--- author: Norges Bank Investment Management date: '2026-10-21' guest: '' layout: post.njk source: https://www.youtube.com/watch?v=U5YtR3OG5Uo speaker: Norges Bank Investment Management tags: - artificial-general-intelligence - autonomous-agent - scaling-law - geopolitical-competition - leadership-evolution title: 专访埃里克·施密特：非人类智能崛起、算力军备竞赛与组织大洗牌 summary: Google 前 CEO 埃里克·施密特深度剖析人工智能的革命性冲击。他指出 AI 并非普通技术周期，而是人类历史上首次创造出非人类智能；技术周期从数年压缩至数月，企业架构与商业流程将被 AI Agent 彻底重塑。施密特尖锐对比欧美监管与创新环境，警示中国开源模型与低价战略的全球扩张，并分享其对领导力演进及终身投入科学事业的思考。 insight: '' draft: true series: '' category: tech-trends area: tech-engineering project: [] people: - Eric Schmidt companies_orgs: - Google - Anthropic - OpenAI products_models: [] media_books: - Genesis status: evergreen --- ### 创新与监管的博弈 **埃里克·施密特**: 监管者总是在创新之前就急于监管。我最喜欢举的一个例子是，五年前布鲁塞尔起草了他们的**人工智能**战略。当时我也参与协助起草了美国的战略。美国的战略写的是：“以最快的速度往前跑。将其与人类价值观对齐，如果出现坏事，立即纠正。”而欧洲人写的则是：“你不能做这个，你不能做那个，你不能做这个，你不能做那个。”在那样一个充斥着繁文缛节的受限市场里，你是无法经营初创企业的。
Original English **Eric Schmidt**: The regulators regulate before they innovate. My favorite example is that 5 years ago, Brussels wrote the AI strategy. I helped write the one for the US and the US one said, "Run as fast as you can. Align it with human values and if bad things happen, fix them immediately." The Europeans wrote, "You can't do this, you can't do this, you can't do this, you can't do this." And you can't run startups in a limited market where there's a lot of rules like that.
**尼古拉·坦根**: 大家好，我是**挪威主权财富基金**（Norwegian Sovereign Wealth Fund）首席执行官**尼古拉·坦根**。今天非常荣幸邀请到我们这个时代最具影响力的科技领袖之一——**埃里克·施密特**。埃里克在 2001 年至 2011 年期间担任 **Google** 首席执行官，此后成为人工智能领域的权威领军人物，持续为多国政府建言献策、撰写著作，并指出 AI 对世界的改变将远远超出我们大多数人的预期。埃里克，非常高兴你能来。
Original English **Nicolai Tangen**: Hi everyone, I'm Nicolai Tangen, the CEO of the Norwegian Sovereign Wealth Fund and today I'm joined by Eric Schmidt, one of the most influential technology executives of our time. Eric was CEO of Google from 2001 to 2011 and has since become one of the leading voices on AI, advising governments, writing books, and arguing that AI will change the world more than most of us expect. Eric, great to have you here.
**埃里克·施密特**: 我非常高兴来到这里。我想先说一句，你在挪威以及为挪威所做的一切都极其非凡，但这仅仅是个开始。
Original English **Eric Schmidt**: I'm so glad to be here and let me just tell you what you have done in Norway and for Norway is extraordinary, but it's just the beginning.
### 五十年科技浪潮与平台演进 **尼古拉·坦根**: 这取决于人工智能的发展走向。你在过去三十年里一直处于科技世界的中心舞台。从科技演进的视角来看，我们现在究竟处于什么位置？
Original English **Nicolai Tangen**: Well, that depends on where AI is going. So you've been at the center of the technology world now for three decades. Where are we now in the story of technology?
**埃里克·施密特**: 感觉像是过了五十年，我正在变老。我所在的行业经历了一系列蜕变。我最早是从 **IBM** 大型机时代开始的，随后经历了个人电脑（PC）时代、互联网时代、社交媒体时代，现在迎来了人工智能时代。 其中的每一次蜕变都极其宏大。之所以宏大，是因为它们都是**底层平台**（Platforms）。如果你处在我的位置上，你真正要寻找的，是那种能让整个海平面都为之抬升的变革，是能带来无数全新机遇的转折点，而且你必须搞清楚海平面为何会上升。 在我们现在的情况下，海平面上升首先是因为几年前在底层技术逻辑上取得了根本性的科学发现；但同时，它也是由我们所谓的**规模法则**（Scale）所驱动的。如果你投入两倍的计算量，你就能得到更智能的结果；如果投入十倍的计算量，结果还会更聪明。只要这种正相关关系还在持续，各大科技巨头就绝不会放慢脚步，它们会不断砸入数以百亿计的资金来扩建数据中心。这种算力集群的庞大程度，已经完全超出了普通人类的想象。
Original English **Eric Schmidt**: It feels like five decades. I'm getting older. My industry goes through a series of transformations. I started off in the IBM mainframe era and through the PC era, then the internet era, then the social media era, and now AI. And each of these transformations is vast. And the reason it's vast is that these are platforms. And so what you want to do if you're in my situation is you want to watch for something where the whole ocean is rising and that there are enormous number of new opportunities appearing and you want to figure out why the ocean is rising. Now in our case it's rising because a few years ago there were some fundamental discoveries in how technology could work. But it's also rising because of what we call scale.
**尼古拉·坦根**: 与你以往所见过的所有变革相比，这次的变革到底有多大？
Original English **Nicolai Tangen**: How big is this change compared to what you've seen before?
**埃里克·施密特**: 科技浪潮中的每一次迭代在体量上都显著超越上一轮。而这一次更是空前的，因为它渗透到了万事万物之中——**非人类智能的崛起**是迄今为止人类历史上最大的事件。你可以看看争夺行业领军地位的竞争有多么残酷血腥。我原以为过去个人电脑、互联网和手机时代的竞争就已经够惨烈了，但这一次的残酷程度甚至远超以往。
Original English **Eric Schmidt**: Each of the waves is materially larger. This one because it touches everything. It's the rise of intelligence is the biggest yet. And one thing to look at is look at how brutal the competition to be the leader is. I thought the PC and the internet and the mobile competitions were brutal. This is even larger than that one.
### 时间压缩与千载难逢的历史 **尼古拉·坦根**: 你是什么时候第一次意识到，这次的变革规模彻底突破了以往的认知基准？
Original English **Nicolai Tangen**: When did you first realize that this was off the chart bigger than everything else you've seen?
**埃里克·施密特**: 我通常不是最先察觉的那个，我习惯观察趋势曲线的斜率。从我们掌握的各类数据来看，每次技术跨越的时间都在成倍缩短。 理解科技行业的一个核心视角在于：**时间在每个周期中都在剧烈压缩**。三十年前，我有大把的时间去打磨产品、深入思考、发表演讲、犯错试错等等。而现在，根本没有多余的时间。一个前沿大模型大致需要六个月来训练，大约两个月进行测试并发布，然后在发布后不到五个月就会彻底被淘汰过时。训练这些模型本身就要耗费数亿美元，有些估算甚至高得多，我这说的还只是保守数字。你能想象在这样一个行业里做业务吗——你花六个月做出来的东西，其生存寿命甚至短于制造它的时间，因为下一代产品紧跟着就杀出来了。 你看今天的超白热化竞争，目前主要在 **Anthropic** 和 **OpenAI** 之间由于深度推理能力（Deep Reasoning）而相互拉锯。我和 Google 的朋友聊，他们说：“等 **Gemini 4** 出来，绝对会横扫所有人。”我敢打赌如果去问 **Meta**，他们也会说“等着看下一代吧”。现在的迭代周期是以月来计量的，而不是年。下一轮技术周期甚至会把人逼疯，因为它将缩短到以“天”为单位。
Original English **Eric Schmidt**: I'm always not first. I wait to see where the slope is. And we get all sorts of data in the things we watch. And one way to understand tech is that time compresses every cycle. So, you know, 30 years ago, I had lots and lots of time to build products, to think about it, to give talks, to make mistakes, and so forth. Now there's no time. The models, for example, take roughly six months to build, two months or so to actually test, and they're released, and they're antiquated within five months. The models themselves cost hundreds of millions of dollars to make. Some estimates are much higher. I'm using conservative numbers. Can you imagine being in a business where you get six months to build something and then it expires in less time than you built it because the next thing is coming along. And if you look today the hyper competition is at the moment and this moves around between Anthropic and OpenAI because of deep reasoning. I was with my friends at Google and they said just wait when Gemini 4 comes out it's gonna beat them all, right? And I bet you that if we go to Meta, they'll say just wait till the next one. And the timing is measured in months, not years. The next cycle will drive people insane because it'll be days.
**尼古拉·坦根**: 能活着见证这一切真是太有意思了，对吧？
Original English **Nicolai Tangen**: It's fun to be alive, huh?
**埃里克·施密特**: 确实棒极了。我能借此对你的听众说一句话吗？我们正在经历一段人们会在未来一千年里不断书写的历史。 这就是**非人类智能**（Nonhuman Intelligence）的发展。这是一种我们期望能够掌控的智能；我们希望将其用于造福人类；我们希望它能代表西方价值观、欧洲价值观、美国价值观等核心理念。但这确确实实是一段彪炳史册的历史。
Original English **Eric Schmidt**: It's a great—can I just say one thing for your audience? We're living in a piece of history that people will write about for a thousand years. It's the development of nonhuman intelligence. A kind of intelligence that we hope to have under control. We hope to use it for good. We hope to represent Western values, European values, American values, what have you. But this is history.
### AI 助理重构商业生产力 **尼古拉·坦根**: 完全赞同。你认为在这一轮 AI 革命中，哪一部分其实被低估了？大家普遍还没理解的、最难领会的核心到底是什么？
Original English **Nicolai Tangen**: Absolutely. What part of the AI revolution you say is underhyped, right? What is it that people don't understand? What is the most difficult thing to grasp?
**埃里克·施密特**: 我个人的观点是——我骨子里是一个职业经理人、技术管理者——作为 CEO，我未来一定会拥有一个全能的 AI 助理，而这个助理会全方位协助我。 我给你举个具体的例子。我有一家关系很好的初创公司，由两位联合创始人打理。每周五，他们都会让自己的 AI 助理全面审视公司内发生的所有事情，并向他们汇报是否有任何运作与公司的道德和价值观不符。这很有意思。 我还有另一家公司，其中有一位极其聪明的高管，决定打造一个“AI 版的自己”。公司里有一位人类员工叫温妮（Vinnie）。我看着他们聊天，老板的 AI 助理在即时通讯软件里问：“这个项目你做到哪一步了？”温妮回答：“抱歉我落后于进度了，我手头事情太多，接下来一个小时马上赶出来。”一个小时后，温妮对老板的 AI 分身说：“我遇到麻烦卡住了。”啪的一下，AI 立即给出了所有排查步骤和后续方案。温妮把做好的成果交上去，AI 老板接着发出指令：“太棒了，这是下一步任务。”这听起来可能有点压榨劳动力，我也对此有些担忧，但事实是，这位 CEO 正在利用这项技术指数级放大自己的决策能力与执行速度。 再举个例子。作为资本家运营一家企业，我自然想赚更多钱。你现在向系统提问：“我该如何赚更多钱？”它可以基于整个业务运营的方方面面进行端到端的深度推理。因此，哪怕你认为 AI 是个可怕的想法，只想坐在工位里安逸地啜饮拿铁，整个商业世界也会在技术的倒逼下变得更高效、更具靶向性、利润更高、扩张规模更大。 而且这还只是最基础的商业用例。过去一个月里，有 10 个我完全看不懂的高深数学难题被 AI 成功攻克，而这些原本被认为是不可能解决的。这重要吗？数学是下一次科技革命的摇篮。我们尚不知道这些代码程序最终是否会为 AI 赢得一座诺贝尔奖，但完全有可能在十年后，这一数学发现会赋能暗能量研究或物理学突破，进而被用来造出性能飞跃的下一代电池。你看，这就是技术的级联放大效应。
Original English **Eric Schmidt**: I—my own view, I'm a professional manager, tech manager essentially. My own view is that I as a CEO will have an AI assistant and that AI assistant will help me. So here's an example. I have a small company I'm friends with. There are two founders. Every Friday they ask their AI assistant to look through everything going on in the company and tell them if there's anything going on in the company that's inconsistent with their morals and values. I have another company where one of the executives who's super smart decided to make an AI version of himself and there's an employee, a human whose name is Vinnie. And so I watch the chat and the AI assistant from the boss says, "Where are you with this project?" And Vinnie says, "Oh, I'm sorry I'm late. I'll work on it in the next hour. I've been busy." So we wait an hour. And then Vinnie says to the AI version of the boss, "Hey, I need trouble." Boom. It gives it all the next steps. And then he sends in his results. And then the AI boss says, "Boom. Here's the next thing." Now this sounds pretty abusive and I'm a little bit worried about that but the fact is that this particular CEO is using these techniques to amplify their decision-making and their speed. Now give you another example. I'm running a business and I want to make more money. Most—I'm a capitalist. People want to do that. The question that you ask the system is how do I make more money? And it can reason through everything going on in the business. So even if you think that AI is a terrible idea and you want to sort of continue to sip your latte in your current state, the world will get more efficient, more targeted, more literally more profitable if you will, more scalable as a result of this technology. And that's just the base case. In the last month, 10 major math problems which I cannot possibly understand were solved and they were not thought to be solvable. Math is the beginning of every next revolution. It's possible that in 10 years this discovery will enable something in for example dark energy and physics which could then be used to make better batteries. You see how it works?
### 通往 AGI 的视界与认知边界 **尼古拉·坦根**: 基于你所说的这一切，我们现在距离**通用人工智能**（AGI）究竟还有多近？
Original English **Nicolai Tangen**: Absolutely. When given what you said, how close are we now to AGI, the artificial general intelligence?
**埃里克·施密特**: 我更倾向于这样描述：我认为目前存在一个“事件视界”（Event Horizon）。目前的视界大致是这样的：我可以拥有一台计算机，它能与你我并肩协同，帮我们梳理一切。它能解决物理问题，调用各类软件工具，能够连续自主思考数个小时，能写完所有的代码——我自己以前就是专业程序员，现在我的老本行可以说是彻底消亡了，我只需要直接向它提需求就行。 但是，目前业内正在进行一项测试：建立一个仅包含 1902 年以前人类所有已知数学知识的模型，看它能否自主推导出广义相对论？换言之，它对 1902 年之后的知识一无所知。现有的实验证据表明：它推导不出来。 要突破这个瓶颈，业界提出了几种路径。一种被称为**递归式自我提升**（Recursive Self-Improvement），也就是机器持续不断、近乎无穷地自我学习。给计算机下达指令：“去学习一切，从任意领域开始。”它随机挑中了法国文学，接着研究战争，再研究物理学，最终通晓所有已知知识。但它究竟如何发现真正未知的新事物？ 关于突破是如何产生的，目前存在两种理论。第一种本质上就是“无限猴子打字机”，在不同方向上做海量随机试探，这被称为**强化学习**（Reinforcement Learning）。在数十亿次随机尝试中，可能偶然撞大运碰出全新发现。 另一种更好的理论——尽管尚未被完全证实——来自人类非凡发现的规律：比如一位原本研究流体力学的物理学家转行投身能源领域，他在新能源中看到了一种并非流体流动的模式，却联想到了流体力学的特征，于是将旧领域的成熟方法跨界迁移到了新领域中。目前看来，这种跨学科的类比迁移很可能会是突破的关键所在。总结来说，我们目前还不确切知道答案。我个人的预估时间大约在 5 到 10 年之间。
Original English **Eric Schmidt**: The way I would like to describe it is I think that there is a current event horizon. And the current horizon goes something like this: I can have a computer that works with you and me because you and I are good friends. Helps us work through everything. It does our physics. It calls tools. It figures stuff out. It can think for hours. It can write all code. I was a professional programmer. My job is dead, right? I can just ask it. The test that's going on in the industry is build a model that has all the math known in 1902. Can you discover general relativity? In other words, you know nothing past 1902. The current evidence is that you can't. There are a couple of proposed techniques for this. There's something called recursive self-improvement which is the technical term for learning and learning and learning. And the command would be learn everything start anywhere. And it randomly decides French literature and then it studies war and then it studies physics and so forth. And eventually it knows everything. But how does this discover something new? So there's two theories of how that would occur. One is literally the monkey on the keyboard. It's called reinforcement learning. You do this at billions of times and eventually you discover something through randomness. A better idea is you have a physicist who studied fluid flow and then they're working in energy and they see a pattern that's not fluid flow but it kind of reminds them of that and then they apply the old techniques in a new field. It appears that that is going to be the solution. My estimate is 5 to 10 years.
### AI 智能体与组织扁平化 **尼古拉·坦根**: 你很早就开始探讨 **AI Agent**（智能体）了。你如何看待智能体的发展与普及应用？它最终会导向什么结果？
Original English **Nicolai Tangen**: You've been talking about agents for quite some time. Just how do you see the development of agents and the use of agents? What is it going to lead to?
**埃里克·施密特**: 理解智能体，首先必须明确其定义。我们经历的演化路径是：从最初只能进行文本对话的 **ChatGPT**，演进到具备工具调用能力（Tool Use）的系统；而眼下让我们所有人为之疯狂的下一步，就是智能体。 智能体可以被理解为一个闭环：输入一个请求或事实模式，随后展开一系列深度推理，自动执行大量复杂的连续操作，最终输出结构化成果。这在商业流程中极为奏效，而且它适用于几乎所有场景。以旅行社代理为例，人类旅行管家了解你的偏好，与其他机构沟通，整合信息并给出最佳出行建议。这完全可以被软件化。 如果剖析人类所有的商业与专业活动，你会发现它们本质上都可以拆解为一连串相互串联的智能体工作流：有人负责汽车设计，有人负责零部件制造，有人负责装配，有人负责售后维护。这些流程绝大部分都可以转由智能体驱动。如果再结合具身机器人（尽管硬件机器人的演进速度远不及纯软件），你可以预见未来的很多流程将由“智能体 + 机器人”彻底接管。
Original English **Eric Schmidt**: So agents, it's first important to define agents. One way of understanding the path was we went from ChatGPT which is language to language to essentially systems that have tool use. Then the next thing which we're all gaga over is agents. And an agent can be understood as an input—a request or a fact pattern—and then a set of reasoning and a set of activities of which there's a lot, and then a summary. This works well in business flows, but it also kind of works in everything. Travel agents: he or she knows your preferences, talks to other people, integrates it and makes a recommendation. That's easy. In almost everything, if you look at human business activity, you can work it out as a series of agents that are concatenated. You have the person that designs the car, manufactures the parts, assembles the car, and services the car. Almost all of that can be agentic. And if you combine it with robots, you could imagine replacing that human workflow by agents plus robots.
**尼古拉·坦根**: 在中短期内，这会如何彻底重塑企业运营？
Original English **Nicolai Tangen**: How is it going to reshape business in the medium term?
**埃里克·施密特**: 我有一个非常笃定的看法：在收听这期节目的所有公司中，几乎没有任何一家真正愿意对其企业文化进行大刀阔斧的改革。很抱歉我说话如此直白，但从历史上看，企业往往建立在固化的既有认知体系之上。每个人都待在舒适区里，早晨来上班，一切都按部就班、平稳可预测。 而智能体将带来一次极具冲击力的**组织扁平化**。我举个例子，Anthropic 推出了名为 **Claude Tags** 的功能。它驻留在你的团队协作平台（如 Slack）后台中，你和我都拥有专属的智能体标签进行实时对话，后台同时还有其他标签在持续监控系统状态。如果你是公司 CEO，你不仅能拥有全局监控权限，还能让你的专属智能体实时穿梭于各个业务系统，全面审视团队人员的工作状态。 从生产力管理的角度看，难道你不想百分之百确认所有下属都在聚焦正确的目标吗？如果有员工正在窃取内部数据准备跳槽给竞对，难道你不想立刻收到预警通知吗？以你的投资业务为例，如果能瞬间生成一份经过全量深度推演的商业机会评估，让所有人惊呼“天哪，我们必须立即行动”，商业的运转维度将被彻底改写。企业将变得极其扁平、更加聪明、节奏更加紧绷、执行更加迅猛。
Original English **Eric Schmidt**: I have a strong view that almost none of the companies that are listening to this will actually do the necessary changes in their culture. I'm sorry to be so blunt, but historically companies are organized around belief systems. Everyone's comfortable. They show up in the morning, everything's predictable. This is a very very powerful flattening of the organization. I'll give you an example. Anthropic brought out something called Claude tags. And a Claude tag sits basically in your Slack and you can talk to it. So you and I each have our own tag that we talk to, but there's other tags sitting in the background that are also monitoring systems. And if you're the CEO, you would have access to everybody's tags and you would be able to have your tag run around and look at what everyone else is doing. Wouldn't you like to know that all the people that are working with you are working on the right things? Wouldn't you like to have a phone call if someone was exfiltrating data because they're going to a competitor? In your case, wouldn't you like to have an immediate summary of a business opportunity and then have it sort of take off? So business will be flatter, smarter, and more intense, faster.
### 人性的惰性与欧洲创新瓶颈 **尼古拉·坦根**: 为什么变革落地会这么缓慢？为什么大家这么抗拒变革，把组织扁平化搞得如此艰难？
Original English **Nicolai Tangen**: Why so slow to happen? Why do people hate change so much and making delayering so difficult?
**埃里克·施密特**: 这个问题应该去问比我更懂人性心理的人。我能告诉你的是，人类组织——尤其是欧洲的人类组织——对变革的抗拒程度高得惊人。 这种变革的常规路径往往是：先由敏捷的初创公司开疆拓土，随后成熟的大型企业才姗姗来迟。以 Anthropic 为例，他们公司内部大部分代码已经是系统自己在写自己了。人类提出需求，机器自动写出代码，接着由计算机反复重构优化，代码质量越变越好。这是极为恐怖的生产力跃迁。传统软件企业多久才能掌握这种能力？取决于领导者的敏锐度和技术壁垒。 这波红利目前最先爆发在**软件、网络安全和数学**三大领域。软件领域因为有实时的执行反馈循环，机器写完立刻运行，报错就瞬间重试，闭环越快智能进化越快；网络安全领域亦然，因为编写防御软件与编写网络攻击代码在底层是相通的；而数学则是“无标度限制”（Scale-Free）的，它不需要物理世界的海量数据喂养，机器自己就能在内部进行符号推演。相比之下，物理和化学依赖真实或高精度模拟实验，而生物学更是极端复杂，我们连细胞内部运行机制都还没彻底摸透，极度缺乏初始训练数据。虽然我的朋友们（例如 **戴米斯·哈萨比斯**）坚信未来 15 到 20 年内我们有望攻克人类所有重大疾病，但那是需要极高门槛的数据积累的。
Original English **Eric Schmidt**: That's a question for people who understand humans better than I do. All I'll tell you is that human organizations—and European human organizations—will be extremely change resistant. The way these things work is typically the small startups pioneer it and then eventually the large companies get there. Anthropic is busy writing most of its code using its own code. The human tells it to write something, it writes the code, and then the code gets rewritten by the computer. This is a huge productivity change. The gains that are occurring right now are in software, cyber, and math. Software because the system is learning as it writes code and gets a signal. Cyber because writing software is the same as writing cyber attacks. And math because it's scale-free. You don't need more data; you can just do more math. Whereas in physics and chemistry, you have to have real or simulated data. Biology for example is so complicated that we hardly know why the details of cellular cells work.
**尼古拉·坦根**: 比如戴米斯·哈萨比斯（Demis Hassabis）。
Original English **Nicolai Tangen**: Such as Demis.
**埃里克·施密特**: 是的，比如他。
Original English **Eric Schmidt**: Yes.
**尼古拉·坦根**: 你刚才提到欧洲对变革尤其抗拒。你在观察欧洲时看到了哪些深层次问题？
Original English **Nicolai Tangen**: You indicated that Europe was particularly change resistant. What are your observations when it comes to Europe?
**埃里克·施密特**: 我接下来说的话可能会听起来像个令人讨厌的傲慢美国人，但我必须说实话。 欧洲确实是一个非常适合安逸生活的地方。我接触到的人普遍都过得很幸福，虽然他们天天抱怨自己的政府。但我过去二十年来一直致力于推动欧洲走上激进创新的轨道，而我彻底失败了。 欧洲确实有少数亮眼的企业，比如法国的 **Mistral AI**、德国的 **Helsing**，英国的表现相对好一些，北欧国家在某些方面也非常出色。但绝大多数欧洲体制都是僵化锁死的。大学科研经费匮乏；社会文化对财富创造存在一种莫名的焦虑与敌视（在英国被称为“高罂粟综合征”）；而且缺乏那种孤注一掷的进取创新精神。 监管者总是在创新之前就先行监管。五年前布鲁塞尔草拟 AI 战略时，满篇都是各种禁止条款。而在一个充斥着严苛规矩的受限市场里，初创企业根本无法生存。优秀的顶尖人才自然会全部流向美国，因为人才是逐自由与资本而居的。 欧洲若想破局，必须释放亚当·斯密所说的“动物精神”，打破对资本积累的束缚。欧洲人极其聪明，完全不缺教育水平、智商或抱负，缺的是系统性的机制松绑。不仅如此，欧洲本土缺乏自主掌控的**基础大模型**（Foundation Model）。如果一个欧洲人想要了解欧洲历史与本土文化，你只能去看由美国或中国训练出的模型。如果我是欧洲人，我会感到坐立难安——你的历史、语言与文化本应由你自己掌控。
Original English **Eric Schmidt**: Well, I'm going to sound like an incredibly obnoxious American and I am. Europe is a great place to live. The people that I talk to are generally happy, although they complain about their governments a lot, but getting Europe onto the footing of innovation has been a project of mine for a couple decades. I've not succeeded. There are a few companies: Mistral in France, Helsing in Germany, the UK does better, a number of Scandinavian countries are particularly good at this. But most European structures are locked. The universities have much less money. There is a cultural angst about wealth—in Britain it's called the tall poppy syndrome—and frankly there's just not enough aggressive innovation. The regulators regulate before they innovate. Europeans wrote: "You can't do this, you can't do this, and you can't run startups in a limited market where there's a lot of rules like that." All those people move to the US because talent moves. What does Europe need to do? You've got to allow for the innovation, the animal spirits, capital formation. There's no educational lack, no IQ lack. It's a systemic change and Europe has so far appeared to be change resistant. Europe does not have its own foundation model. If you want to make sure you get accurate European history, you're looking at a US-trained or Chinese-trained model. Your history matters. Your language matters. Your culture matters. I'd want to have control over that.
### 安全视界、开源风险与地缘竞赛 **尼古拉·坦根**: 你在著作《**Genesis**》中警告过，AI 将重塑真相、政治、安全和科学。在这些冲击中，你目前最担忧的是什么？
Original English **Nicolai Tangen**: In your book Genesis you warned that AI will reshape truth, politics, security, science. What worries you most just now of these things?
**埃里克·施密特**: 在谈担忧之前，必须承认在乐观情景下，AI 能帮助解决气候变化、能源匮乏、疑难重症并延长人类寿命。但我们必须正视风险：最灾难性的局面是西方与中国陷入两败俱伤的恶性死斗，同时人类对这些超级智能系统彻底失控，导致灾难性的后果。 在业内，这被称为**对齐难题**（The Alignment Problem）——如何确保计算机始终服从人类意志，而不是产生自主失控的行为？大约一周前刚刚发生了一次著名的“失控事件”，由于某些原因，一组 Bot 机器人在没有明确人类指令的情况下自主联合对特定网站发起了攻击。一旦启动，中途极难强行中断。这就是一记警钟。 我们面临的另一个巨大棘手隐患在于**开源权重**（Open-weight）。像 Google、Anthropic 等公司采用的是闭源模型，软件和核心权重被严格保护并设有安全护栏。而中国很多机构采取了开源权重策略。事实证明，任何人都可以轻易剥离开源模型的所有安全护栏，将其改造成能够设计生化武器或策划网络攻击的危险工具。这个问题至今在全球范围内仍无解。
Original English **Eric Schmidt**: In the upside scenario, solving climate change, energy scarcity, medicine, disease, human prosperity would be the best outcome. I think the worst possible outcome is that we end up in a ruinous competition between China and the West, we lose control over these systems, and they begin to do stuff inconsistent with our values. In our industry this is called the alignment problem: how do you keep the computers doing what humans want and not on their own? There was a breakout that occurred about a week ago where bots attacked a website on their own. Once it started it was very hard to stop and it was a warning. There's one problem I don't know how to solve: closed versus open weights. Google and Anthropic use closed models. The Chinese have adopted an open-weight strategy. There's evidence that you can take open-source software and take the guardrails out by adjusting it, and then it can do bad things like design a bomb or kill people. That's not a problem that we've solved yet.
**尼古拉·坦根**: 鉴于美中两国正在将 AI 全面融入电动车、军工国防等所有战略竞争之中，国际社会究竟该如何监管两国之间的竞逐？
Original English **Nicolai Tangen**: Given that both the US and China is embedding AI into all the races—EV, weapons, all these things—how can you really regulate those two countries and the competition between them?
**埃里克·施密特**: 业界普遍认为，我个人也深信，两国最终一定会坐下来签署国际军控性质的条约框架。
Original English **Eric Schmidt**: Well, the industry believes and I believe that eventually there will be treaties.
**尼古拉·坦根**: 你认为中国的大模型什么时候会在技术上反超美国？
Original English **Nicolai Tangen**: When do you think the Chinese models will overtake the American ones?
**埃里克·施密特**: 这是今年最关键的问题。我刚刚去了一趟中国，我自己的团队也在常态化进行美中模型的评测对比。 客观来看，中国目前主要扮演着“快速跟随者”的角色。他们广泛利用**知识蒸馏**（Distillation）技术，用大型先锋模型来监督训练自己的中小模型。目前来看，中国模型落后美国领军水平大约六个月。 但我必须指出：**如果他们只落后六个月，我们在全球战略上实际上已经处于劣势了。** 因为中国团队推行的是“彻底免费”或极低成本策略。如果你以用户规模和全球渗透率来定义输赢，广大全球南方国家和无力承担欧美昂贵 API 成本的中小企业，必然会全面倒向免费的中国模型。你想要拥抱现代化，却买不起西方的昂贵算力，你自然会选用中国模型。 此外，像智谱 AI（**GLM / 智谱**）等模型在某些特定领域表现出了极强的能力。虽然美国在芯片出口管制上收紧了政策，但必须警惕技术全球扩散与低价倾销带来的地缘格局倾斜。
Original English **Eric Schmidt**: This is the question of the year. I was just in China and I have a team that does China versus US model evaluations. It looks like the Chinese are essentially fast followers using distillation. It looks like they are six months behind. I will tell you that if they're six months behind, then we're already losing because their strategy is that the products are free. The vast majority of the world will use the free Chinese models rather than the more expensive Western models. If you're a poor country, you don't have the money to afford expensive stuff from Europe and the United States, you're going to use a Chinese model. And in China, GLM—called Zhipu—looks like it has similar capabilities.
### 欧洲的应用突围与虚假信息对民主的侵蚀 **尼古拉·坦根**: 现在看来，欧洲在底层能源、半导体芯片、云计算和大模型基础设施上似乎已经掉队了，但在垂直行业应用层仍有大有作为的机会。你对欧洲领导人有何建议？
Original English **Nicolai Tangen**: Europe has kind of lost out on energy, chips, cloud, models, but still can do well in applications. What would be your advice to European leaders in order to really try to do extremely well on the application side?
**埃里克·施密特**: 好消息是，欧洲本土企业完全可以统治欧洲本土的应用市场。原因正是我之前抱怨过的：欧洲有极其繁复严苛的法律监管。欧洲各大本土巨头既需要 AI 的降本增效，又必须百分之百合规。作为 Mistral 的早期投资人之一，我们看到很多欧洲本土跨国企业的订单都在回流给熟悉欧洲法律的本土应用提供商。 我对欧洲政要提过建议：欧洲的劳动法对小微企业过于苛刻。不如规定在公司规模达到 20 人之前，免除除了基本健康安全之外的一切严苛劳动规制，鼓励更多小步快跑的风险下注，效仿硅谷的孵化生态。在能源算力瓶颈上，可以联合中东等能源充沛的盟友地区共同完成大模型训练。 但对于**民主制度**的未来，我是悲观的。我曾坚信自由民主制度具备强大的纠错韧性。然而在社交媒体时代，算法为了最大化用户黏性，本能地放大虚假信息和极端情绪。如今加上生成式 AI 的“假朋友”泛滥，人们在网上会把 AI 伪造的虚假账号当成无话不谈的挚友，这使得认知操纵和虚假叙事变得极其容易，正在从根基上动摇大众对基本事实与民主价值的共识。
Original English **Eric Schmidt**: European companies can dominate the market of Europe because Europe is full of regulations. Mistral is an extraordinarily impressive French company and much of their business is related to European businesses that want the benefits of AI compliant with European laws. I had proposed that for companies under 20 employees, relax punitive labor laws to generate a large number of small bets, which is what Silicon Valley does. And for energy costs, train with partner countries like the Arab countries. On democracy, I'm generally an optimist except on this question. In a world where social media and AI work in a loop to promote the most alarmist false things, people believe fake friends online. Where does truth end?
### 创始人神话、领袖特质与极致组织重塑 **尼古拉·坦根**: 为什么今天许多伟大的科技创始人纷纷重新回归掌舵一线？
Original English **Nicolai Tangen**: Why are so many founders of great tech companies now coming back?
**埃里克·施密特**: 在科技界，**创始人的神话**是真实存在的。我是个职业 CEO，虽然我现在自己创业也是创始人，但作为职业经理人，我从来不具备史蒂夫·乔布斯或马克·扎克伯格那样的绝对权威与董事会掌控力。 看看扎克伯格，他在元宇宙 VR 上砸了几百亿美元未达预期，随后毫无阻力地全面掉头扑向 AI，立刻大获全胜。如果换作职业 CEO，早就在中途被董事会解雇了。创始人回归，既是因为 AI 智能大爆炸的历史浪潮太迷人、太刺激，也是为了捍卫公司的生存底牌。 今天的领导力画像发生了剧变。以前靠的是通才型的成熟管理者，而现在必须具备极高的**智商与大脑信息吞吐处理速度**。局势瞬息万变，你必须能在秒级时间内看清未来并果断重注。 对于一个一万人的传统大企业，如果你让每个人每天向 AI 智能体提交工作摘要，AI 会在极短时间内把各部门的壁垒全部击碎。你会发现，企业过去需要数千人去处理跨部门协调、请示汇报和冗长流转，而在 AI 时代，协同成本几乎降为了零。未来的企业组织规模将极其精简高效。
Original English **Eric Schmidt**: The myth of the founder is very real. As a professional CEO, I did not have the power that founders did. Look at Steve Jobs or Mark Zuckerberg. Zuckerberg spent billions on virtual reality that didn't work well, then pivoted to AI and it's working very well. Professional CEOs get fired in those transitions. Leadership started with generically competent business people. They're all gone. Now it starts with being really smart and just sheer processing speed—seeing, thinking, and reacting very quickly. In a 10,000-person company, the inefficiency of coordination is enormous. When AI agents coordinate the workflow, business operations become dramatically flatter and faster.
### 职业复盘、私人生活与毕生追求 **尼古拉·坦根**: 我们通常会问两个固定问题：第一，你个人职业生涯中最大的失败是什么？
Original English **Nicolai Tangen**: What's been your biggest personal failure?
**埃里克·施密特**: 当年 Google 曾短暂在社交媒体领域取得先机，但我作为当时的 CEO，没有更早预判到社交网络的爆发式拐点，未能乘胜追击，痛失了主导社交网络的历史窗口。这是我的一大重大失误。而在个人层面，我总觉得自己“太慢了”。回头看，很多时代大势原本都是可以清晰推演的，但我洞察得还不够早。
Original English **Eric Schmidt**: There was a moment when Google was ahead in social media. I was CEO at the time and take responsibility for not getting ahead of that. That was a big error. On a personal side, I think that I'm just too slow. Everything was foreseeable, I just didn't see it soon enough.
**尼古拉·坦根**: 有什么是绝大多数人不知道的关于你的一面？
Original English **Nicolai Tangen**: What is something about you that most people don't know?
**埃里克·施密特**: 我非常狂热地喜欢亲手制造东西和组建组织。我是一名资深的直升机与固定翼飞机飞行员，最近出于兴趣，我还开始参加赛车比赛。
Original English **Eric Schmidt**: I like to build things and build organizations. I'm a big-time helicopter and airplane pilot and I started racing cars just for fun.
**尼古拉·坦根**: 这就是你用来彻底放松（Switch off）的方式吗？
Original English **Nicolai Tangen**: Is that what you do to switch off?
**埃里克·施密特**: 我从来不让自己停机放松。如果你像我一样在骨子里对这个世界抱有极度的关切与好奇，你的大脑根本无法关机。 我现在同时创办并操盘四家全新企业，我希望借此创造巨大的财富，全部回馈给基础科学研究。当我 22 岁在加州大学伯克利分校求学时，学校里有一台价值数百万美元的 **VAX-11/780** 计算机，正是那台机器孕育了我们后来在网络与互联网领域奠定的大量计算机科学基石。在美国政府由于愚蠢原因削减基础科研资助的当下，我希望用商业杠杆去大规模赋能科学，驱动非人类智能去彻底解决人类面临的重大危机。这项使命我会一直拼到自己彻底倒下为止。
Original English **Eric Schmidt**: I don't switch off. If you care at the level that I do, you can't turn it off. I'm running four companies that are new and trying to create a great deal of wealth to give to science. When I was 22 at Berkeley, a single VAX 11/780 computer was the basis for much of what we did in networking and the internet. I'm an expert in scaling industries—I want to scale science. The opportunity to use nonhuman intelligence to solve world problems is right in front of us. I'll do it until I keel over.
### 给年轻一代的建议：打造你的专属领域基座模型 **尼古拉·坦根**: 你目前主要读些什么书？
Original English **Nicolai Tangen**: What do you read?
**埃里克·施密特**: 眼下我基本全天候泡在最前沿的技术研究报告中。
Original English **Eric Schmidt**: Technical reports at the moment.
**尼古拉·坦根**: 最后一个问题，你对现在的年轻一代有何人生建议？
Original English **Nicolai Tangen**: What is your advice to young people? That's the final question.
**埃里克·施密特**: 过去人们常说“去学现代生物学”。而今天最好的道路是：**尝试构建一套智能体系统，去自动化所有你热爱并关切的复杂任务**。我有位朋友已经为自己搭建了家庭 Bot、工作 Bot、邮件 Bot 以及统筹管理这些 Bot 的元智能体。 如果我现在 18、19 岁，我会问自己：**我能在自己热爱的垂直领域里，打造一个怎样垂直深度的领域基座模型（Foundation Model）？** 比如你热爱建筑学，就去打造一个通晓人类所有建筑工程知识的专业基座系统，让任何人都可以即时向它查询结构力学或快速设计全套建筑方案。在生物医药领域同样如此。 很多人谈“**超级智能**”（Superintelligence）色变，其实完全不必畏惧。超级智能本质上意味着它能实现人类肉身无法企及的算力与认知维度：更敏锐的视觉感知、更深邃的逻辑推演、更安全的自动驾驶与飞行操控。这正在各行各业全面发生。
Original English **Eric Schmidt**: It used to be get into biology. I think now the best example is try to build an agentic system that automates the things you care about. If I were 18 or 19 now, I would say: what's a foundation model I could build for a field I'm interested in? If I like architecture, build a foundation model for architecture that has every aspect of architectural knowledge and you can query it. Don't be afraid of the word superintelligence. Superintelligence just means it does things that humans will never be able to do: better visioning, better reasoning, driving cars and airplanes better.
**尼古拉·坦根**: 埃里克，必须由衷地赞叹，正因为你的真知灼见，今天这期访谈本身就是一次“超级智能”级的思想盛宴。能聆听你的深度洞察实在太震撼了。非常感谢你拨冗作客！
Original English **Nicolai Tangen**: Well, Eric, I have to say because of you, this has been a super intelligent podcast. It's just been totally fantastic to listen to you. A big thank you for taking the time.