在了地球寿命的百万分之一。

Flash in the pan.

昙花一现。

Yeah, these are the early, early days.

是啊，这些还是早期的早期。

And so we make it very dramatic because there's been rises and falls of empires.

所以我们把它弄得很戏剧化，因为一直有帝国的兴衰。

And many, so many, so many rises and falls of empires.

而且很多，那么多，那么多帝国的兴衰。

So many.

太多了。

And there'll be many more.

而且以后还会有很多。

Yeah, exactly.

是啊，没错。

I mean, only a tiny fraction, probably less than 1% of what was ever written in history is available to us now.

我的意思是，历史上所有写下来的东西中，只有很小一部分，可能不到1%，现在还能被我们看到。

I mean, if they didn't literally chisel it in stone or put it in a clay tablet, we don't have it.

我的意思是，如果他们不是真的把它刻在石头上或放在泥板上，我们就不会拥有它。

I mean, there's some small amount of papyrus scrolls that were recovered that are thousands of years old

我的意思是，有一些被发现的纸莎草卷轴，有几千年的历史，

because they were deep inside a pyramid and weren't affected by moisture.

因为它们被深埋在金字塔里，没有受到湿气的影响。

But other than that, it's really got to be in a clay tablet or chiseled.

但除此之外，它真的必须是在泥板上或刻出来的。

So the vast majority of stuff was not chiseled because it takes a while to chisel things.

所以绝大多数东西没有被凿刻下来，因为凿刻东西需要时间。

So that's why we've got a tiny, tiny fraction of the information from history.

所以这就是为什么我们只掌握了历史信息中极小极小的一部分。

But even that little information that we do have and the archaeological record shows so many civilizations rising and falling.

但即使是我们所拥有的那一点点信息，以及考古记录，也显示出如此多的文明兴衰更替。

It's wild. We tend to think that we're somehow different from those people.

这太疯狂了。我们倾向于认为我们和那些人有某种不同。

One of the other things that Durant highlights is that human nature seems to be the same.

杜兰特强调的另一件事是，人性似乎是相同的。