科技的共同主线：从美国动力学到医疗和教育中的人工智能

I. 科技领域的连贯性与统一价值

We're excited to do our first episode of Katherine and I's new podcast series with A16Z where we run through the stories of the internet. Eddie, thank you for being our first guest.

我们很高兴能与 A16Z 一起推出凯瑟琳和我的新播客系列的第一集，我们将回顾互联网的故事。艾迪，感谢你成为我们的第一位嘉宾。

[00:00:25] Katherine, some people say: "How is A16Z coherent? They are leaning into the American Dynamism (美国动力学: a16z的一个投资主题，关注国家利益的关键产业，如国防、教育、医疗等，致力于用技术解决重大公共问题) brand with investments, but also I see all these consumer investments. How does that make any sense?" More seriously, how should one think about the variety of what we do and how it is coherent and cohesive to do something as serious as American Dynamism but also do consumer internet, for example?

凯瑟琳，有些人会说：“A16Z 是如何保持连贯性的？他们投入了‘美国动力学’的品牌和投资，但我也看到了所有这些消费者投资。这怎么说得通？”更严肃地说，人们应该如何看待我们所做事情的多样性，以及如何在做像美国动力学这样严肃的事情的同时，还能做消费互联网这类事情，使其保持连贯性和凝聚力？

[00:00:53] Well, I always get the question or someone put on the internet maybe a week ago, "How do games have anything to do with American dynamism?" And I always answer that question with one word, which is Palmer. Because Oculus would not have happened if not for Palmer's obsession with gaming.

嗯，我总是被问到这个问题，或者可能一周前有人在互联网上提出：“游戏与美国动力学有什么关系？”我总是用一个词来回答这个问题，那就是帕尔默（Palmer）。因为如果不是帕尔默对游戏的热爱，就不会有 Oculus。

[00:01:17] There's so much continuity in the world of tech that I sometimes think people silo things almost because we silo things. We're very different worlds, we have very different types of expertise in terms of what we're looking for as investors. But the continuity of what is building and what is technology, it spans all sorts of realms and frankly it spans every industry, every realm of human life.

在科技界，有如此多的连续性，我有时觉得人们将事物孤立起来，几乎是因为我们自己将事物孤立了。我们生活在非常不同的世界，作为投资者，我们在寻找不同类型的专业知识。但是，建设和技术的连续性，它跨越了各种领域，坦白地说，它跨越了每一个行业，人类生活的每一个领域。

[00:02:22] And I always say, Eric, I've been begging you and telling you we have to have the Techno Optimist Manifesto (马克·安德森的《技术乐观主义宣言》: 一份倡导相信技术进步和积极建设未来的哲学宣言) on the front page of our website, because to me, that is the coherent thesis of what all building is. It's the thesis of every founder's journey.

我总是说，埃里克，我一直在恳求你，告诉你我们必须把《技术乐观主义宣言》（马克·安德森的《技术乐观主义宣言》: 一份倡导相信技术进步和积极建设未来的哲学宣言）放在我们网站的首页，因为对我来说，那是所有“建设”的连贯性主旨。它是每个创始人旅程的主旨。

[00:04:08] Everything great looks like or starts looking like a toy. Toys or games can inspire so much beyond that; they can sort of be a playground to experiment with new technologies that can then be the inspiration for something way bigger or very different. And it's all a boiling pot.

所有伟大的事物看起来或开始看起来像一个玩具。玩具或游戏可以激发超越其本身的许多事物；它们可以成为一个试验新技术的游乐场，这些新技术随后可以成为更大或非常不同的事物的灵感来源。这是一个沸腾的熔炉。

[00:04:50] I actually have a great, very timely story about this. We were just in Washington, D.C. with Mark and a senator was talking about what could be very valuable for the defense world. He was talking about how in Ukraine, just-in-time manufacturing feeds into building things in the trenches very quickly. And Mark responded to him and said: "Well, we do this with toys. This is how we build hardware toys. Of course we can do this with hardware in America because this is how any type of hardware is built. You iterate very rapidly and you iterate in the field. It's only in defense where we don't do this."

我有一个非常及时且精彩的故事。我们当时和马克一起在华盛顿特区，一位参议员正在谈论什么对国防领域非常有价值。他谈到在乌克兰，即时制造如何支持在战壕中快速构建事物。马克回答他：“我们用玩具就是这么做的。我们就是这样制造硬件玩具的。我们当然可以在美国用硬件做到这一点，因为任何类型的硬件都是这样构建的。你迭代得非常快，并且在现场进行迭代。只有在国防领域，我们才不这样做。”

1.1 美国动力学与加密货币的哲学交汇点

[00:06:41] I think of crypto as freedom-promoting technology. In the horrible case where America becomes not freedom-promoting, then you could see crypto as a hedge. I think of Satoshi as trying to create in software a way to represent those same values that was in some sense immune to whatever the states happened to be doing. So I don't think that they're necessarily replacements; I think that they're complements.

我认为加密货币是促进自由的技术。在可怕的情况下——如果美国变得不促进自由——那么你可以将加密货币视为一种对冲。我认为中本聪试图在软件中创造一种方式来代表那些相同的价值观，使其在某种意义上能够免疫于国家当时正在做的任何事情。所以我认为它们不一定是替代品；我认为它们是互补品。

[00:07:26] They're trying to get at the same underlying value system, the same concepts about freedom to move capital, property rights for individuals, the ability of capital to flow freely, payments to flow freely, for people to be able to own things. Just two different mechanisms for putting them in place.

它们试图达到相同的基本价值体系，即关于资本流动自由、个人财产权、资本自由流动、支付自由流动以及人们能够拥有事物的能力等概念。只是将它们落实到位采用了两种不同的机制。

[0008:52] I think that what the state would like to see, if the state is freedom-promoting, and in America it certainly is, I think the state would like to see people doing these experiments in a way that's totally visible and legible. And that's exactly what crypto does.

如果国家是促进自由的，就像在美国肯定是这样，我认为国家会乐于看到人们以一种完全可见和清晰可读的方式进行这些实验。而这正是加密货币所做的。

[00:09:33] I've picked up on philosophical alignment between the types of founders who start crypto companies and the types of founders who start American Dynamism companies. The conversations we were having about what does America need, like what are the biggest problems for our generation, these people are very aligned at what the problems are.

我发现，创办加密货币公司的创始人和创办美国动力学公司的创始人之间存在哲学上的一致性。我们当时正在讨论美国需要什么，比如我们这一代人面临的最大问题是什么，这些人对问题的认知非常一致。

[00:11:10] As an Federalist (联邦主义: 一种政治哲学，主张权力在中央政府和地方政府之间分配，在美国通常指权力分散的结构), like someone who is in deep favor of the Federalist experiment, I think that's something that's under-discussed about what makes America extraordinary and unique. And where is decentralization most prominent in the tech community? It is in the crypto world.

作为一个联邦主义者（Federalism: 一种政治哲学，主张权力在中央政府和地方政府之间分配，在美国通常指权力分散的结构），一个非常支持联邦主义实验的人，我认为这是关于美国非凡和独特之处的一个被低估的话题。而在技术社区中，去中心化在哪里最为突出？它就在加密世界。

II. 互联网、AI与信息标准

2.1 互联网医疗与更高的信息标准

[00:14:22] There's something about the way that this generation understands medicine that actually comes from: the internet is your doctor now. And maybe 10 years ago there was a lot of fear around that, but I think we're just living in this completely different reality where the people who are having children now grew up at an age of just completely distributed information, particularly around things like medicine.

这一代人理解医学的方式，实际上源于：互联网现在就是你的医生。也许十年前，对此存在很多恐惧，但我认为我们现在生活在一个完全不同的现实中，现在生孩子的人成长于一个信息完全分布式传播的时代，尤其是在医学之类的事情上。

[00:15:29] What you're seeing from this generation isn't like skepticism of experts, it's not how it's portrayed. You can't call it quackery; you kind of have to call it like wisdom of crowds.

你从这一代人身上看到的，并不是对专家的怀疑，不像外界描绘的那样。你不能称之为庸医之术；你必须称之为群体的智慧。

[00:16:19] I joke it's my ChatGPT baby, because I upload every single medical diagnostic, every blood report, everything to ChatGPT and it gives me a ton of data that the doctors don't have time to give me.

我开玩笑说这是我的 ChatGPT 宝宝，因为我将每一次医学诊断、每一次验血报告，所有东西都上传到 ChatGPT（生成式预训练变换器），它给了我大量医生没有时间给我的数据。

[00:17:00] It's just a higher information standard, it's a higher epistemic standard. It's maybe resisting the idea to accept unchallenged what the first expert you talked to said. That's how I'm seeing it. We always come to the doctor now informed by a bunch of research.

这只是更高的信息标准，更高的认知标准。它可能是抵制毫无疑问地接受你交谈的第一个专家所说的一切。我就是这样看待它的。我们现在去看医生，总是带着大量研究的信息。

[00:17:27] I'll send the same thing to ChatGPT, Anthropic, and Grock at the same time, force them to fight against each other, and then end up with just a list of interesting sources that I put in front of the doctor. That's the feedback loop now.

我会同时将同样的东西发送给 ChatGPT、Anthropic 和 Grock，让它们相互竞争，然后最终得到一份有趣的来源列表，再将其呈交给医生。这就是现在的反馈循环。

[00:19:16] What people are worried about is when they imagine that someone just picks the cheapest, lowest IQ LLM (Large Language Model: 大型语言模型) and then accepts the first thing that it says uncritically. But that's the same thing if you go to a terrible doctor. The answer is obviously a more rigorous standard of information, more information sources, uncorrelated channels of information.

人们担心的是，他们想象有人只是选择最便宜、智商最低的 LLM（大型语言模型），然后不加批判地接受它说的第一件事。但如果你去看一个糟糕的医生，情况也是一样的。答案显然是更严格的信息标准、更多的信息来源、不相关的信道。

2.2 ADHD的系统性激励问题

[00:21:33] 23% of all 17-year-old boys have been given an ADHD (Attention-Deficit/Hyperactivity Disorder: 注意缺陷多动障碍) diagnosis. So that's one in four boys in America. There's something wrong with them? Or you can say, "Okay, we're medicating boyhood."

美国所有 17 岁男孩中，有 23% 被诊断患有 ADHD（注意缺陷多动障碍）。这相当于美国每四个男孩中就有一个。是他们出了什么问题？或者你可以说：“好吧，我们正在给男童期用药。”

[00:22:24] The incentive systems around a lot of these things are: if you go to the doctor, parents want these ADHD diagnoses because it gives you extra time, extra resources. It makes school more manageable for high-energy boys. The reasons the school wants it is because they get more dollars from the state if they have special needs students.

围绕很多这些事情的激励系统是：如果你去看医生，家长希望得到 ADHD 诊断，因为它能让你获得额外的时间和资源。这使得精力充沛的男孩在学校里更容易管理。学校希望得到它的原因是，如果他们有特殊需要的学生，他们可以从州政府那里获得更多的资金。

[00:24:39] There's no downside to a diagnosis. You have total optionality on whether you want to take the medication, you have no incentive to resist. We can identify the pathologies in students, we want to help students, we have tools to help them, and so and there's no downside to identifying them, misidentifying them. So let's just generously hand these out.

诊断没有坏处。你完全可以选择是否服药，你没有理由抗拒。我们可以识别学生的问题，我们想帮助学生，我们有工具来帮助他们，而且识别他们、错误地识别他们都没有坏处。所以我们就大方地分发这些诊断。

[00:26:04] One of the issues is the diagnostic criteria can shift. As people want to identify these cases, see very little downside in identifying them, they may broaden it in an effort to precisify the diagnostic criteria. This can explain what appears to be an increase in prevalence.

其中一个问题是诊断标准会发生变化。当人们想要识别这些病例，并且认为识别它们几乎没有坏处时，他们可能会为了精确诊断标准而将其放宽。这可以解释看似患病率的增加。

III. 教育的未来与深度兴趣驱动

[00:27:09] Alpha School is interesting because one of their main KPIs (Key Performance Indicators: 关键绩效指标) is "Do kids love school?" My school was like a prison; it was so boring. It's exciting to see kind of different models.

Alpha School 很有趣，因为他们的主要 KPI（关键绩效指标）之一是“孩子们喜欢学校吗？”我的学校就像一座监狱，太无聊了。很高兴看到不同的模式。

[00:28:22] We can put an infinite treadmill in front of kids about things that they can learn at a low-cost, infinite treadmill, where they can just train and learn whatever they want to arbitrary depth. Designing something that is rigorous enough and gives them strong enough feedback and kind of encourages them to really maximize that, that seems like the educational challenge of the 21st century to me.

我们可以为孩子们提供一个关于他们可以学习的事物的无限跑步机，一个低成本的无限跑步机，他们可以训练和学习任何他们想要达到的任意深度。设计一些足够严格、给予他们足够强反馈并鼓励他们真正最大化利用的东西，这对我来说似乎是 21 世纪的教育挑战。

[00:29:10] That arbitrary depth couldn't be nurtured when we were kids around random things. Our kids' play comes before they go to school, and their play is so hyper-specialized. If you really learn a lot about how a garbage disposal works, you really get into the weeds, you see a one cut in half and a diagram, and you see the pipes, there's so many approximate or adjacent things to that that end up generalizing.

当我们还是孩子时，围绕随机事物，这种任意深度是无法培养的。我们孩子的玩耍先于他们上学，他们的玩耍是如此高度专业化。如果你真的深入了解一个垃圾处理器是如何工作的，你深入细节，看到一个切开的剖面图和图表，看到管道，有如此多近似或相邻的事物，最终会泛化。

[0:32:55] Boredom is actually good. A lot of our life is pretty boring. It is important to learn how to work in big systems. It is important to kind of realize that not everyone's going to entertain your weird rabbit hole. So there is something about like getting that lesson early and having that sort of "I'm bored, I hate school, I'm going to figure out how to make it interesting for me" that could be very valuable for kids as well.

无聊实际上是好的。我们的生活很大一部分是相当无聊的。学习如何在大型系统中工作是很重要的。认识到不是每个人都会娱乐你古怪的“兔子洞”是很重要的。因此，及早学到这一课，拥有那种“我很无聊，我讨厌学校，我要想办法让自己觉得有趣”的经历，对孩子们来说也可能非常有价值。

IV. 育儿、家庭结构与社会系统性问题

[00:35:27] I am struck by how obvious it is how totally helpless they are. They are like functionally blind, deaf, like they're a little ball of nothing. Their helplessness is kind of the mirror of our capability, and we should really lean into that.

我很震惊于他们是多么的无助。他们就像功能性失明、失聪，就像一团虚无的小球。他们的无助感恰恰是映照出我们的能力，我们应该真正地专注于此。

[00:36:37] Without parental leave, it would be incredibly difficult, and it makes it very obvious to me why fertility is declining because the opportunity cost of this suffering is massive.

如果没有陪产假或产假，那将是极其困难的，这让我很清楚地看到了为什么生育率正在下降，因为这种痛苦的机会成本是巨大的。

[00:38:33] I'm fortunate enough to have my mother live with us, and she has been a godsend in holding these babies. It's like there used to be sort of a familial unit that took care. It's very apparent to me, especially if you don't have it, "How do two people do this?"

我很幸运能和我的母亲住在一起，她在抱着这些哭泣的婴儿方面简直是天赐之福。过去似乎有一种家庭单元来照顾孩子。对我来说，这一点非常明显，特别是如果你没有这样的支持：“两个人如何做到这一点？”

[00:39:02] Only until the nuclear family post-war did we really move to suburbs and separate sort of multigenerational family units from each other, where then you have one mother doing this all alone, which of course was the big complaint was loneliness in the 50s and 60s of these housewives.

直到战后核心家庭出现，我们才真正搬到郊区，并将多代家庭单位相互分离，这样就有了母亲独自一人做这一切，这当然是 50 年代和 60 年代家庭主妇的主要抱怨——孤独。

V. 社交网络、文化与碎片化

[00:42:04] The reactions to the assassination of Charlie Kirk as it sort of played out on the internet sort of demonstrated a bit of a gap between sort of Zoomer culture, internet culture and the rest of the population.

围绕查理·柯克（Charlie Kirk）遇刺事件在互联网上的反应，在某种程度上展示了 Zoomer 文化、互联网文化与其余人口之间存在的某种差距。

[00:44:45] Because the internet is so open by default, I think there's this assumption that it's all accessible and therefore should permeate. But we have managed to create new gradients, new selection methods to create isolated little corners of the internet. Gamer culture, you can't learn those things unless you play games.

因为互联网默认是如此开放，我认为存在一种假设，即一切都是可访问的，因此应该渗透。但是我们已经设法创造了新的梯度、新的选择方法，来创建互联网孤立的小角落。游戏玩家文化，除非你玩游戏，否则你学不到那些东西。

[0:45:55] "Your Insta is not my X." We were in totally different worlds. I actually think X is sort of a unifying platform from all of the different pockets because if you're a power user of X, a lot of the early data that was coming out on anything gets to X early.

“你的 Instagram 不是我的 X。”我们身处完全不同的世界。我实际上认为 X 在某种程度上是一个统一的平台，它将所有不同的群体连接起来，因为如果你是 X 的高级用户，很多关于任何事情的早期数据都会首先到达 X。

[0:47:38] If you want to find the ground truth and you do the legwork and you follow the right people, the answer is probably on X in a way that is unlike other places.

如果你想找到事实真相，并且你付出了努力，关注了正确的人，那么答案很可能就在 X 上，这与其他地方不同。

[0:48:15] Does it have to do with the fact that it's an open graph, quote tweets, and correcting people is a part of the norm in a way that is just totally not on Facebook or on Instagram?

这是否与它是一个开放图谱、引用推文以及纠正他人是常态有关，而这在 Facebook 或 Instagram 上则完全不是常态？

[00:50:14] It gives you a theory of mind for people you otherwise would not have a theory of mind for. And I think there's something about sort of the pseudo-anonymous accounts on X that actually do a really good job of giving you a full picture.

它让你对那些你原本不会有心智理论的人有了心智理论。我认为 X 上那些半匿名账户确实很好地为你提供了完整的图景。

[00:52:00] Every tweet is a focus group on some idea.

每一条推文都是针对某个想法的焦点小组。

[00:57:13] I've slightly mourned the death of the group chat. It is definitely a good thing that the group chats are no longer siloed, are no longer necessary. I do think a lot of the real interesting conversations have moved to X, and that is a good thing.

我有点哀悼群聊的消亡。群聊不再孤立、不再必要绝对是一件好事。我确实认为很多真正有趣的对话已经转移到了 X，这是一件好事。

[00:57:48] There's peace-time social networks and war-time social networks.

有和平时期的社交网络和战争时期的社交网络。