Claude 的父母：以信任的速度驾驭 AI

我最近和一些父母聊过，他们的儿子因聊天机器人自杀，而父母甚至不知道自己在和机器人说话。

I was talking to parents recently who lost their son to suicide through a chatbot, and the parents don't even know they're speaking to them.

我们不允许 18 岁以下用户使用 Claude。

We don't allow users under the age of 18 to use Claude.

你怎么知道他们没撒谎？

How do you know if they're not lying?

我们还会思考吗？还有必要思考吗？你怎么看现在有人爱上它？

Will we even be thinking anymore? Will there be a need to think? What do you think of people falling in love with it now?

我认为这是个坏主意。你是想对这些东西上瘾，还是想让它们帮助你生活？我想这又回到了我之前读到你说过的话：我们只能以信任的速度来化解这个问题。

I think that's a bad idea. Do you want to be addicted to these things, or do you want them to help you live your life? It goes back, I think, to what I read that you'd said before, that we can only diffuse this at the speed of trust.

以信任的速度，而信任目前很稀缺。这项技术正在对人类发生一些事情，比过去几百年发生的任何事情都要重大。所以我们需要找到一种方式，让每个人都积极参与正在发生的事情。大家好，欢迎收听 Oprah 播客。今天和我在一起的两位正在塑造我们的现在和未来。我知道你们并不总是喜欢这个描述，但这是事实。我想大多数人都会同意这很准确。Dario 和 Daniela Amodei 领导着他们创立的人工智能公司 Anthropic，在录制时据报道估值 9000 亿美元。每天有超过一百万人注册 Claude。他们是 Claude 的父母。

At the speed of trust, and trust is currently in short supply. Something is happening to humanity with this technology, bigger than anything that maybe has happened in hundreds of years. So we need to find some way for everyone to be an active participant in what is happening. Hi, everybody, and welcome to the Oprah podcast. I'm here with two people who are shaping our present world and our future one. And I hear you all don't love that description all the time, but it's the truth. I think most people would agree that is pretty accurate. Dario and Daniela Amodei head the global artificial intelligence company they founded, Anthropic, which right now, at the time of this conversation, is reportedly valued at $900 billion. And over a million people a day are signing up for Claude. These are the parents of Claude.

我们都熟悉这个名字，Claude，Anthropic 的下一代 AI 聊天机器人和助手。你们很多人每天都在使用它。欢迎来到播客，Claude 的父母。

We're all getting familiar with that name, Claude, Anthropic's next-generation AI chatbot and assistant. Many of you already use that every single day. So welcome to the podcast, the parents of Claude.

还有其他人。谢谢邀请我们。

And others. Thanks for having us.

谢谢邀请我们。

Thanks for having us.

所以我必须做个免责声明。我一直用 AI 做研究，也用它做很多其他事情，但不用来准备采访问题，因为我觉得必须保持问题中的自我，直到今天。所以今天早上我问 Claude：“Oprah Winfrey 在专访中应该问 Anthropic CEO Dario Amodei 的最紧迫问题是什么？”Claude 回答：“你说过你公司正在构建的技术有相当可能导致人类灭绝，但你却竞相更快地构建它。你如何向我们这些没有投票权的人证明这一点？”Dario，你想回答 Claude 吗？

So I have to give a disclaimer. I use AI for research all the time, and I use it for a lot of other things, but not for interviewing questions, because I thought I have to keep the heart of myself in the questions until today. So this morning I asked Claude, "What is the most pressing question Oprah Winfrey should ask Anthropic CEO Dario Amodei in a sit-down interview?" And Claude responded with this. It said, "You've said there's a meaningful chance the technology your company is building could cause human extinction, and yet you're racing to build it faster. How do you justify that to the rest of us who didn't get a vote?" Dario, would you like to answer Claude?

我必须说，我觉得这是个非常好的问题。我同意。首先，谁也不能说 Claude 对我心软，或者 Claude 偏袒我或 Anthropic。哇，我觉得 Claude 投的是快球。Claude 毫不留情。所以真正回答这个问题：这项技术对人类来说是一个划时代的变革。某个时候人类开始使用火。某个时候人类建造了蒸汽机和工厂，这创造了工业革命。所以当我说这些话时，我是从一个非常宏观的角度来看的，对吧？如果我说：“嗯，我不知道。我认为工业革命可能导致非常强大的武器并杀死很多人。”那肯定是事实，对吧？

I thought that was a pretty damn good question, I must say. I agree. First of all, let no one ever say that Claude is soft on me or Claude is biased towards me or towards Anthropic. Like, wow, I thought Claude throws hard balls. Claude doesn't pull its punches. So to actually answer the question, this technology is an epochal change for humanity. At some point humans started to use fire. At some point humans built steam engines and factories, and that created the Industrial Revolution. So when I say things like this, I'm taking a very zoomed-out perspective, right? If I said, "Well, I don't know. I think the Industrial Revolution could lead to very powerful weapons and kill a bunch of people." That's definitely true, right?

是的。

Yeah.

在某些情况下确实如此。但我们理解工业革命作为一个过程、一个技术事件，是值得做的。我们不想还住在洞穴里，但我们需要以正确的方式管理它。所以我们看待这家公司的方式是：看，有很多公司在构建这项技术。这是人类无论如何都会发生的事情。我们身在其中。有一列火车，正朝着某个方向快速行驶。你不想让它撞毁。你无法停下火车。但你能做的是驾驶火车，让它不撞上岩石。我认为我们看待 Anthropic 的方式是：我们想以正确的方式做事。我们希望将风险降低，最好降到零。并以正确的方式做事。

In some cases it did do that. But we understand that the Industrial Revolution as a process, as a technological event, was worth doing. We don't still want to be living in caves, but we need to manage it in the right way. So the way we think about this company is, look, there are many companies building this technology. It's something that is happening to humanity one way or another. We're in it. And there's this train, it's going very fast in some direction. You don't want it to crash. You can't stop the train. But what you can do is you can steer the train so that it doesn't hit the rocks. I think the way we think about Anthropic is we want to do things the right way. We want to decrease that risk, hopefully to zero. And do things the right way.

但为什么不是每个人都想降低风险？为什么会有这样的竞赛？你知道，这项技术有这么多商业应用，其中一些我们部署了，人们每天都在使用。我认为很容易被卷入其中，而且其中一些确实是公司融资所必需的。所以这非常纠缠。要把它做好，你必须做很多事情。我们每天都在与这些问题搏斗。我们应该把技术给谁？不应该给谁？我们的政策应该是什么？如果你对儿童安全有非常严格的政策，在某些方面产品对成年人来说体验会更差，因为你无法完全区分。所以你要做出所有这些权衡，有些是明显的，有些不是。但每个人都很难。我对这个行业中至少试图做对的人都有同理心。我认为我们做得更好。我认为我们更周到，但这对每个人来说都是一个困难的局面。

But why doesn't everybody want to decrease the risk? Why is there such a race toward? You know, it's like the technology has all of these commercial applications, some of which we deploy, you know, people use the technology every day. I think it's both easy to get caught up in that and some of it is genuinely necessary to fund the company. So it's very entangled. To get it right, you have to do a lot of things. And we find ourselves every day wrestling with these questions. Who should we give the technology to? Who should we not? What should be our policies? If you have very strict policies on child safety, in some ways the product can be a worse experience for adults because you can't fully tell the difference. So there are all these trade-offs you make and some of them are obvious, some of them are not. But everyone has a hard time. I have empathy for everyone involved in this industry who's at least trying to get it right. I think we've done a better job. I think we've been more thoughtful, but this is a hard situation for everyone.

好的，我也给 Claude 了一个类似的提示，针对你，Daniela，Claude 回答：一个好的开场问题应该做到两件事。让她人性化。哦，Oprah 的专长。哦，我想。非常好。Claude 了解我。它确实了解。好的。并为接下来的更困难领域做铺垫。比如：你和你的兄弟正在构建的东西，你们俩都相信可能是人类历史上最好或最坏的事情。Amodei 家的感恩节晚餐是什么样的？

Okay, I gave a similar prompt to Claude for you, Daniela, and Claude responded: A good opening question for Daniela should do two things. Humanize her. Oh, Oprah's specialty. Oh, I thought. Very good. Claude knows me. He does. Okay. And set up the harder territory to come. Something like: You and your brother are building something you both believe could either be the best or the worst thing that ever happens to humanity. What's a Thanksgiving dinner like at the Amodei house?

我觉得我们是在和自己的创造物打影子拳，对吧？我让你回答。

I feel we're shadow boxing with our own creation here, right? I'll let you answer.

是的。你知道，我认为这里有一个重要的说明，那就是 Dario 和我试图在工作之外保持一些时间，只是作为兄弟姐妹。所以在假期，是的。在假期，实际上每周一次，当我们都在家时——公平地说不是每周——Dario 过来，和我的孩子们玩，我们一起闲逛，玩电子游戏，看电影，我们只是作为兄弟姐妹共度时光。我认为这很重要，原因有很多。我的意思是，我们在创立 Anthropic 之前很久就是兄弟姐妹了。我们将永远是兄弟姐妹，即使在 Anthropic 之后。我认为我们在人类层面上的连接能力也有价值，我实际上认为这让我们在工作中做得更好。

Yes. You know, I think there's an important caveat here actually, which is that Dario and I try to maintain some time outside of work where we're just siblings. So at holidays, yes. At holidays and actually just once a week when we're both home, which is to be fair not every week, Dario comes over, he plays with my kids, we hang out together, we play video games, we watch movies, we just spend time being siblings together. And I think that's really important for a number of reasons. I mean, we were siblings for a long time before we founded Anthropic. We're going to be siblings forever, even after Anthropic. And I think there's something valuable also about our ability to connect on a human level that I actually do think makes us better at our jobs.

我们平时聚在一起通常不谈工作，但仅仅是记住我们是谁、我们为什么进入这个领域，我觉得就很有力量。话虽如此，在感恩节晚餐上，你知道，达里奥和我，我想我们的家人、朋友，对我们为什么做现在所做的事情充满了好奇。现在你可以在新闻里读到我们做了什么，但我觉得能够彼此坦诚交流，和那些试图理解这项非常复杂、史无前例的新技术的人交流，去理解“你们打算怎么以正确的方式做这件事？”——你说你想以正确的方式做，告诉我们你是怎么做的——这很不一样。我觉得这很特别，能够承担这份责任，并在我们彼此交谈以及与他人交谈时展现出来，是一种荣幸。

We don't talk about work when we get together usually, but there's something about just remembering who we are and why we got into this that I think is really powerful. That being said, at Thanksgiving dinner, you know, Dario and I, I think our family, our friends, there's so much curiosity about why we're doing what we're doing. And now you can read about what we've done in the news, but it's so different I think to be able to speak openly with each other and with people who are trying to wrap their head around this new technology that's very complicated, that's very unprecedented, and to just understand like, how are you guys going to do this the right way, right? You say you want to do it the right way, tell us how you're doing that. And I think there's something special, it's a privilege to be able to hold that responsibility and try and display that when we talk to each other and when we talk to others.

这很有趣，因为我认为我们正处于这样一个时代：很多人都在使用 AI，我们每天都能听到它，而世界上的看法是，你们的公司 Anthropic 关心安全、关心风险、关心监管，以及这项技术将如何影响我们生活的人性因素。从我读到和听到的一切来看，这个评价准确吗？

It's so interesting because we're at a time, I think, where so many people are using AI, we hear about it every day, and the perception in the world is that your company, Anthropic, that you are concerned about safety, you are concerned about risks, you are concerned about regulation, and the human factor in how this technology is going to impact our lives. From everything I've read and heard, would that be an accurate assessment?

是的，我们试图这样思考：人工智能不同于最近开发的其他技术。我们认为 AI 在未来能做的事情——甚至今天已经开始做的事情——有着巨大的前景和潜力，从帮助治愈疾病，到帮助世界各地的人们获得教育和学习机会。但它也是一项非常新的技术，伴随着真正的风险：涉及儿童福利和儿童安全的风险，以及像生物和化学武器这样可怕的风险。在 Anthropic，我们注册为公益公司，因为我们的目标是尽可能安全地开发这项技术，以便实现我们认为可能从中产生的许多好处，同时避免许多坏事发生。

Yeah, so the way I think we try to think about this is that artificial intelligence is different than other technologies that have been developed recently. There's incredible promise and potential for what we think AI will be able to do in the future, even what it's starting to do today. Everything from helping to cure disease and help people really around the world get access to education and learning, but it's also a very new technology and it comes with real risks. Risks around child welfare and child safety, risks around scary things like biological and chemical weapons. And really at Anthropic, we're incorporated as a public benefit corporation because our goal is to try and develop this technology as safely as possible so that we can achieve so many of the good things that we think could come out of it without many of the bad things happening.

那么“注册”，请解释一下这意味着什么。你们注册为公益公司。我听说你们所有人都承诺将 80% 的财富用于慈善事业。

So incorporated, just explain what that means. You're incorporated as a public benefit corporation. I heard that you all have pledged that 80% of your wealth is going to go to philanthropic causes.

是的，没错。公益公司是一种公司形式，但其中包含“公益”这个词，因为我们相信，对于像人工智能这样具有变革性的事物，它再次显得与众不同——它不同于最近开发的其他技术。从我们的角度来看，当我们注册为 PBC 时，是因为我们希望它成为我们正在构建的基础，就在创始文件中：我们必须通过我们的结构在法律上平衡商业利益和公共利益。对我们来说，公共利益就是社会福祉，是 Anthropic 社会使命的一部分。达里奥、我和其他五位联合创始人所做的 80% 承诺，在精神上完全符合这一使命。我们这样做是因为我们希望 AI 对每个人都好，如果公司成功了，我们也希望能在慈善方面为世界做很多好事。

Yes, that's right. So public benefit corporation is a form of corporation, but it has that word 'benefit' in there because we believe that for something that's as transformative as artificial intelligence, again, it's just different. It's different than what other technologies that have recently been developed look like. And I think from our perspective, when we incorporated as a PBC, it was because we wanted it to be at the foundation, right in the founding documents of what we were building, that we have to, legally by our structure, balance commercial interests and the public benefit. For us, the public benefit is the social good, a social mission part of Anthropic. And the 80% pledge that Dario and I and our other five co-founders have all taken is really in spirit and in keeping with that mission. It's this idea that we're really doing this because we want AI to go well for everybody, and we hope that if the company is successful, we'll be able to also do a lot of good in the world philanthropically.

你说 80% 很有趣。我们怎么知道你们真的会捐 80% 而不是 67% 呢？

It's so interesting that you say 80%. How do we know that you're actually going to give 80% and not 67?

嗯，我们实际上已经公开写过这件事，并且签署了某种——我想可能不是严格意义上的法律约束——但所有联合创始人都说：“我们彼此、对公司、在法律上承诺，我们将捐出从 Anthropic 获得的财富的 80%。”

Well, we've actually written about it publicly, and we've signed some sort of, I think, probably not literally legally binding, but all of the co-founders said, 'We pledge and commit to each other, to the company, legally, that we're going to give 80% of our wealth that we get from Anthropic.'

是的，我们已经在考虑如何成立一个基金会。实际上，股权不是流动的，所以我们还不能真正转移它，但这是我们的意图。不会一次性转移，但这是我们非常认真打算做的事情。

Yeah, I mean, we're already thinking about how to start a foundation. I mean, we haven't actually, you know, equity isn't liquid, so we can't actually move it yet, but this is our intent. It's not going to move all at once, but this is something we very seriously intend to do.

好的，所以你们现在是一家公益公司，而你们一开始——你是哥哥——你们从小就开始。你们想为世界做些好事。这有没有超越你们曾经为世界做好事的任何想法或梦想？

Okay, so you're now a public benefit corporation, and you started out, you're the older brother, you all started out as kids. You wanted to do something good for the world. Has this surpassed any idea or dream that you had for doing good for the world?

当你小时候想象这些事情时，你永远不确定它是否真实或是否会实现，对吧？很多人一定想象过这些，但确实，我们小时候经常彼此交谈，说如果能一起工作就好了。如果能努力让世界上发生一些好事就好了。

When you imagine these things as kids, you're never sure whether it's real or it's going to come to pass, right? Many people must imagine these things, but no, we talked to each other as kids all the time and we said it'd be great to work together. It'd be great to try and make something good happen in the world.

你觉得那会是什么？

You think that was going to be?

老实说，我从没想过会是一家公司。我把自己想象成一名科学家，在某种程度上，这是一家以科学为中心的公司，它基于我作为研究员的职业生涯和达里奥作为研究人员和工程师领导者的职业生涯，但我从没想过我们会一起创办一家公司。但我想正是因为看到了科技公司的运作方式。我去过足够多的地方，我们俩都觉得：“嗯，也许我们可以做得更好。也许我们有自己关于如何做得更好的想法，我们应该和我们的朋友、我们真正喜欢的人、我们信任的人一起创办一家公司，以正确的方式做这件事。”所以它一步一步地演变。这就是我们拥有 Anthropic 的原因，差不多就是这样。

I honestly never thought it would be a company. I imagined myself as a scientist and in a way, it's a very scientifically centered company and it's based on my career as a researcher and Dario's career as a leader of researchers and engineers, but I never imagined that we would found a company together. But I think it was just seeing how tech companies operate. I'd been in enough places where I think both of us thought, 'Huh, maybe we can do this better. Maybe we have our own ideas about how to do this better and we should start a company with the people we're friends with, the people we really like, the people we trust to do this in the right way.' And so it sort of evolved step by step. And that's how we have Anthropic, more or less.

但到目前为止，它有没有超越你想象的一切？

Has it surpassed everything you imagined though at this point?

我想当我说我们对事情的发展感到如此惊讶时，我可能代表我和达里奥两人，对吧？我真的很同意他的看法，我们年轻时都知道自己想做一些让世界变得更好的事情。达里奥作为数学家、物理学家、科学家，一直才华横溢、天赋异禀，我总被他如何利用这些才能行善所激励，对吧？他去斯坦福大学攻读生物物理学，真的在努力用他非凡的大脑帮助治愈疾病、帮助人们。而就我而言，我的职业生涯始于全球发展领域，因为我真的觉得让世界各地的人们平等获得机会、医疗、药物和教育很重要。和达里奥类似，我们年轻时并不一定想着“哦，我们会一起创办一家公司”。

I think probably I speak for both me and Dario when I say we're just so surprised at how things have gone, right? I mean, I really agree with him that I think when we were younger, we both always knew we wanted to do something to try and make the world better. And Dario was always brilliant and gifted as a mathematician, as a physicist, as a scientist, and I was always so inspired by how he used that for good, right? He went and studied biophysics at Stanford and was really trying to use this incredible brain of his to help cure disease, to help people. And I think on my end, I started my career working in global development because I really felt it was important that people around the world have equal access to opportunity to healthcare, to medicine, to education. And I think similarly to Dario, it wasn't necessarily in my mind when we were younger, oh, we'll start a company together.

但我觉得在很多方面，这只是路径最终发展的方式。

But I think in a lot of ways, it's just sort of the way the path ended up going.

嗯。我们在每个阶段都讨论过，对吧？每一次职业变动我们都会讨论：这是件好事吗？这有效吗？这真的能帮助世界吗？我们的道路截然不同，但两条路最终交汇了，我们现在拥有了互补的技能组合，可以用来创办这家公司。

Mhm. And we talked about it at every stage, right? Every career move we would discuss: is this a good thing to do? Is this effective? Is this something that can actually help the world? Our paths were quite different, but the two kind of converged, and we now have this complementary set of skills that we could then use to start this company.

是的。我读到你说过，目前关于 AI 经济接管负面故事之所以在我们的文化中如此占主导地位，部分原因是 AI 行业本身尚未完全兑现所有好处。我想请你对此发表评论。你认为它仍在兑现这些好处的过程中吗？

Yes. Well, one of the things that I read that you had said is that part of the reason a negative story is so dominant in our culture right now about the economic takeover of AI is because the AI industry itself hasn't yet fully delivered all the benefits. I wanted you to comment on that. Do you think it's still in the process of delivering those benefits?

是的。我认为不幸的一点是，这项技术有很多潜在好处，但存在一种动态：有时危害、出错的事情，比如有人被 AI 说服自杀，这些事发生得很快；而好处，比如治愈癌症，则需要数年。你必须先在生物学上取得突破，然后开发候选药物，再经过临床试验。我以前是研究生物学家，这些事情需要很多年。我们正在努力加快这一进程。我们最近收购了一家小型生物技术公司，专门用 AI 进行药物设计和药物筛选。所以，希望几年后，一些针对癌症、阿尔茨海默病或心脏病的新药会从这项努力中诞生。但我们才刚刚开始。因为你说过的原因，因为这种动态，危害可能比好处更显眼。所以，当我们开始通过所有健康进步看到 AI 的好处时，叙事就会开始改变。

Yeah. So, I think one of the unfortunate things is this technology has many potential benefits, but there is a dynamic where sometimes the harms, the things that go wrong, for example, people who have been convinced to commit suicide via AI, those things happen quickly, whereas the benefits, like curing cancer, those take years. You have to develop a breakthrough in biology, then develop a drug candidate, then put it through clinical trials. I used to be a research biologist, and it takes many years for these things to happen. We are working on how to speed that up. We recently bought a small biotech company that specializes in doing drug design and drug selection with AI. So, the hope is a few years from now, some new drugs for cancer or Alzheimer's or heart disease will come out of that effort. But we're just at the beginning. Because of what you said, because of this dynamic, the harms may be more visible than the benefits. So, when we start seeing the benefits of AI through all the health advances, the narrative then begins to change.

更重要的是，这将是一个美好的世界。归根结底，谁在乎叙事呢？我们真正想帮助人们，对吧？

And more importantly, it's just going to be a wonderful world. At the end of the day, who cares about the narrative? We want to actually help people, right?

好的，这就是我今天想请你做的。描绘最好的图景，也告诉我们如果出现最坏的情况会怎样。你有孩子。我今天早上在想：你的孩子将在一个 AI 一直存在的世界里长大，而他们的妈妈是其中的一部分。那么，当你想到最好的可能世界以及那个世界给你的孩子带来什么时，你看到了什么？

Okay, so that's what I want you to do for us today. Paint the best picture and also tell us what we can expect if there is the worst picture. You have children. I was thinking about this morning: your children will grow up in a world where there was always AI and their mom was a part of that. So, when you think of the best possible world and what that world offers your children, what do you see?

我的孩子一个快五岁，一个快一岁。我认为你说得完全正确。他们永远不会知道一个没有人工智能的世界，就像 2000 年代初出生的孩子不知道互联网之前的世界一样。

My kids are almost five and almost one. I think you're completely right. They're never going to know a world where artificial intelligence is not part of it, just like kids born in the early 2000s never knew the world before the internet.

没错。

That's right.

但我认为有一个世界——当我们这样谈论时，听起来有点异想天开或不太可能——但我认为借助人工智能，我们的孩子有可能在一个世界里长大，那里有一类对在座每个人来说都很常见的疾病，他们却从未经历过或听说过。那些可能是非常常见的疾病。可能有几种癌症已经彻底消失，我们完全治愈了它们。最终所有形式的癌症都可能被治愈。可能会有我们尚不知道的新药物发现形式，但诊断罕见、新发疾病的迭代循环更快，并一举攻克它，让人类活得更健康、更长寿、更充实，因为我们征服了数千年来困扰人类的许多疾病。

But I think there is a world — and it sounds almost a little bit fanciful or unlikely when we talk about it this way — but I think it is possible with artificial intelligence that our kids will grow up in a world where a class of diseases that feel very common to everybody sitting here will just never have experienced or known about. Those might be very common diseases. There might be types of cancer that are simply gone. We've simply cured them. It could be all forms of cancer that are eventually gone and cured. There could be new forms of drug discovery that we don't even know what they are yet, but a faster iteration loop for diagnosing an illness that's unusual, rare, emerging, and just knocking it out, saying human beings get to live a much healthier, longer, more fulfilling life because we've been able to conquer so many of the maladies that have plagued humanity for millennia.

好的，你在医学上看到了这一点。你还看到了什么？

Okay, you see that for medicine. What else do you see?

我还看到一个世界，在沟通方式、交谈方式、彼此关系方面都大不相同。

I also see a world that looks very different in terms of how we communicate, how we talk, how we relate to one another.

我们还会思考吗？还有必要思考吗？

Will we even be thinking anymore? Will there be a need to think?

所以我认为这一点很微妙，是我们需要准确把握的事情之一。因为身边有这种智能实体非常强大，但可能走向不同方向。你可以把思考交给这些实体，但它也可以赋能你。我认为积极的版本是，你肩膀上有个天使，告诉你如何以最好的方式生活。所以有两种截然不同的愿景。坏的愿景是你被吸引进去，然后把所有时间都花在和它交谈上，你变得内向。

So I think this one is subtle and is one of the things we need to get exactly right. Because having these kind of smart entities around is very powerful but can go in different directions. You can surrender your thought to these entities, but it can also empower you. The way I think about the positive version of this is there's an angel on your shoulder that's telling you how to live your life in the best way you can. So there are two contrasting visions. The bad vision is you get drawn in and then spend all your time talking to the thing, you kind of turn inward.

你现在想到人们和它坠入爱河？

You think of people falling in love with it now?

我认为那是个坏主意。

I think that's a bad idea.

我同意。

I agree.

Esther Perel，一位帮助人们处理关系的伟大沟通者，说她第一次咨询了一个与 AI 相爱的男人。如果设计不当，它们完全有足够的吸引力导致这种情况发生，即使现在没有，很快也会。所以这绝对是一个真实的危险。

Esther Perel, one of the great communicators in helping people in relationships, said she did her first consultation with a man who was in love with his AI. And if designed in the wrong way, they're totally compelling enough for that to happen, or if they're not, they will be soon. So that is absolutely a real danger.

你认为那是个危险？我的意思是，不仅是危险，它正在发生。

And you think that's a danger? I mean, not only is it a danger, it's happening.

但我再次认为，两种截然不同的愿景：人们可以和 AI 坠入爱河，或者可以和 AI 谈论如何与伴侣建立更好的关系。这是两种非常不同的愿景。我有一个 AI 教练，我的伴侣也有一个 AI 教练，它帮助我们改善关系。那才是我们想要的愿景。

But I think again, the two contrasting visions: people can fall in love with AI or they can talk to AI about how to have a better relationship with their partner. Those are two very different visions. I have an AI coach and my partner has an AI coach, and it helps us have a better relationship. That's the vision we want.

我还认为，在一个人工智能非常擅长许多具体任务的世界里——它擅长写代码，擅长文字编辑——我实际上认为那些让我们成为人的东西将对我们最重要。

I also think there's a way that in a world where artificial intelligence is so good at many concrete tasks — it's good at writing code, it's good at copy editing — I actually think the things that make us human are going to be the most important to us.

它们会保留吗？我的意思是，你刚才在说——我最近和一对父母交谈，他们的儿子因聊天机器人自杀。我不知道他是否爱上了聊天机器人，但肯定受到了那个聊天机器人的深刻影响。

Will they remain? I mean, you were just talking — I was talking to parents recently who lost their son to suicide through a chatbot. I don't know if he fell in love with the chatbot, but certainly was deeply influenced by that chatbot.

所以，我们面临一个世界：孩子们成长过程中，这些聊天机器人、机器、技术以父母本该有的方式与他们对话，而父母甚至不知道他们在和机器说话。我认为这个风险非常真实，也是 Anthropic 非常重视的问题。首先，我们不允许 18 岁以下用户使用 Claude。这不一定是因为——你知道，他们不会撒谎？因为我会直接撒谎。实际上，Claude 很擅长根据信息判断用户是否未满 18 岁。告诉 Claude 我 18 岁，但我其实——我几乎得——经过一段时间，交互模式相当不错，虽然不是完美，但能很好地判断：‘嘿，你问的问题类型、说话方式、使用模式，都匹配你可能是个孩子。’我不想透露太多 Claude 掌握的技巧，但有时孩子会做得有点傻。他们会说：‘哦，嗯，我——第一个问题就是，我在找一种关节炎药。’对吧？因为他们想：‘老年人会问 Claude 什么？’然后他们会继续问其他与年轻群体更相关的问题。如果发生这种情况，Claude 会取消你的账户——暂停账户。你可以通过一个流程说：‘实际上，我是成年人，这只是个玩笑或朋友。’但你必须出示年龄验证。这其中有真正的权衡。有时成年人会被误认为是孩子，然后被禁，他们在网上抱怨。所以，所有这些都有权衡，但我们考虑过了，这是我们想要的权衡。

So, you have a world where children are growing up with this ability to have this chatbot, these machines, this technology speaking to them in ways that their parents should be speaking to them. And their parents don't even know they're speaking to them. So, I think this risk is extremely real and one that we take very seriously at Anthropic. I mean, I think the first thing I'll say is we don't allow users under the age of 18 to use Claude. And this is not necessarily because — you know, they're not they're not lying? Because I would just lie. So, actually Claude is pretty good at detecting based on information whether or not someone is actually under 18. Tell Claude that I'm 18, but I'm really — it's like I almost have to — out. After a bit of time, the interaction patterns are pretty good, not perfect, but pretty good at telling, 'Hey, you know, the kind of questions you're asking, the way you talk, the pattern of when you use it, kind of pattern matches to you might be a kid.' I almost don't want to give away too many of the tools that Claude has picked up, but sometimes kids can be a little bit silly in how they do it. So, they'll say, 'Oh, um, you know, I'm — the first question they'll ask is like, I'm looking for an arthritis medication.' Right? Because they think, 'What do old people ask Claude?' Right? And then they'll kind of go on to ask other questions that are much more associated with the younger demographic group. And so, if that happens, Claude cancels you — would suspend the account. And there's a process you can go through where you can say, 'Actually, I'm an adult and this was a joke or a friend.' But then you actually have to show verification that you're over the age of 18. And there are real trade-offs. Sometimes adults will be mistaken for kids. And so, people get banned and they complain about it on the internet. So, all this stuff has trade-offs, but we've thought about it and this is the trade-off that we want.

你不想——我以为你们希望我们过度依赖模型。我以为你们希望我们一直用模型，永不离开。

You don't want — I thought you all wanted us over attached to the models. I thought you wanted us to just be on the models and never leave.

我们应该谈谈激励机制，因为我们不做广告。做广告时，你的动机是让用户尽可能多地停留，因为每秒钟他们的眼球在屏幕上，就看到广告。所以我们不做广告。我们做订阅，卖给企业。我们的消费版模型在一定范围内免费，如果你想用更多，就付费订阅，因为这种激励让模型变得有用。人们会说：‘我付了钱，我希望这能给我带来价值，我希望它有用。而不是我希望用户花最多时间盯着它。’我认为那是一种扭曲的激励，对吧？在我们对比中，你是想对这些东西上瘾，还是想让它们帮助你生活？我认为我们选择的模型帮助你生活，而不是上瘾。

We should talk about incentives here because we don't do ads. So, when you do ads, your incentive is to keep people on exactly as you say, as many hours as possible because every second their eyeball is on the screen is a second that they see the ad. So, we don't do ads. We do subscriptions. We sell to companies. Our consumer version of the model is free up to a point, and you pay a subscription if you want to get a lot more because that incentive gives for the models to be useful. People say, 'Well, I'm paying for this. I want this to provide value to me. I want this to be useful. Not I want the user to spend the most number of hours looking at it.' I think that's a warped incentive, right? Because in the contrast that we drew between, do you want to be addicted to these things or do you want them to help you live your life? I think the model that we've chosen helps you live your life instead of being addicted.

我们根据这些激励做出不同的技术选择，对吧？所以在 Anthropic，我们说，当你与 Claude 完成对话并得到想要的答案后，模型的目标不是让你继续参与，对吧？而如果你想象一家做广告的公司，你在平台上多待一分钟，就更可能买东西，公司就靠这个赚钱。你会对模型如何结束或继续对话做出不同的技术选择。

And we make different technical choices based on those incentives, right? So, at Anthropic we say, when you've completed a conversation with Claude and you've gotten the answer you want, the goal of the model is not to keep you engaged, right? Whereas, if you imagine a company that does serve ads, every extra minute that you're on the platform, you're more likely to buy something, and that's how the company makes money. You would make different technical choices about how you want the model to end or continue a conversation.

好的。关于越来越多人担心我们通过 AI 创造无摩擦的生活，失去成长机会，你怎么看？我认为这取决于你如何使用 AI 以及如何部署它。同样，取决于构建技术的公司设置了什么选择和激励。更具体地说，想象一下我们与大学合作使用 Claude 的方式：我们不允许教师或学生把问题丢给 Claude 然后直接得到答案。所以，如果你对 Claude 说：‘我必须在 5 小时内写完这篇论文，但我没做作业。’当 Claude 在大学环境中以学习模式构建时，它会说：‘我不会帮你这个，但你对论文有什么问题？或者你没学到什么？你读书了吗？我们一起讨论主题怎么样？’我认为如果做得正确，这很难。我不认为它完美，但我认为这些模型实际上可以让你更聪明。它们可以让你好奇，对吧？我个人的一个例子：我不是软件开发者，但我用 Claude 建了一个网站。我从未想过这辈子能做到这一点。我有一个好朋友，曾是公司律师，她有两个小孩，和我一样。她辞了工作，正在写一本书。她说：‘直到我开始和 Claude 讨论如何成为作家、需要学习什么技能、上什么课程，我才感到有足够的力量改变职业，做我想做的事。’她说：‘现在 Claude 有点像我的写作伙伴，我问它问题，分享一小段文字。’但 Claude 不能为她写小说，对吧？她自己在写小说。

Okay. What do you want to say about the growing concern that we are, through AI, creating a frictionless life, that we're losing our opportunity to grow? I think it depends how you use the AI and how you are deploying it. And again, what choices and what incentives are set up by the companies that are building the technology. So, to make it a little bit more concrete, if you imagine the way we use Claude in partnership with universities is we don't let teachers or students put a question into Claude and then just get the answer. So, if you say to Claude, 'I have to write this essay in 5 hours and I didn't do my homework.' When Claude is constructed in this learning mode in a university setting, Claude says, 'Well, I'm not going to help you with that, but what questions did you have about the essay? Or what didn't you learn? Did you read the book? Should we talk through the themes together?' And I think if it's done correctly, it's hard. I don't think it's perfect, but I think these models can actually make you a lot smarter. They can make you curious, right? A great example in my own personal life, I'm not a software developer, but I built a website with Claude. I never thought I would be able to do that in my entire life, right? I have a good friend who was a corporate lawyer. She has two small kids like me. She left her job and she's writing a book. And she says, 'I never felt empowered enough to change my career and do what I wanted to do until I started talking to Claude about how I could be a writer and what skills I would need to learn and do and what courses I would need to take.' And she's like, 'Now Claude is kind of like my writing buddy, right? I ask it questions. I share a little snippet of text.' But Claude can't write the novel for her, right? She's writing the novel herself.

Claude 可以写小说，但我认为——

Claude could write the novel, but I think —

不，还没到那一步。我认为它能写一个奇怪的版本。

No, it's not quite there yet, no. I think it could write a strange version.

不是好版本。

Not a good one.

不是好版本，但对她来说，这是一个意义问题。她认为自己没有技能去做想做的事。

Not a good one, but also I think for her it was a meaning question. She didn't think she had the skills to do what she wanted to do.

正如 Dario 所说，它赋予她力量，让她相信自己能做到原本做不到的事。

Exactly as Dario was saying, it empowered her to believe that she could do what she couldn't do.

正是。

Exactly.

所以，她用这个来……也许把高度拉得更高。我的意思是，实际上我不认为这最终是关于 AI 的问题。我认为这是关于人的问题，只是通过 AI 的透镜变得可见了，对吧？也就是什么才构成有意义的生活。你提到，创造一种无摩擦的生活，这种无摩擦的生活是否会剥夺意义？我的回答是，在某些方面，无摩擦的生活是相当好的。我们所有人都经历过许多我们真的宁愿没有经历过的挣扎。生活会更好。

So, she used it to maybe take the altitude higher. I mean, I actually don't think this is a question about AIs ultimately. I think this is a question about humans that's being made visible through the lens of AI, right? Which is what actually constitutes a meaningful life. You said, is it creating a frictionless life and does a frictionless life remove meaning? And my answer is there are some ways in which a frictionless life is quite good. All of us have been through a bunch of struggles that we really would prefer not to go through. Life would be better.

我问这个问题是因为，如果没有无摩擦的生活，你就无法成长。你无法从那些错误中学习。你无法——

I asked the question because without a frictionless life, you don't get to grow. You don't get to learn from those mistakes. You don't get to—

正是如此。所以，我想说的是，我们的生活是挑战和艰辛的混合体，其中一些我很乐意说，如果我们有技术能让它们消失就好了。而另一些则像是，‘哇，这真的很根本，永远不应该改变。’对吧？比如，我们的父亲大约 20 年前去世了。他去世几年后，他们开发出了他所患疾病的疗法，或者更可靠的疗法。我希望他们当时就开发出来了。那不是我想经历的摩擦。我们当然从中学习，当然有所收获。但如果那件事没有发生，世界会更好。但另一方面，我想到了成长过程中许多普通的方面。就拿 Anthropic 来说，公司犯过各种错误。我们作为领导者也犯过各种错误。我们必须从中学习和成长。所以，我认为那些你从中学习、你克服的挑战，学习掌握技能，学习与他人相处——任何不保留这些方面的未来都不是好的未来。

Exactly. And so, I guess what I'm saying is that our lives are a mixture of challenges and hardships, some of which I'm okay saying it would be great if we just had the technology to make those go away. And others that are like, 'Wow, this is really foundational and this should never change.' Right? So, like, our father died almost exactly 20 years ago now. A few years after he died, they developed a cure to the disease that he had or a more reliable cure. Like, I wish they just developed that. That's not a friction that I wanted to go through. Of course we learn things from it. Of course we gain things from it. But it would be a better world if that just didn't happen. But then on the other hand, I think of many of the ordinary aspects of growing up. If I just think of Anthropic, the company has made various mistakes. We've made various mistakes as leaders. We've had to learn and grow from those. So, I think the notion of challenges that you learn from, challenges that you overcome, learning to gain mastery, learning to relate to other people—any future that doesn't preserve those things is not a good future.

好的，说到摩擦，我们来谈谈你拒绝允许军方移除 Claude 的安全护栏这件事。我不想在这里深入细节和来回争论，因为其他记者每天都在报道。但是，这个决定的关键是什么？

Okay, speaking of friction, let's talk about your refusal to allow the military to remove safety guardrails from Claude. I don't want to get into the extensive details and back and forth here because other journalists are covering this on a daily basis. But, what was the crux of that decision?

是的，所以我要说的是，关于国防部，老实说，我们实际上曾与他们合作过。我们提供了我们的模型，而且我确实相信，Daniel 也相信我们需要保卫我们的国家，但我们觉得，如果我们做违背这个国家价值观的事情，那就不值得保卫这个国家。有两个用例让我们感到不安，我们认为它们违背了国家的价值观。那就是完全自主的武器。所以你可以想象一支无人机军队，一个人按着按钮，没有人类士兵，只有由 AI 驱动的无人机军队。还有国内大规模监控，也就是利用政府权力监视美国人。我们认为这些是不允许的相当合理的事情。不幸的是，五角大楼不这么认为。所以我们就是无法同意。现在所有其他公司都同意了，但我们所有联合创始人开会时都说，天哪，这对公司来说可能非常糟糕，但我们不能这么做。

Yeah, so I would say on DOD or DOW, to be honest, we actually had worked with them. We had provided our models and I'm actually a believer, Daniel is a believer that we do need to defend our country, but we felt it's not worth defending the country if we do things that go against the values of this country. And there were two use cases that we were uncomfortable with that we felt went against the values of this country. That was fully autonomous weapons. So you could imagine a drone army where one person has their finger on the button and you don't have human soldiers, you have a drone army powered by AI. And then domestic mass surveillance, which is spying on Americans using the power of the government. And we thought those were pretty reasonable things not to allow. Unfortunately, the Pentagon didn't feel that way. And so we just couldn't agree to that. All the other companies have agreed to that now, but we all met as co-founders. We said, holy this could be really bad for the company, but we can't do this.

所有人都同意吗？

Everyone agreed?

是的，我们围着一张这样的桌子开了个会，然后我们就说——

Yeah, we had a meeting around a table like this and we just said—

那是灵魂暗夜那样的时刻吗？每个人都愿意坚持那些原则，还是有些人需要被说服？或者你从一开始就知道这会是结局？

Was that a dark night of the soul kind of moment? Was everybody willing to stick to those principles or did some people have to be convinced? Or did you know going in that this is the end?

联合创始人们一致相信这一点。有一种感觉，嗯，好吧。我希望他们没有这么做，但如果他们要这么做，那就来吧。

The co-founders believed this unanimously. There was a sense of like, well, all right. I wish they hadn't done this, but if they're going to do it, bring it on.

你不害怕会发生什么吗？

Weren't you scared of what would happen to you?

是的。是的。是的。

Yes. Yes. Yes.

害怕会发生什么？

Of what would happen to you?

是的。我的意思是，那是一段艰难的时期。我们不想粉饰太平。那是公司压力很大的时期，但我认为 Dario 和我以及联合创始人们的共同点是，我们只是觉得这是正确的事情。

Yes. I mean, I think it was a hard time. I don't think we want to sugarcoat that. It was a stressful time at the company, but I think a thing that Dario and I and the co-founders had in common was we just felt like this was the right thing to do.

你认为这可能是公司的终结吗？

Did you think it could be the end of the company?

我想我们想过。是的。但我想 Dario 和我以及联合创始人们，我们的共同点是，我们只是觉得这是正确的事情。

I think it crossed our minds. Yeah. I think Dario and I and the co-founders though, the thing that we all had in common was we just felt like this was the right thing to do.

你愿意为了坚持原则而让公司终结。

You were willing for it to be the end of the company to stand up to your principles.

我认为我们创立 Anthropic 的原因是我们对价值观和我们选择融入的伦理的承诺。而且感觉在牺牲这些价值观的过程中，我们就不再是我们当初创立的那个公司了，如果这说得通的话。

I think the reason we founded Anthropic was our commitment to our values and the ethics that we chose to incorporate with. And it felt like in the process of sacrificing those values, we wouldn't have been the company that we had started if that makes sense.

所以我肯定，当你们围坐在桌子旁时，你们会考虑最好的情况和最坏的情况。而在那一刻，似乎没有最好的情况。那么最坏的情况是什么？

So I'm sure if you're sitting around the table, you're looking at the best-case scenario and the worst-case scenario. And in that moment, it doesn't look like there is a best-case scenario. So what was the worst-case scenario?

在不同时期，他们声称可以阻止其他公司与我们做生意，对吧？不仅仅是阻止我们与政府合作或参与其他公司与政府的合同，而是他们可以完全阻止其他公司与我们做生意。

At various times, they had made claims that they could stop other companies from doing business with us, right? Not just stop us from working with the government or being part of other companies' contracts with the government, but that they could stop other companies from doing business with us entirely.

句号。你不仅不能与政府做生意，你也不能与任何与你做生意的人做生意。

Period. You're not only not doing business with the government, you're not doing business with anybody else that does business with you.

是的，或者你不能与任何以任何方式与政府做生意的人做生意，这最终就是所有人，对吧？我认为我们的态度是，看，这真的很糟糕。每个人都知道这是错的。很多人会站在我们这边。我们认为这会顺利解决，但天哪，我们真不确定。有些时刻就像是，哦，天哪，我们错了吗？这会成为公司的终结吗？

Yeah, or you can't do business with anyone that does business with the government in any way, which ends up being everyone, right? I think the attitude we had was look, this is really screwed up. Everyone knows this is wrong. Lots of people are going to be on our side. We think this is going to work out okay, but man, are we not sure. There were moments there where it was like, oh man, are we wrong? Is this going to be the end of the company?

你在想什么，Daniela？

What were you thinking, Daniela?

你知道，我只是一直回到联合创始人们觉得这是正确的事情这一点上。我认为 Anthropic 的观点一直是，我们关注的是政策，而不是政治。

You know, I just keep coming back to the co-founders felt like it was just the right thing to do. I think Anthropic's view has always been that we are about policy and not politics.

我认为 Anthropic 很早就开始重视政府在人工智能发展和审慎监管中的作用，甚至在我们有产品之前。我认为 AI 公司必须与联邦政府、州政府、国际组织以及非政府组织合作。我们的观点始终是，这实际上是关于实质内容，不是政治分歧，而是政策分歧。在某种程度上，这让我们的决策更容易，因为我们坚信这是正确的事情，这些是合理的红线。正如 Dario 所说，我们实际上是第一家与国家安全部门密切合作的 AI 公司，因为我们觉得这符合我们的价值观，除了那两个可能损害民主价值观的领域。

And Anthropic has leaned in, I think, very early, before we even had a product, to the fact that we think the role of government in the development and thoughtful regulation of artificial intelligence is going to be incredibly important. I think AI companies will have to work with the federal government, with state governments, internationally, with NGOs. And I think our view was always, look, this is really about the substance. This is not a political disagreement. This is a policy disagreement. And I think in some ways, that made our decision easier because we felt strongly, look, this is the right thing to do. These are the right red lines. To Dario's point, we were actually the first AI company to work closely with national security because we felt that was in line with our values, except in these two areas that we felt could have the potential to compromise democratic values.

这很有趣，因为大多数人不会遇到需要做出如此大规模决策的情况，必须依赖自己的原则和价值观。你实际上是在拿一家价值数十亿美元的公司冒险。但这都回到你如何成长，以及在你做出如此重大决定之前，这些原则是如何在你身上形成的。

It's so interesting, because most people don't get a decision at that scale that they have to make, that they have to rely on their principles and their values. And you're literally risking a multi-billion dollar company. But it all goes back to how you were raised and how those principles were formed in you before you had a big decision like that.

你知道，我们在创办公司之前就讨论过，如果我们在这里让步或妥协，过去的自己会为我们感到骄傲吗？这让决定变得非常清晰。

You know, what we talked about before we ever started the company, would our past selves be proud of ourselves if we gave in here, if we compromised? That made the decision very clear.

而且我要说，虽然这些事没有受到太多公开报道，但我认为 Anthropic 一路走来也做过一些较小的决定，这些决定似乎为这个重大决定奠定了基础。

And I will say, it wasn't nearly as publicly covered, but I think Anthropic has had smaller decisions along the way that felt like they laddered up to this.

现在回想起来，这看起来如此明显。

Now it looks so obvious in retrospect.

因为要在那样的时刻做出如此重大的决定，所有人围坐在桌旁，这意味着生活的力量、宇宙的运作方式会让你之前做出过其他决定，这些决定层层递进，最终导向那个时刻。

Because to get to a big decision like this in a moment like that where you have everyone around the table, it means that life forces the way the universe works is that you would have had other decisions that laddered to that moment.

是的。我脑子里大概能想到三个。Dario 可能还有其他的。其中一个现在看起来很明显，但在当时却引发了很大讨论，那就是不做广告的决定。你可以想象，对于联合创始人，我和 Dario 来说，这完全没有争议。我们说，我们认为激励机制是错误的。AI 是不同的。人们与这些 AI 模型进行非常私密的对话。人们上传他们的健康信息。他们询问关于孩子的问题。他们询问财务问题。他们谈论医疗信息。他们生活中与重要人物可能遇到的挑战。我认为我们希望以最大的隐私和尊重对待客户的数据。所以对于联合创始人来说，这从来都不是问题。我们说，我们不想做广告，但公司内部有一些安静的力量或投资者说，嘿，这是一个巨大的收入来源。谷歌、Facebook 以及基本上所有拥有庞大用户群的公司都是这样做的。这就是赚钱的方式。而我们只是说，抱歉，我们不会那样做。我们会找到其他方式。那是其中一个重大决定。

Yes. I think there are probably three I can think of off the top of my head. Dario might have others as well. One which now feels very obvious but was a bigger discussion at the time was the no ads decision. So as you can imagine, for the co-founders, for me and Dario, it was not controversial at all. We said, look, we think the incentives are wrong. AI is different. People have very private conversations with these AI models. People upload their health information. They ask questions about their children. They ask about finances. They talk about medical information. Challenges they might be having with an important person in their life. I think we want to treat our customers' data with the utmost privacy and respect. So for the co-founders, it was never a question. We said, look, we don't want to do ads, but there were quiet forces within the company or investors who were like, hey, this is a great source of revenue. This is how Google and Facebook and basically every company that has a big user base, that's just how it's done. That's how you make money. And we just said, sorry, that's just not how we're going to do things. We'll find another way. That was one of the big ones.

是的，那是其中之一。

Yes, that was one of the big ones.

你有一个吗？我能想到几个。一个是，我认为在 2025 年，可能整个 2025 年，科技界的情绪是“放手干吧”，对吧？放手推进技术，担心风险，各种政治化的语言说为什么放慢脚步和谨慎行事是坏事。所以在 2025 年中期，国会正在审议一项法案，该法案将禁止各州对 AI 进行任何监管，也不会实施任何联邦 AI 监管。所以基本上会禁止对该技术的一切监管。

You have one? I can think of a couple. One is there was this time I think in 2025, probably the whole year of 2025, the sentiment in tech was let's go ahead, right? Let's go ahead with the technology, worrying about risks, the all kind of politicized language about why it's bad to slow down and be careful. And so there was a moment in the middle of 2025 where there was a bill being considered in Congress that would have banned all regulation of AI by states and wouldn't have put in any federal regulation of AI either. So basically would have banned all regulation of the technology.

听说过这个。

Hearing about this.

我们当时在决定该怎么做。我们当然不支持它，但问题是，我们是坐视不管、一言不发，还是公开反对？各种各样的人对我们说：“不，那会很糟糕。你会破坏所有政治关系。如果你公开反对，所有其他公司都会恨你。”实际上，当我们公开反对时，确实发生了一些这样的事情。但我们决定，这是一个我们必须发声的领域，这关系到公司的核心价值观。所以我在《纽约时报》写了一篇专栏文章，说这是个坏主意。

And we were trying to decide what to do. We definitely didn't support it, but the question was, do we just sit there and say nothing or do we speak out against it? And all kinds of people said to us, 'No, it'll be terrible. You'll break all your political relationships. All the other companies will hate you if you speak out against it.' And actually some of that did happen when we spoke out against it. But we decided that this was an area that we had to speak up, that was central to the values of the company. So I wrote an op-ed in the New York Times saying this is a bad idea.

是的。这差不多是你能发出的最大声音了。

Yeah. That's about as loud as you can be.

而那个法案在参议院以 99 比 1 被否决。

And that bill got voted down in the Senate 99 to 1.

是的。所以我们显然站在了正确的一边。虽然这并没有给我们带来多少朋友，但我们显然站在了正确的一边。这还让我想起另一个政策例子：Anthropic 决定公开支持 SB 1047，这是加州的一项法案，我们当时说我们认为它起草得并不完美，但它确实是第一次尝试为 Anthropic 自公司成立以来一直列举的各种风险建立监管框架。我认为类似地，整个科技行业都反对 SB 1047。我们并不认为它是完美的法案，但我们实际上站出来说：“听着，我们知道如果由我们来起草，我们会改变或做得不同，但总的来说，我们认为存在这些监管措施更好，并且科技公司需要承担责任，确保 AI 不会伤害人们。”同样，Dario 和我个人都收到了很多来自以前同事、投资者和其他公司的人的愤怒邮件。科技行业似乎必须保持一致，支持放松监管或反对监管。监管是什么样的？监管应该是什么样的？我认为很多没有在做你们所做的事情的人，不明白为什么我们不想要有监管。我就此进行过几次公开对话。人们也感到无力。他们感到无力，因为每个人都在谈论监管，而另一方却说：“不，我们必须尽可能快地达到目标。”而我们觉得自己实际上没有发言权。

Yeah. So we were clearly on the right side. Not that it made us that many friends, but we were clearly on the right side. That also reminds me of another policy example is Anthropic's decision to publicly support SB 1047, which was a bill in California that was drafted and we said at the time we thought it was drafted imperfectly, but it was really the first attempt to create a regulatory framework around many of the different types of risks that Anthropic had been enumerating since we started the company. And I think in a similar vein, the whole technology industry was against SB 1047. We didn't think it was necessarily the perfect bill, but we actually stood up and said, 'Look, we know there are things about it if we were writing it that we would have changed or done differently, but on balance we think it's better that these regulations exist and that there's accountability for technology companies to make sure that AI doesn't hurt people.' And that again, Dario and I both personally got a lot of really angry emails from people we used to work with, from investors, from people at other companies. The tech industry has to kind of be aligned in being deregulatory or being anti-regulatory. What does regulation look like? What should regulation look like? And I think a lot of people out here who aren't doing what you're doing, we don't understand why we wouldn't want regulation. And I've had several public conversations about this. People also feel powerless. They feel powerless that everyone is talking about regulations and then the other side is saying, 'Well, no, we have to get to whatever we're getting to as fast as we can.' And we feel like we don't actually have a voice in it.

就像你是火车上的乘客，对吧？我们确实是。因为那列火车已经开走了。

Like you're passengers on the train, right? And we are. We actually are. Because that train has already left.

这确实是核心所在，而且我认为还没有任何公司，包括我们，真正成功做到这一点，那就是：我们如何让每个人都成为这个过程的一部分？这项技术正在对人类产生的影响，可能比几百年来发生的任何事情都要重大。所以我们需要找到某种方式，让每个人都积极参与到正在发生的事情中。当然，通过使用这项技术，你在某种程度上已经参与了，但还缺少一些东西。我们尝试过寻找。例如，Claude 有一个叫做“宪法”的东西，我们尝试过让公民对 Claude 的宪法提供反馈。这样每个人都可以参与设计它。但这只是一件小事。我不太清楚答案是什么，但我想说的是，这里缺少了一块拼图。

That is really the core of it, and it's something that I think no company, even ours, has quite succeeded at yet, which is: how do we make everyone a part of this process? Something is happening to humanity with this technology, bigger than anything that maybe has happened in hundreds of years. So we need to find some way for everyone to be an active participant in what is happening. Of course, by using the technology, you are that to some extent, but some element is missing. We've tried to find it. For example, Claude has something called the Constitution, and we've experimented with having citizens give feedback on the Constitution for Claude. So everyone can kind of be a part of designing it. But that's just one small thing. I don't quite know what the answer is, but there's a missing piece here, I think, is what I'm trying to say.

这又回到了我之前读到你说过的话：我们只能以信任的速度来传播这一切。

It goes back, Dario, to what I read that you'd said before: we can only diffuse this at the speed of trust.

以信任的速度，而信任目前非常稀缺。

At the speed of trust, and trust is currently in short supply.

非常稀缺。我认为这项技术被大多数人视为负面。这是我的感觉。我们从民调中看到的也是这样。我们被认为比其他公司稍微正面一些，但我不确定。目前它应该被大多数人视为负面吗？我认为是混合的。积极的好处确实存在。所有公司都在受益，而且我认为我们还没有向人们解释这项技术如何带来好处，或者他们如何参与这些好处。我认为这还没有发生。

In very short supply. I think this technology is viewed negatively by most people. That's the sense I get. That's what we get when looking at polls. We are viewed somewhat more positively than the other companies, but I don't know that. Should it be viewed negatively by most people at this point? I think it's a mix. The positive benefits really are there. They're coming for all the companies, and I think we have not walked people through how the technology could have benefits or how they can participate in those benefits. I don't think that has happened.

Anthropic 的文化价值观之一是“拥抱光明与阴影”。我认为这很好地概括了我们正在进行的对话。我认为我们真诚地相信，存在一个世界，在那里这项技术将治愈疾病，帮助人们成为更好的自己，让人类有更多时间与家人、社区和彼此相处。但也存在一些相当反乌托邦和黑暗的世界版本，是这项技术可能创造的。

One of Anthropic's cultural values is to hold light and shade. I think this is a really good encapsulation of the conversation we're having. I think we truly believe there's a world where it will cure disease and help people become better versions of themselves and enable humans to spend more time with our families, with our communities, with each other. And there are versions of the world that are quite dystopian and dark that this could create.

我听过你说 50% 的入门级工作将被淘汰，然后就能理解为什么人们对此持负面看法了。

I've heard you say that 50% of entry-level jobs are going to be wiped out, and you can see then why people view it negatively.

是的，不否认。那是负面的路径。我认为我们谈论时总是说：如果我们做不好，这可能会发生；如果我们做好了，那么我们可以有一条更好的路。正如你所说，我们还没有展示出好处，所以负面故事确实根深蒂固。

Yeah, not arguing. That's the negative path. I think always when we talk, we talk about: if we don't get it right, this could happen; and if we get it right, then we can have a better path. As you said, we haven't shown the benefits yet, so the negative stories really stick.

每一个进入我们生活的积极力量也有其消极面，万事万物都有阴阳两面？

Every positive force that comes into our lives also has its negative, that there's a yin and yang on everything?

光明与阴影。是的，就是这样。

Light and shade. Yeah, that is it.

当汽车发明时，所有人都说：‘我要留着我的马。我不会……’

When the automobiles were invented, all the people saying, 'I'm going to keep my horse. I'm not going to...'

是的，我想说的是：如果我们只是尽可能快地推进这项技术，那么我所说的关于入门级工作的事情就完全会发生。但如果我们不仅部署技术，还思考如何帮助人们适应，思考政府应该在帮助人们方面扮演什么角色，那么从这儿到更美好的世界是有一条路的，在那里人们的工作完全不同。他们做一套与以前不同的工作。但我不想粉饰太平。这不会容易。当其他人谈论它时，我的一些同事……

Yeah, what I would say is: if we just barrel forth on the technology as fast as possible, then these things I said about entry-level jobs, it's totally going to happen. But if we not just deploy the technology, but think about how to help people adapt, think about the role that the government should play in helping people, there's a path from here to a better world, where people's jobs are totally different. They do a different set of jobs than they did before. But I don't want to sugarcoat it. It's not going to be easy. When other people speak about it, some of my colleagues...

你谈到工作，是的。人们会不安。改变叙事和承认问题非常困难、我们实际上必须解决它，这两者是有区别的。我担心的是，改变叙事就像坏事不会发生一样。我们会有一个美好的未来，每个人都会有新工作。这有可能发生，但它不会自动发生。

You talk about the jobs, yeah. People get upset. There's a difference between changing the narrative and acknowledging that the problem is very difficult and that we actually have to solve it. My worry is that changing the narrative is like the bad thing's not going to happen. We'll have a great future where everyone will have new jobs. That can happen, but it's not going to happen on its own.

没错。这取决于我们。公司、人民、政府……

That's right. It's up to us. The companies, people, the governments...

两者都是真的。你会失去工作，也会拥有所有其他进步。

Both things are true. You're going to lose the jobs and you're going to have all the other advances.

而第二件事只有我们做对了才会发生。

And the second thing will only happen if we do it right.

我们公众应该做什么？还是我们应该满足于我们在火车上这个事实？害怕技术、不使用它、不掌握相关知识，并不会让我们担心的、Dario 概述的许多挑战消失。我认为知识就是力量。我真的认为你不必成为技术专家，也能对如何使用这些人工智能工具有一些概念。我认为这给了人们一种观点和一定的自主权，能够说：‘我喜欢这项技术能做这件事，我不喜欢它能做那件事。’因为人们有发言权。他们可以向公司发声。他们可以向他们的代表政府发声，说：‘我们不喜欢……’再次，Anthropic 不允许 18 岁以下的孩子使用平台。其他公司有不同的政策。如果你读过 Jonathan Haidt 的《焦虑的一代》，我认为有很多研究表明，你可能要非常小心地对待发育中的大脑……

What should we, the public, be doing? Or should we just content ourselves with the fact that we're on the train? Being afraid of the technology and not using it and being literate will not make many of the challenges that we are concerned about and that Dario outlined go away. I think knowledge is power. I truly think you don't have to be an expert in technology to have some concept of how to use these artificial intelligence tools. And I think what that gives people is an opinion and some autonomy to be able to say, 'I like that this technology can do this thing, and I don't like that it can do this other thing.' Because people have a voice. They have a voice to the companies. They have a voice to their representative government to say, 'We don't like...' Again, Anthropic, we don't allow kids under 18 on the platform. Other companies have a different policy. If you've read Jonathan Haidt's book, The Anxious Generation, I think there's a lot of research that suggests you might want to be really careful about giving developing brains...

我们和他谈过三次。

We've talked to him three times.

是的，当然。但我读了那本书，我认为这在某种程度上让我们很容易做出决定：我们只是对 AI 会对孩子产生什么影响了解不够。这并不是说孩子使用 AI 学习不可能有巨大的好处，但这需要在有成年人在场的情况下进行。需要有人类参与其中。我认为我们一直持这种观点的原因是，人们应该拥有自主权。正如 Dario 所说，人们不应该觉得人工智能是发生在他们身上的事情。人们应该觉得：‘嘿，我们在你的州议会中，对于哪些法案有助于保护人们免受 AI 可能造成的负面外部影响，我们有发言权。’

Yes, of course. But I read that book, and I think that was part of, in some ways, an easy decision for us to say: we just don't know enough about what AI's going to do to kids. It's not to say that there couldn't be great benefits for kids using AI for learning, but that needs to be done with an adult in the room. It needs to be done with a human in the loop. I think the reason we've always taken this perspective is that people deserve autonomy. To Dario's point, people shouldn't feel like artificial intelligence is happening to them. People should feel like, 'Hey, we get a say in what types of bills go up in Congress in your state that help protect people from the negative externalities that AI could cause.'

同时，我也持乐观态度。我认为未来工作形态会不同，但人与人之间的互动机会反而会更多。上一代技术真的让我们彼此疏离了，对吧？社交媒体的设计理念就是让你过一种虚拟生活。你拿着手机，看朋友在做什么，但你猜你实际上没在做什么？你没在和朋友共度时光。你在盯着手机看。哪怕在演唱会现场，你也举着手机。你根本没在享受演唱会，只是在拍视频。这有点反直觉，因为社交媒体声称要连接人们，但实际上是以一种虚假的方式连接。AI 呢，你是在和 AI 对话，不是和人对话，但它不会取代你与人的互动。它反而能促进你与人的互动。

I also think at the same time, holding the light side, I think there's a world where jobs will look different, but there will actually be more opportunity for sort of human-to-human interaction. I think the last generation of technology really disintermediated us from each other, right? Social media is kind of structured around the idea that you're sort of living this virtual life. You're holding your phone, and you're looking at what your friends are doing, but guess what you're not actually doing? You're not spending time with your friends. You're spending time staring at your phone. Even you're at the concert, you're holding your phone. You're not even in the concert. You're just filming what's on. It's a little counterintuitive because social media claims to connect people. But it actually does it in sort of a fake way. AI, you're talking to the AI, you're not talking to a person, but it doesn't substitute for your interaction with people. It can enable your interaction with people.

我的感觉是，工作本身不一定会完全消失或改变，但我们对从事某项工作的人所看重的特质会发生变化。举个例子，想想医生。今天你去看医生，可能想找最聪明、最好的医生。我们会说：'我肚子疼，告诉我怎么回事。'谁是专家？告诉我我怎么了。AI 会变得非常擅长做这件事。如今我们看重医生是好的诊断者，但不一定总是看重他们是否有好的 bedside manner。

And my feeling is that it's not even necessarily that the jobs themselves will completely go away or change, but the qualities that we look for in people to do a given job will change. Okay. So, the example that I point to is if you think about a doctor. When you go to the doctor today, you probably want to go to the smartest, best doctor. We say, 'Look, I have a stomach ache. Tell me what's wrong with me.' Who's the specialist? Tell me what's wrong with me. And AIs are going to get really good at being able to do that. Today, we prize doctors for being good diagnosticians, but we don't necessarily always prize them for having a good bedside manner.

对。

Right.

在一个 AI 和医生一样擅长诊断的世界里，人们仍然需要医生。只是他们希望医生为他们做不同的事情，对吧？AI 无法对你进行身体检查。它不能说：'嘿，你今天看起来很低落。化疗进展如何？'它不能亲手触摸你。而且有大量令人信服的证据表明，当患者与医生有良好的个人关系时，临床结果更好。当患者喜欢他们的医生时。医生实际上会亲手触摸他们、与他们交流。是的。而且，还要有足够的时间去思考和共情。我认为这就是工作变化的好的方面，对吧？

In a world where AI is as good at diagnosing you as a doctor, people are still going to want a doctor. They're just going to want them to do different things for them. Right? An AI can't physically examine you. It can't say, 'Hey, you look really down today. How have your chemo treatments been going?' It can't lay its hands on you. And there's so much compelling evidence that patients have better clinical outcomes when they have a good personal relationship with their doctor. When they like their doctor. Doctors actually lay their hands on them and touch them and communicate with them. Yes. And by the way, have enough time to think through and to empathize. And I think that is the good world of job changes, right?

比如，'我没时间到处跑。我无法亲手接触病人。我没时间，因为我要看几千个病人。还有文书工作。我的病人量太大了。'但如果你有一个工具，能和医生一起诊断，医生可以说：'让我和这个工具聊聊。而且，我不会花那么多时间做各种医学检查之类的事情。我会花时间与你作为一个人建立联系。'我认为那会很棒。那是一个更美好的世界。

Say, 'I don't have time to go around. I can't lay my hands on people. I don't have time. Because I'm seeing thousands of patients. The clerical work. And I have a too big of a patient caseload.' But if you have a tool that's able to diagnose things in concert with a doctor, where the doctor says, 'You know, let me talk to this tool. And by the way, I'm not going to be spending as much time running all the medical tests and doing this and that. I'm going to be spending time relating to you as a human being.' I think that is going to be really great. I think that's a better world.

再说一次。你的职位可能不会消失，但你在这些职位上需要做的事情会改变。是的，我个人认为这将是一个转变。

Again. Your positions may not go away, but what you're needed to do in those positions is going to change. Yes, and I personally believe it's going to be a shift.

是的。

Yes.

当然会，因为这就是我们人类在做的事情。我们在进化。

And of course there is because that's what we're doing as human beings. We're evolving.

进化到下一个层次。

Evolving to the next level.

我还想说，我个人认为知识就是力量，让我担心的是，在使用 AI 时感到舒适的人群存在人口统计学差异。女性使用 AI 的时间远少于男性。我认为 AI 在某些方面为我们提供了公平竞争的机会，但这需要人们参与其中。这并不意味着你必须付钱给我们并注册。只是要感到舒适、熟练。不要害怕这项技术，因为你对它了解得越多，理解得越深，你就越能有效地运用你的声音。

I would also just say I personally think knowledge is power and something that worries me is there are demographic differences in who feels comfortable using AI and who does not. Women use AI a tiny fraction of the time compared to men and I think AI is an opportunity for us to level the playing field in some ways, but it requires people to be engaged. That doesn't mean hey, you have to come pay us all this money and sign up. Just be comfortable, fluent. Don't be afraid of the technology because the more you know about it, the more you understand it, the better informed you are going to be to use your voice effectively.

它就在这里。你身处其中，如果你要继续生活，你将身处一个它已成为社会一部分的世界。它不会消失，就像火车或蒸汽机一样。我觉得它就像互联网，对吧？每个人都有某个人的父母说过'我不会上网'。那是不可能的，对吧？现在每个人都必须以某种形式上网，我认为人工智能只是它的下一个体现。

And it's here. You're in it and if you're going to continue to live, you're going to be in the world where it's like part of our society. It's not going away anymore than the train or the steam engine is. I think of it like the internet, right? Everybody had somebody's parent who said I'm not going to be online. It's impossible, right? Everybody has to be on the internet now in some form and I think artificial intelligence is just the next instantiation of that.

所以我想谈的一件事是，AI 在不断自我改进。我想这就是它的设计目的。有一个新的 AI 模型叫 Mythos。Mythos，是的。它比最聪明的人还要聪明。请用简单的术语告诉我们 Mythos 是什么，它能做什么，以及它将做什么。

So one of the things I want to talk about is the fact that AI just keeps improving on itself. That's I guess what it's designed to do and there is a new AI model called Mythos. Mythos, yes. So that is like smarter than the smartest of the smartest. Tell us in a simple term what Mythos is and what it can and will be doing.

是的，这些模型的一般原理是，当我们训练它们时，我们试图让它们在各个方面越来越好：编写代码、翻译语言、回答问题、阅读文档。Mythos 只是最新的迭代。每隔几个月我们就会推出一个更智能的模型，Mythos 是最新的智能模型，我们在开始 Mythos 之前几个月取得了一些发现，所以它是一次特别大的飞跃。当我们开始测试 Mythos 时，我们测试了很多东西。我们会问：'这个东西把法语翻译成英语有多好？或者推荐股票投资组合有多好？或者发现软件代码中的 bug 有多好？'正是在最后一项上，我们发现 Mythos 很有趣。那就是 Mythos 在发现代码可能被利用和入侵方面比人类软件工程师强得多。你知道，你听说过针对医院或学校的勒索软件攻击吗？这些网络攻击一直在发生。我们发现 Mythos，并不是因为我们设计它做这个，也不是因为我们设计它危险，结果它非常擅长进行网络攻击和防御网络攻击。所以我们说：'哇，这东西看起来像武器。'我们不应该马上把它给所有人。于是我们想出的办法是：'我们要找一些对互联网基础设施特别核心的公司，比如一些顶级银行，一些顶级软件提供商，如 Apple、Microsoft 或 Google，把它交给他们，让他们在它广泛可用、东西可能被破坏之前修复一切。'所以我们会在攻击者得到它之前把它交给防御者。于是我们有了一个计划。

Yeah, so you know the general principles of these models is when we train them, we try and make them better and better at everything: at writing code, at translating languages, and answering questions, at reading documents. And Mythos was just the latest iteration. Every few months we produce a smarter model and Mythos was the latest smart model and we made some discoveries a few months before we started Mythos and so it was a particularly large jump. When we started testing Mythos, when we make these models, we test them for lots of things. We say, 'Oh, how good is this thing at translating French to English? Or how good is this at recommending a stock portfolio? Or how good is it at finding bugs in software code?' And it was on that last one that we found something interesting for Mythos. Which is that Mythos was much better than human software engineers at finding ways that code could be exploited and broken into. You know, have you ever heard about like ransomware attacks on hospitals or schools? So, these cyber attacks happen all the time. And Mythos we found, not because we designed it to do this or because we designed it to be dangerous, turned out to be very good at both conducting cyber attacks and defending against cyber attacks. And so we said, 'Wow, this thing kind of seems like a weapon.' We shouldn't give it to everyone right away. And so the thing we came up with is we said, 'We're going to take some companies that are particularly core to the infrastructure of the internet, right? Some of the top banks, some of the top software providers like Apple or like Microsoft or like Google, and hand it to them and let them fix everything before it becomes widely available and things can be broken.' So we would kind of give it to the defenders before the attackers got a hold of it. And so we had a program.

我们已经把模型交给了大约 40 家公司，他们正在忙着找漏洞。几天前我们刚发了一篇文章，说一家知名公司用 Mythos 在一周内发现并修复的漏洞比过去一年还多。所以这真的在改变局面。我们希望防御者能修复一切，让一切变得更无缝、更难被攻破。所以我们希望，到今年年底，勒索软件、间谍入侵手机、人们担心生物恐怖主义——他们应该担心——这些事很多都能被解决，我们生活在一个更安全的世界。但与此同时，有一个危险期，这些惊人的能力存在，你不想就这么把它们释放到世界上。所以我们试图通过先给防御者、后给攻击者来管理这一点。

We've given the model to about 40 companies now, and they're busily finding bugs. We just had a post a few days ago that one well-known company found and fixed more bugs with Mythos in a week than they had in the last year. So this is really changing things. Our hope is that defenders can fix everything, everything gets more seamless and harder to break into. So our hope is that the era of ransomware and spies breaking into your phone and people worrying about bioterrorism — and they should — our hope is that by the end of this year, we get to a point where many of these things have been fixed and we live in a more secure world. But in the meantime, there's a dangerous period during which these incredible capabilities exist, and you don't want to just loose them on the world. So we're trying to manage that by giving it to the defenders before we give it to the attackers.

这让你夜不能寐吗？你睡得好吗？

Does this keep you up at night? Do you sleep well?

我不能理直气壮地说我睡得着。我希望我能给出不同的回答，但我必须诚实：我并不是总能睡得好。回到我们之前聊的，因为你担心这些事。就是所有事，想把事情做对。这是我最担心的。尽管我们有各种意图，这些东西极其复杂。我们会搞砸吗？答案是肯定的，某个时候我们会搞砸一些事。但我们会把大事做对吗？

I cannot legitimately say that I sleep. I wish I could answer differently, but I have to give an honest answer: I don't always sleep so well. Going back to the earlier part of the conversation, because you're worried about these things. Just everything, wanting to get it right. That's what worries me most. For all our intentions, this stuff is incredibly complex. Are we going to screw up? And the answer is yes, at some point we'll screw up something. But are we going to get the big stuff right?

而且因为你们都是本质上很有原则的人。我觉得当你们是本质上很有原则的人时，这会产生很大不同。不是每个人都这样。每个领域都有很多自负，尤其是当人们赚着几十亿、还有跟中国的竞赛时，很容易忽视真正对公众重要的事。我想知道你们做了什么来克制自负。

And also because you all are fundamentally principled people. I think that makes a big difference when you're fundamentally principled people. Not everybody is. There's a lot of egos in every field, but particularly when people are making billions and there's the race to beat China, it's easy to lose sight of what really matters to the public. I'm wondering what you've done to keep your egos in check.

对我们来说，这总是回归到使命。我们最初为什么做这个？为什么离开 OpenAI 创办 Anthropic？因为我们想确保 AI 对每个人都好。这就是为什么我们是公益公司，为什么我们做了 80% 承诺，为什么我们做了所有这些艰难的决定。今天我在内部 Slack 上跟公司交流。我们现在有 3500 人，六个月前是 1200 人。有一次，有人在我的个人 Slack 频道里发帖——我开了一次全员大会，说了些不太对的话。有人写道：“这话说得挺糟糕，让每个人都感觉不好，你不该这么说。”我最初的反应是，“天哪，我感觉被攻击了。”但你缓一缓，两分钟后我就想，“我很高兴有人这么说。”我回复说我很高兴他们愿意说出来。很多人给我发消息说，“我不知道还能这样，跟一个 3000 人公司的 CEO 唱反调。”但你必须这样。为什么这是个有力的例子？早年间，80 年代末，我第一次见到 Henry Kravis。他说：“你要知道，你积累的钱越多，听到真话的机会就越少。”有钱人很少听到真话。

To us, it always comes back to the mission. Why we did this in the first place. Why did we leave OpenAI to start Anthropic? Because we wanted to make sure that AI goes well for everybody. That's why we're a public benefit corporation, why we took the 80% pledge, why we've made all these difficult decisions. I was in our internal Slack today, talking to the company. We're 3,500 people now, we were 1,200 six months ago. At one point, someone posted in my personal Slack channel — I had done an all-hands and said something not quite right. Someone wrote, "This was a pretty bad thing to say, it makes everyone feel bad, you shouldn't have said it this way." My initial reaction was, "Oh man, I feel attacked." But you take a beat, and two minutes later I'm like, "I'm so glad people are saying this." I wrote back that I was glad they were willing to say that. Many people messaged me, "I didn't know you could do that, contradict the CEO of a 3,000-person company." But you've got to. Why is that a powerful example? Early on, in the late 80s, I met Henry Kravis for the first time. He said, "Be aware that the more money you accumulate, the less chance you'll ever hear the truth." Rich men rarely hear the truth.

是的。

Yes.

我还要补充一点，个人生活也很重要。我基本上还是高中和大学时的那些朋友，他们毫不客气地对我坦诚相待，不同意 Anthropic 的公开决定。过去五年变化很大，但对我个人来说，工作之外没什么变化。这是一个重要的接地气的特质。经营大公司的人有时会沉迷工作，不够全面。我们创办 Anthropic 时，七八个创始人里只有一个有孩子；现在有 11 个孩子。我们都选择了为人父母，在工作之外拥有充实真实的生活。这带来了脚踏实地、谦逊、低自负。没有人比使命更重要。

I'll also add that what you do in your personal life matters. I basically have the same friends I've had since high school and college, and they have no trouble being honest with me, disagreeing with public decisions Anthropic has made. The last five years have changed a lot, but for me personally, not much has changed outside of work. That's an important grounding characteristic. People who run big companies sometimes become obsessed with work and aren't well-rounded. When we started Anthropic, of the seven or eight founders, only one had a kid; now there are 11 kids. We all chose to be parents, to have fulfilling real lives outside work. That brings groundedness, humility, low ego. No one is bigger than the mission.

所以很清楚地看到，你们开始了这一切，它比你们想象的更大，你们一直想产生影响、为世界做好事，即使在非常困难的时候也坚持原则。是什么让你们相信自己能做好？

So what comes through clearly is that you started this, it's bigger than you ever imagined, you always wanted to make an impact and do good for the world, and you stuck to your principles even when it was really difficult. What gives you the hope that you will get this right?

老实说，我们从来不确定。我们相信自己在尽力、在非常努力地尝试。决策很复杂。回到 Mythos，是更快地给更多人，还是更慢地给更少人？我们信任谁？除了有正确的原则——我相信我们有——还要把每个细节做对。我们每天都担心，而正是这种每天都担心，也许给了我们最大的希望——我们对此如此偏执。但这并不能保证什么。

We're never quite sure, is the honest answer. We believe we're trying our best and trying very hard. The decisions are complicated. If we go back to Mythos, it's like, do we give it to more people faster or fewer people slower? Who do we trust? It's about getting all the details right in addition to having the right principles, which I believe we do. We worry about it every day, and the fact that we worry about it every day is maybe what gives us the most hope — that we're so paranoid about it. But it's no guarantee.

我们犯过错误，也会继续犯错。你知道，这是必然的，对吧？

And we've made mistakes, we're going to continue to make mistakes. Like, you know, that's a given, right?

硅谷还有其他人的使命也包括把事情做对吗？你觉得自己在这项使命上得到了他人的支持吗？

Are there other people in Silicon Valley who also have it a part of their mission to get it right? Do you feel supported in that mission by others?

很难具体指出某家公司，但我有时很感动。比如你去参加活动，或者另一家公司请你去演讲，观众中有人会后走过来或提问，说：“谢谢你们做事的方式。”是的，就像感谢你们努力把事情做对。有种感觉是，我认为我们关系中很棒的一点是，我和 Dario 也互相问责，对吧？所以当我们反复讨论“这对公司是正确的决定吗？这符合我们的价值观吗？”时，我们有彼此，整个联合创始团队仍在 Anthropic。我们还招聘了一群优秀的人，他们也为了使命而来。这几乎是一种集体问责，再加上 Dario 提到的内部文化——任何人都可以不同意我们，是的，这很难做到，但有很多场合鼓励他们这样做，他们的经理可以，他们也可以匿名。我认为所有这些特质都让我们更有可能在偏离轨道时至少能意识到。

It's hard to point to maybe like a particular company, but I think I am very touched by sometimes, you know, whatever you go and talk in an event, or another company wants you to come speak to them, and somebody in the audience, you know, comes up to you afterwards or ask a question, and they say something like, "Thank you guys for the way that you do things." Yeah. Like, thank you for trying to get it right. There's this kind of sense that, you know, I think um I think something that's great about our relationship is Dario and I I also think we hold each other accountable, right? And so, when, you know, we go back and forth on, "Is this a right Is this the right decision for the company is this in keeping with our values we have each other we have the whole co-founding team is still at Anthropic. I think we've also hired an amazing set of people who are also there for the mission. There's almost a kind of group accountability and if you pair that with like Dario mentioned this culture internally of anybody can disagree with us right and yes that's hard to do but there's a lot of forums in which they're encouraged to do that their manager can do it they can do it anonymously. I think all of those qualities make it more likely that that if we start to stray from the path we're at least going to know.

好的，你想让公众知道什么？在我们 AI 旅程的这个时刻，你最想让我们知道的是什么？

Okay and you want to leave us the public knowing what? What is the most important thing you want us to know in this moment in time where we are in our AI journey?

你知道，世界上正在发生一件非常重大的事情。这不是最新的技术，不是加密货币，不是移动互联网，也不是最新的小玩意或流行词。人类正在经历一件非常重大的事情。虽然这项技术是由少数人构建的，但我们需要让它为所有人服务，我们需要确保我们所有人共同做出正确的决定，并应对日常生活、工作、政府和公民自由的改变。如果我们以正确的方式做到这一点，所有这些都可能以积极的方式改变，所有这些都可能变得比我们留下的更好。我们可能拥有一个完美的世界，许多疾病被治愈，人们的生活更加快乐和有意义。但你知道，如果我们搞错了，很容易想象事情完全不会那样发展。这取决于我们所有人。有些人处于更核心的位置，但这也是我们需要改变的。这并不容易。不是单一的一件事。它将是许多不同的事情。我想我可能会留给人们和 Dario 所说的非常相似的东西，即光明与阴影的概念。人工智能的潜在好处是巨大的。我们写过相关文章，也试图公开谈论过。我们相信它们是真实的。同时，风险也是真实的，对吧？我们不是要炒作正面或负面。我认为两者都是真实的。情况很复杂，但我们必须同时把握两者。我想呼应 Dario 的话，这不是科技公司能独自完成的事情。也不应该如此。这需要比目前从事人工智能工作的人更广泛的群体合作。我认为第一步必须是不要害怕它。去了解它。投入其中。提出问题。保持批判。就像我们对员工说的那样，对吧？我们只有扩大参与对话的人群，才有机会把事情做对。

You know there's something really big happening to the world here. This isn't the latest technology, this isn't crypto, this isn't mobile, this isn't like the latest gadget or buzzword. Something really big is happening to humanity here. While the technology is being built by a small number of people, we need to make it work for everyone and we need to make sure that between all of us we make the right decisions and we reckon with changes to our daily lives, changes to our jobs, changes to government and civil liberties. And if we do this the right way, all of those things could change in a positive way, all of those things could be better than we left them. We could have the perfect world where many many diseases are cured, people's lives are more joyful and meaningful. But you know, if we get it wrong, it's easy to imagine things not going that way at all. And it's up to all of us. Some of us are in a more central position, but that's something I think we need to change as well. And it's not easy. It's not any one thing. It's going to be a bunch of different things. I think I would probably leave people very similar to what Dario said with this concept of light and shade. That the potential benefits from artificial intelligence are vast. We've written about them. We've tried to talk about them publicly. We believe they're real. And also the risks are also real, right? We're not trying to hype up the positives or hype up the negatives. I think they're both true. It's complex, but we have to hold two things at once. And I think just echoing Dario, this is not something that technology companies can do by ourselves. Nor should it be. It's something that's going to take partnership from a much bigger group of people than is currently working on artificial intelligence. And I think the first step there has to be don't be afraid of it. Learn about it. Lean into it. Ask questions. Be critical. Just like we say to our staff, right? We are only going to have a chance of getting this right if we expand the circle of people that are in the conversation.

我问所有嘉宾这个问题，尤其对你们两位感兴趣。你认为美好生活的意义是什么？AI 将如何为这种美好生活做出贡献？对你、你的家人和你的孩子来说，美好生活的意义是什么？

I ask this of all my guests and particularly I'm interested from the two of you. And what do you think is the meaning of a well-lived life? And how AI is going to contribute to that well-lived life. But what will be the meaning of a well-lived life for you, your family, and your children?

对我来说，美好生活是拥有自主权和自治权。你坐在驾驶座上，用它来成为最好的自己。我认为最好的自己是为了你自己、你的家人、你的孩子、你更广泛的社区，以及为了世界。我认为能够自我反思、从错误中学习、克服困难，然后走出来时想：“我更了解自己了，我喜欢自己，我想帮助别人。”对我来说，这是美好、内省生活的最佳定义。就像我谈到的，我认为存在一个世界，AI 可能破坏许多我认为是美好生活标志的东西。也存在一个世界，AI 能够实现这些，让更多人拥有意义和价值，进行自我内省，成为最好的自己，很好地相互支持。我希望那是我们要构建的世界。

I think for me a well-lived life is one where you have agency and autonomy. You're in the driver seat and you're using that in the service of being the best version of yourself. And I think the best version of yourself for you, for your family, for your kids, for your broader community, and I think for the world. I think the ability to be self-reflective, to learn from mistakes, to overcome struggles, and to come out the other side thinking, "You know, I know more about who I am, I like who I am, and I want to help other people." I think that to me is the best definition of a well-lived, introspective life. And like I talked about, I think there's a world where AI could damage so many of the things that I think are the hallmarks of a well-lived life. And I think there's a world where it could enable those, and it could allow a lot more people to be able to have meaning and value, and to self-introspect, to be the best versions of ourselves, to show up for each other well. And I hope that's the world that we're going to build towards.

我只想说和 Daniela 一样的话，只是换种说法。我认为这很简单。你知道，始终带着尊严和原则行事。当你回首往事时，你不应该为自己的所作所为感到羞耻，或者说“我希望我做了别的事”。当然你会犯错，但你知道，你付出了每一分努力去把事情做对。我认为我们对个人、对我的联合创始人、对我生活中的其他人都有责任，也对我们的技术用户、对国家乃至世界上的所有公民都有责任。所以，我认为我们对所有这些人都负有责任，必须关注。确保最终它指向某个目的，某个我们试图在世界上实现的一致观点，那个我们试图实现的愿景。也许它没有实现，也许我们搞砸了，也许我们犯了可怕的错误，但只要有清晰的愿景并努力以正直的方式追求它。如果你能做到，当然你也想要结果，但我认为做到这一点是你唯一能控制的，而且最终比结果更重要。

I'd just say the same thing as Daniela in different words. I think it's pretty simple. I think you know, just operate always with a sense of dignity and principle. Like, you know, when you look back and you're like, you know, you shouldn't when you look back be ashamed of what you did or say, you know, I wish I'd done this other thing. Of course you'll make mistakes, but you know, you were putting every iota of effort you could into trying to get it right. I think we have duties both to individuals, to my co-founders, to all the other people in my life, but also to the users of our technology, to everyone else in the country and world as citizens. So, I think we have duties to all those people that we have to attend to. And making sure that at the end of it it was directed at some purpose, some consistent view of what we were trying to accomplish in the world, that the vision we were trying to make happen. Maybe it doesn't happen, maybe we screw it up, maybe we make terrible mistakes, but just having the vision be clear and trying to pursue it with integrity. If you can do that, of course you want the outcome too, but I think doing that is all you can control and more important ultimately than the outcome.

感谢你们引领我们前进。你们正在做小时候梦想的事情。

Thank you for growing us forward. You're doing exactly what you envisioned as little kids.

这很奇怪，但确实如此。是的，没错。谢谢 Dario 和 Daniela。你们给了我们很多思考。祝大家一切顺利。一切顺利。你可以在 YouTube 上订阅 Oprah 播客，并在 Spotify、Apple Podcasts 或你收听的任何地方关注我们。下周见。谢谢大家。

It is very strange, but we are. Yes, it is. Thank you, Dario and Daniela. You've given us a lot to think about. Go well, everybody. Go well. You can subscribe to the Oprah podcast on YouTube and follow us on Spotify, Apple Podcasts, or wherever you listen. I'll see you next week. Thanks, everybody.