---
author: All-In Podcast
date: '2026-09-26'
guest: ''
layout: post.njk
source: https://www.youtube.com/watch?v=cvP_1jmnkmM
speaker: All-In Podcast
tags:
- ai-development
- open-source-model
- life-science-research
- risk-assessment
title: 播客回顾与人工智能责任讨论：从峰会幕后到未来市场趋势
summary: 本文回顾了一期播客中关于大型会议的幕后回顾，探讨了人工智能发展对国家对话的影响，以及开源模型和生物技术领域的前沿研究。核心讨论集中在如何通过技术发展来应对对人工智能的恐慌，以及在生命科学领域利用人工智能进行蛋白质发现和基因编辑的潜力。
insight: ''
draft: true
series: ''
category: science
area: knowledge-meta
project: []
people: []
companies_orgs: []
products_models: []
media_books: []
status: evergreen
---
### 播客回顾与人工智能责任讨论
**Jason**: 好了，各位，请。请等到第290期最佳剧集、全球最佳播客——包罗万象的播客。再次对我来说，这是主要四位，与我们David Friedberg、Chamath Palihapitiya和David Sacks。我是你们的主持人，Jason Calacanis。我们今天一整天都很忙。日程。然而，如果我不寻求大家一起分享他们最愉快的第五个记忆，即由David Friedberg先生组织的年度全面峰会，那将是我的疏忽。David，你有什么想法？你如何评价你作为包罗万象峰会的管理者、制作人的工作？你有什么我们不知道的吗？他们做了什么，他们忘记了什么？制作？在最后第二天，我们没有表达很多我们衷心的感谢给制作团队，Kimber、Lizzie。我做了。缺口。我认为我们……
Original English
**Jason**: All right, everybody, please. Please stand by for episode 290 of the besties, the best podcast in the world, the all-in podcast. Again, for me, this is the main four, with us, David Friedberg, Chamath Palihapitiya, and David Sacks. I'm your host, Jason Calacanis, and we've been busy all day today. Schedule. However, it would be it would be remiss of me not to seek to have everybody share their their most delightful fifth memory of the annual all-in summit organized by Mr. David Friedberg. David, what do you have? What are your thoughts? How do you rate your job as the as the manager, the producer, the all-in summit? What do you know that we don't know? What did they do that they forgot to. Produce? On the last second day, we didn't express a lot of our heartfelt thanks to the production team, Kimber, Lizzie, I did. Gap. I think we...
**David Friedberg**: 不，Jason，好吧，我认为我们应该邀请他们，但我只想表达巨大的感激之情。注意……把它们带出来。幕后情况是太棒了，真是太好了。今年依然如此，然后，实际上，玩得很开心。你知道，我最喜欢的部分——这游览环球影城工作室。你没事。你正在消磨时间。它完美。这是一场精彩的演出。我又开始担心了。你的安息日没有童年伤病。顺便说一下，我要离开了。向我们的朋友们问好，来自德龙。我购买了其中一台咖啡机，我们当时穿着西方。我把它放在厨房里。你知道我买了什么吗？我不知道你们是不是……看到了，但在贵宾区。那里有一台饮水机。明智之举，零。这是一台饮水机。不含塑料。它通过超滤。在没有水微塑料。现在我家也有一个。办公室。非常感谢他们。谢谢。
Original English
**David Friedberg**: No, Jason, well, I we thought the point was to invite them, but I just want to express great great gratitude to the note... to bring them out on stage. The behind the scenes situation was was fantastic, and it was great again this year, and then, actually, had a lot of fun. You know, my favorite part of this -- this tour of Universal Studios. You're fine. You're killing time. It's perfect. It's a great show. I'm getting worried again. Your sabbatical has no childhood injuries. By the way, I'm going to leave. Say hello to our friends from Delonghi. I bought one of the coffee machines, which we were wearing in the West. I put it on. In the kitchen. Do you know what I bought? I don't know if you guys... saw it, but in the VIP section. There was a water dispenser. Smart move, zero. It's a water dispenser. No plastic. He through ultra filtration. In no water microplastics. Now I have one at home too. Office. Thank you very much to them. Thank you.
**Jason**: 是的。很棒的公司。Sacks，你有什么最喜欢的时刻？
Original English
**Jason**: Yeah. Great company. Sacks, do you have a favorite moment of the show?
**Sacks**: 嗯，我也这么认为。节目的亮点在于……我们仍将铭记很久以前，就接到过一个电话。总统期间Jason的表演。我认为总的来说，Jason的内容和行为真的有助于平静下来这个国家对人工智能感到恐慌。感觉我们好像在这个全国对话的中心，而且我认为这确实发生过很多事。有机地。你是否记得？在峰会前最后几周，这项科学研究是进行的操纵，让所有人相信如果人类灭绝……我们不会那样做。他们想要什么？民主党人，而不是咱们就此打住吧。人工智能的发展。进而，你知道，首先Sasha出来后，就像……他安抚了所有人的情绪，并且然后Jason就这么做了。确切地。还有总统打电话，他做了他必须做的事。结果是最好的，而且刚刚拍下了全部内容对控制叙述，我确信这些团体自杀式炸弹袭击者谨慎行事计划已制定数周，仿佛他们把这一切都撕成了碎片。手术。
Original English
**Sacks**: Well, I think so too. The highlight of the show was... We'll still remember a call that came in a long time ago. During the presidential Jason's performance. I I think that overall, the content and the behavior of Jason really helped calm down the country's panic about AI. It felt like we were at the center of this national conversation and I think that that really happened a lot. Organically. Do you remember? In the last few weeks before the summit, this scientific study was conducted to manipulate everyone into believing that if humanity went extinct... We wouldn't do that. What do they want? Democrats, rather than let's just stop there. The development of AI. And then, you know, first Sasha came out, and it was like... He calmed everyone's nerves, and then Jason did it. Exactly. And the president called and he did what he had to do. The outcome was the best, and it just captured the whole thing to control the narrative, and I I'm sure that these groups of suicide bombers were careful with their plans that had been made for weeks, as if they tore it all apart. Operation.
**Jason**: 无聊够了，返回工作。你有一个超级聪明特朗普的侦察兵——现在这个学期。我们将叫他特朗普的未来球探，但我认为你你你你你你你是对的。许多人们想知道，是Friedberg计划的吗？有预定通话特朗普总统Friedberg？让我们来看，让我们回顾过去。后台。基本上，Jason直接联系了他幕后，他他说：“嘿，“请给我回电话。”然后他我给他打了电话，当时他还在台上。它们有点对应。你知道，那是完全是自发的。
Original English
**Jason**: Enough of the boredom, back to work. You have a super smart Trump's scout -- this term now. We're going to call him Trump's future scout, but I think you you you you you you you're right. A lot of people want to know, was Friedberg? Planned. Was there a scheduled call with President Trump Friedberg? Let's look, let's go back. Backstage. In basically, Jason reached out to him directly behind the scenes and he he said, "Hey, "Please call me back." And then he I called him, and he was still on stage at the time. They kind of corresponded. You know, that was completely spontaneous.
**Sacks**: 是的，的确如此。自发地。那不是什么事计划中，Jason说：“嘿，我们来做点什么吧整个”。所以就是这样。真是令人惊讶这段对话，Jason正在等待总统现场直播正在播出，这显然是当天的话题是……华盛顿，以及在新闻媒体，而且是很高兴看到一切都这么顺利。这件事正在发生和我们一起在台上。事情就是这样。
Original English
**Sacks**: Yes, it was. Spontaneous. That wasn't something that was planned, and Jason said, "Hey, let's do something about the whole thing." So that's it. It was amazing to have this conversation, and Jason was waiting for the president to be live on air, and it was clearly the hot topic of the day in... Washington, and in the news media, and it was great to see that everything went so smoothly. This thing was happening with us on stage. That's how it was.
**Jason**: 当然，我当时是这么想的。这有点奇怪，而且我想法：“好，放大声地对他联系。还有Jason他问我：“你是怎么做到的？”“你在做吗？”而我只是我把它递给了他（我）我总是会多给一个。以防万一，带个麦克风。如果有人逃脱命令）。我只是把它放上免提通话麦克风。和特朗普总统最搞笑历史上最伟大的总统最搞笑的政治家在历史上，你Sacks注意到他说话非常流畅一句话喇叭？
Original English
**Jason**: Of course, I was thinking at the time. It was a little strange, and I thought, "Okay, put it on speakerphone to him. And Jason asked me, "How did you do it?" "Are you doing it?" And I just I handed it to him (I) I always give an extra one. Just in case, bring a microphone. If someone escapes the command). I just put it on speakerphone microphone. And President Trump the funniest the greatest president in history the funniest politician in history, you Sacks noticed that he spoke very fluently in one sentence on the speakerphone?
**Sacks**: 理想情况下。
Original English
**Sacks**: Ideally.
**Jason**: 正确。Jason最好的薯片。全部他们正在尝试重建。不可能的。不可能的。但他做不到了解如何按下按钮喇叭。好的？就像：他怎么样了？它能做到这一点吗？他怎么样了？做？他只是说得完美短语。但是Chamath，如果说真的，无聊上星期他们简直要疯了，而且，呃，大人们来了。爸爸来了，他来了。Elon，Jason来了。，所有成年人走进房间，然后他们只是简单地说：听着，如果你你有软件规定有缺陷且未准备好在发布到全世界之前，或许不是发布它世界，但要让它有点，你知道，你你喜欢说，少一些脆弱的。所以，你的关于我们身处何方思考我们身处其中国家的讨论，从放慢速度，或许只是释放耐受性非脆弱性软件软件。
Original English
**Jason**: Correct. Jason's best chips. All they're trying to rebuild. Impossible. Impossible. But he couldn't understand how to press the button on the speakerphone. Okay? It's like: How is he? Can it do that? How is he? Do? He just spoke perfectly phrased. But Chamath, if seriously, the boredom last week they were going crazy, and, uh, the adults came. Dad came, he came. Elon, Jason came. , all the adults walked into the room, and then they just simply said: Listen, if you you have software that is flawed and not ready before it's released to the world, maybe don't release it to the world, but make it a little bit, you know, you you like to say, less fragile. So, your thoughts on where we are in the national discussion, from slowing down, maybe just releasing tolerant non-fragile software.
**Chamath**: 嗯，我也这么认为。上星期发生分类年轻的组织想这样称呼自己实验室，但是实际上有拥有损益表和损失、股东和必须遵守这些和其他人一样，遵守同样的规则。其他的。然后是那些复杂的公司对许多人来说几十年处于仔细观察并自己苦涩从经验中学到创建过程可靠的解决方案和创作成本不可靠的解决方案。然后我的意思是？和一方面，你在许多方面都看到了这一点，奇怪的关系，时代创造奇怪的同伴，或者错误。你有一群人：Satya、Jason、Sacks、Zuckerberg、Lynn Musk、Lynn公司总裁Khan。我们拥有自己的语言。责任。产品质量。在美国存在这种观点，责任。产品质量。她被称为为了防止公司做了什么他们不应该做的事情，尤其是如果他们知道这些事情可能会有害。而且另一方面，这些人也是人，做我们所做的事的人，我觉得他们有点天真。并实质上表示：“我们不是一家公司，我们是实验室。保护我们。”而我我认为我成熟的答案是可用性、内部控制和彻底测试。还有什么Zuck早些时候说过，而且Elon说过的话过去：“我们只是我们处于仔细观察。”所以，当事情进展不顺利时完美，我们只是我们放慢进程。Elon用FSD做到了。你Jason说的。他可能在FSD 3中或者说四年前，但他公开声明：“我们我们负担不起这个允许，因为每一起事故特斯拉也加入了进来。增加比那多一千倍，任何还有哪家公司？所以我们就需要搬家慢点。过去的Zuckerberg上周说：“我们我们将发射缪斯”，看起来像最成功的产品发布之一最近，实际上有后果极其严重，关于我们应该……说话。而对他来说不得不放慢速度。它。对他来说Alexander，我和他过去曾说过晚上，我不得不放慢速度几个月到使其理想的。就是这样。在你的掌控之中。唯一的问题是，你决定怎么做？嗯……我想是最有趣的观察。我们必须停止请列出这些公司实验室，仅此而已。公司，我们有以相同方式进行评估标准，包括责任产品质量。
Original English
**Chamath**: Well, I think so too. Last week there was a classification of young organizations that want to call themselves labs, but actually have P&Ls and losses, shareholders, and have to abide by these and the same rules as everyone else. Others. And then there are those complex companies that for many people for decades have been under scrutiny and have learned their own bitter lessons from experience in creating reliable solutions and the cost of creating unreliable solutions. And then what do I mean? And on the one hand, you see this in many ways, strange relationships, times create strange bedfellows, or wrong. You have a group of people: Satya, Jason, Sacks, Zuckerberg, Lynn Musk, Lynn company president Khan. We have our own language. Responsibility. Product quality. In the United States there is this idea of responsibility. Product quality. She is called to prevent companies from doing what they shouldn't do, especially if they know that these things can be harmful. And on the other hand, these people are also people, people who do what we do, and I feel that they are a little naive. And essentially say: "We are not a company, we are a lab. Protect us." And I I think my mature answer is availability, internal controls, and thorough testing. And what Zuck said earlier, and what Elon said in the past: "We are just we are under scrutiny." So, when things don't go perfectly, we just we slow down. Elon did it with FSD. You Jason said. He might have been in FSD 3 or four years ago, but he publicly stated: "We we can't afford this permission, because every accident Tesla is also involved. Increase a thousand times more than that, any other company? So we just need to move slower. The past Zuckerberg said last week: "We we will launch Muse", which looks like one of the most successful product launches recently, actually has extremely serious consequences, about what we should... talk. And for him to have to slow down. It is. For him Alexander, I and he have said in the past at night, I had to slow down for a few months to make it ideal. That's it. It's in your control. The only question is, what do you decide to do? Well... I think it's the most interesting observation. We must stop listing these companies as labs, and that's it. Companies, we have to be evaluated by the same standards, including responsibility for product quality.
**Jason**: 是的，就是这一刻，Sacks，有很多照亮。有趣的，一些虚假媒体他们说了类似这样的话：在流行的播客。进而它们繁殖剪辑，但只是不敢提及播客。足够的有趣的。我觉得Chamath已经这引导我们第一个主题，即我们是否会有个人责任作为我们的北极星，或某些集体行动。而总是这样效果最好。个人责任，无论这是否符合标准？个人，或公司。合适的所需的解决方案接受每个情况是……那么，我们应该这样做吗？发布此产品本身，而不是事实上我们能否强行全世界都同意吗？而问题在于谈论这些集体方法是因为他们正在被移除责任实际做出决定。我认为人们感觉到了，你知道的。就像这周的Dario和Sam去了联合国，他们说我们需要一些全球管理人工智能。这有点像我有一些亿万富翁飞往达沃斯，他们乘坐私人飞机，然后反对气候变化。你知道这其中必有缘由。从根本上不切实际的脱离现实。
Original English
**Jason**: Yeah, it's this moment, Sacks, there's a lot of illumination. Interesting, some fake media they said something like this: In the popular podcast. And then they breed clips, but just don't dare to mention the podcast. Enough of the interesting. I feel like Chamath has already this leads us to the first topic, which is, are we going to have individual responsibility as our North Star, or some collective action? And it always works best. Individual responsibility, whether that's to the standard of the individual, or the company. The appropriate solution required to accept every situation is, well, should we release this product as is, versus, in fact, can we force the entire world to agree? And the problem with talking about these collective approaches is because they're being removed from the responsibility of actually making the decision. I think people feel that, you know. It's like this week, Dario and Sam went to the United Nations and they said we need some global governance of AI. It's a little bit like, I have some billionaires flying to Davos on their private jets and then arguing against climate change. You know, there's something fundamentally impractical and detached from reality about this.
**Sacks**: 虚伪。
Original English
**Sacks**: Hypocrisy.
**Jason**: 虚伪。这就好像：各位，注意听！我们需要你被智者所接受，决定什么生产。如果该产品很危险。也就是说，它不可靠。或者表现然后，出乎意料地不要放手。别再看了，一些全球性的管理决定。因为如果你们不，你会做出这个决定吗？对，不会有意义将使联合国。以及在当时联合国达成共识，我们将非常此外，所有这些旅行。所以，确实存在一些问题。方式很奇怪，主要角色继续谈话：“嗯”。因为真的人们会产生这样的印象，他们正在尝试排除责任。
Original English
**Jason**: Hypocrisy. It's like: Listen, folks, we need you to be accepted by the wise to decide what to produce. If the product is dangerous, meaning it's unreliable or performs, then unexpectedly, don't release it. Stop looking for some global governance decision. Because if you don't, you're going to make this decision. Right, there's no point in making the United Nations. And by the time the United Nations comes to a consensus, we're going to be very, plus all this travel. So, there's definitely some ways that are very strange that the main characters continue to talk: "Hmm." Because really, people will get the impression that they're trying to exclude responsibility.
**Chamath**: 嗯，理论是这样：我认为这值得讨论，实验室秘密地……它们是公司。它们……
Original English
**Chamath**: Well, the theory is this: I think it's worth discussing what the boundaries of the labs secretly... They're companies. They're...
**Jason**: 是的，是公司。让我刚结束……
Original English
**Jason**: Yes, they're companies. Let me just finish...
**Chamath**: 不，不，不，但是……Chamath做得非常好。注意，我想要确保我们千万别错过！……好的。它们是公司。
Original English
**Chamath**: No, no, no, but... Chamath made a very good point that I want to make sure we don't miss! ... Okay. They're companies.
**Jason**: 是的。当你打电话给他们的实验室时，听起来像非商业性组织，以及这谬误。
Original English
**Jason**: Yes. When you call them their labs, it sounds like a non-commercial organization, and this is a fallacy.
**Chamath**: 好的，那么这些公司生产这些产品，期望，特朗普总统专门为他们提供他们互联网上拥有的东西。公司设有第230条保护他们免于责任。那么政府方面呢？特朗普，第47任行政部门，或许有这种想法，可以提供这些公司具有一定的利用杠杆。他们捐出10%的主权国家股份给美国福利基金等等。然后是这个，这将导致以下事实：他们将被给予某种放纵的结果，归根结底是他们的责任。它疯狂。这就像给予一些航空给这家公司，将其10%的股份转让给美国政府，不存在的可能性携带责任，他们的陨落飞机从天而降。但是什么你正在考虑这件事？阴谋论？我，我相信你会喜欢听到它，Sacks。这些是什么寻找盾牌。你创造给人一种印象，这是一种想法特朗普政府。这并非事实。每个特朗普政府。这就是他们提供公司Frontier Labs {斜杠}，以及什么可以有传言说，在行政部门可以成为那些对此持开放态度。就这些了。间接渠道。
Original English
**Chamath**: Okay, so these companies produce these products, expecting, President Trump specifically provided them with what they have on the internet. Companies have a section 230 that protects them from liability. So what about the government side? Trump, the 47th administration, has perhaps this idea that it can provide these companies with a certain amount of leverage. They donate 10% of the sovereign wealth fund shares to the United States, etc. And then this, this will lead to the fact that they will be given some kind of indulgence. The result is, in the final analysis, their responsibility. It's crazy. It's like giving some airline to this company that transfers 10% of its shares to the government of the United States, there is no possibility of carrying responsibility for their fallen planes falling from the sky. But what are you considering this? Conspiracy theory? I, I believe you will like to hear it, Sacks. What are these looking for shields? You create the impression that this is an idea of the Trump administration. This is not true. Every Trump administration. This is what they provide. Companies, Frontier Labs {slash}, and what can be rumored that in the administration can become those who are open to this. That's all. Indirect channels.
**Sacks**: 是的，我们目前知道的情况是这样。上次就是这么回事。差不多一周。每位官员特朗普政府他开口说道类似这样的话，我刚才说了什么？具体而言：首先，人工智能公司，情报应该承担责任，他们产品，以及我们绝不放弃。责任产品。我认为发言人说了这番话。Johnson，国会议员。Bassin说了这番话，最后一次面试。大家都说如果你的目标是获得拒绝责任产品质量或反垄断责任，如果你尝试建立卡特尔，如果你尝试参与阴谋，算了吧。我们，我们不会这么做。所以，我认为他们是完全拒绝整个想法。和记住这一点。特朗普总统今天表示推特：我们有司法部。这是一道栅栏。行政管理是栅栏。行政国家，平民诉讼、刑事诉讼——所有这些都存在。作为一种潜在的责任，为了确保你做得对。妥协风险和报酬。再次总之，最后，如果这些人会放手危险产品，这无关紧要，你同意什么？联合国。因此，我们需要：以便他们意识到责任。
Original English
**Sacks**: Yes, that's what we currently know. That's what happened last time. Almost a week. Every official of the Trump administration, he opened his mouth and said something like what I just said? Specifically: First, AI companies, intelligence should bear responsibility for their products, and we will never give up. Responsibility for products. I think the spokesperson said this. Johnson, Congressman. Bassin said this last interview. Everyone said if your goal is to get rejection of responsibility for product quality or antitrust responsibility, if you try to build a cartel, if you try to participate in a conspiracy, forget it. We, we won't do that. So, I think they are completely rejecting the entire idea. And remember this. President Trump said today on Twitter: We have the Department of Justice. This is a fence. Administration is a fence. Administration of the country, civilian litigation, criminal litigation—all these exist. As a potential responsibility, to ensure that you do it right. Compromise risk and reward. Again, in short, finally, if these people will let go of dangerous products, it doesn't matter, what do you agree with? United Nations. Therefore, we need: so that they realize responsibility.
**Jason**: 听着，拜托。坦诚相待。作者是我。你有内部人士信息。他们正在尝试这个大厅？他们是……你认为他们会吗？游说获得这种保障吗？是还是不？因为……
Original English
**Jason**: Listen, please. Be frank. The author is me. You have insider information. They are trying this lobby? They are... Do you think they will? Lobby to get this kind of protection? Yes or no? Because...
**Sacks**: 嗯，我听说……情况并非如此亲身经历的人。他们没有问。我心想，“嘿，我们想拒绝责任产品质量。”但我从一些人那里听说华盛顿，他们渴望某事差不多就是这样。
Original English
**Sacks**: Well, I heard... It's not the case personally. People. They didn't ask. I thought, "Hey, we want to reject responsibility for product quality." But I heard from some people in Washington that they are eager for something that's pretty much like this.
### 关于“实验室”与责任的讨论
**Host**: 很难确定。确定，对吧？
Original English
**Host**: . Hard to determine. Determine, right?
**Freiberg**: 是的。好的，完全同意。
Original English
**Freiberg**: Yes. Okay, totally agree.
**Host**: 没错。弗里贝格，你在这里的想法。我们将继续讨论。这里第一个故事是：代币到达的地方，最高价和价格跌倒，但我会给你最终结论。美国应该与，从互动角度来看，与那些公司，假装是实验室，但不是，可以得到所有权，自己的作品，弗兰肯斯坦。我，我认为是J.D. Vance，我认为他们有令人印象深刻的外观。他意志坚定，看起来很漂亮。如果我以总统身份出现，我来描述一下。你扔了很多东西。快速球。他打败了他们，场地边界。顺便一提，你知道，快速球。全垒打想要快速球。你不是，你可以击中快速球。他已经不在状态了，罢工。所以，我给了他三岁了，最难的，仅他能想出办法，而且他继续倾斜。三人全部翻过围栏，事情就是这样。来自J.D.的好评。它，非常，不，周日节目，展示节选。我问他的问题。问。砰，砰，砰。自由鸟直道，透过栅栏，但是，你知道，他说，通过，本质上，如果你创造弗兰肯斯坦，就是它了。想法，停止。然后，他又补充道：“如果您已经发布，事情败露了，然后创造反弗兰肯斯坦，本质上。创造预防措施。”
Original English
**Host**: Exactly. Freiberg, your thoughts here. We're going to continue to discuss. The first story here is: where tokens arrive, the high price and the price falls, but I'll give you the ultimate conclusion. The United States should, from an interactive perspective, with those companies, pretending to be labs, but not, that can get ownership, of their own work, Frankenstein. I, I think it was J.D. Vance, I think they had an impressive look. He was resolute, looked good. If I were to appear as president, let me describe it. You threw a lot of stuff. Fastball. He beat them, the field boundary. By the way, you know, fastball. Home run wants fastball. You're not, you can hit fastball. He was out of the zone, strike. So, I gave him three years old, the hardest, only he could figure out, and he continued to tilt. All three went over the fence, and that's how it was. Good review from JD. It, very, no, Sunday show, show excerpt. The questions I asked him. Ask. Bang, bang, bang. Free bird straight, through the fence, but, you know, he said, by, in essence, if you create Frankenstein, that's it. Idea, stop. And then, he added, "If you've already released, things have gone wrong, then create anti-Frankenstein, in essence. Create precautionary measures."
**Freiberg**: 是的。这完全合乎逻辑。我认为扎克给了一次不错的采访，关于这件事。他们不要操之过急，新闻稿，星期。他们说，他们想做所有事，正确的。他们想确认一下，它是安全的，它是不给人们做事的机会，坏事。它，责任，他们承担了。它本身就是一家公司。任何创造者，模型，必须采取，我自己也是如此，责任。适当地。而且，很显然，这一点，该组织，明白通过付费，一系列罚款，加利福尼亚州……我们有法律。我们有法律。我们有很多法律，禁止这样做，很多事情。好了，我们走吧。继续前行。在我们有法律，好吗？
Original English
**Freiberg**: Yes. That's perfectly logical. I think Zuck gave a good interview about this. They're not rushing the press release, the week. They said, they want to do everything right. They want to make sure, it's safe, it's not giving people the opportunity to do bad things. It, responsibility, they took it. It's a company in itself. Any creator, model, must take, myself included, responsibility. Appropriately. And, obviously, this point, the organization, understands by paying, a series of fines, California... We have laws. We have laws. We have a lot of laws, prohibiting doing, a lot of things. Okay, let's go. Move forward. In we have laws, okay?
### 竞争与安全：哲学时刻
**Chamath**: 对了，我可以吗？加一，哲学时刻，在此之前？所以，我……我，我认为这是不言而喻的。嵌入其中，我们姑且称之为，这个，达里奥的立场，那个，我们姑且称之为，可调节自由，我们拥有的市场。其基本声明是，因为这会导致，陷入竞相压价的恶性竞争。这竞争带来，安全面临风险。他的所有博客里都有提到。以及其他许多人，体现这一理念，什么比赛，等于缺席，安全。这是不言而喻的。对我们自身的批评，经济体系。这基本上是左翼。对我们自身的批评，系统。它到达，遥远的历史。我，我记得那些日子。冷战时期，我们真的有过这种经历，两人之间的决斗，不同的系统。你，你想要一个系统吗？自由市场或，共产主义制度？许多人，声称，资本家，系统会这样做，诸如威胁之类的事情，安全与美丽。我们的建筑总是，会更丑，因为贪婪，资本家不会想要，花一分钱，为了改进或，美丽的建筑，等等。实际上，结果证明，确实如此。苏联体制，制造寒冷，灰色的风景。为什么？因为人们想要，美丽。他们想要，安全。因此，如果你还好吗？运转良好的市场，然后是经济。给予人们他们，他们想要。和，明显地，美国体系，赢得了这场战斗。所以我就，从根本上不同意，考虑到这一点，竞争意味着，缺乏安全保障。正如扎克伯格和，詹森，有，巨大的激励措施，例如，从积极方面来看，以及从，负面，为了安全起见。和，积极的一面是，没有顾客想要，该产品，不可预测的，不可靠，这并非，他们做他们该做的事，承诺。没有任何，公司不希望，特工，即将泄露，他们的数据或，入侵竞争对手。以及从负面角度来看，各方，正如明确所述，了解总统，他们有很多，责任。责任，产品质量，民用，责任，行政，责任，刑事，责任。所以，我断然否认，我不同意这种说法。竞争的理念，——那很糟糕。竞争是，这件 事能做到真是太好了。用于，成就，正确的结果。
Original English
**Chamath**: Right, can I? Add one, philosophical moment, before this? So, I... I, I think it's implicit. Embedded in, let's call it, this, Dario's position, that, let's call it, adjustable freedom, market that we have. The underlying, statement is, because this will lead to, a race to the bottom. And this, competition brings, safety at risk. It's in all his blogs. And many others, embody this idea, what competition, equals absence of, safety. It's implicit. A criticism of our own, economic system. This is fundamentally left-wing. A criticism of our own, system. It reaches, far back in history. I, I remember those days. Cold War days, where we really had this, duel between two, different systems. Do you, do you want a system of, free markets or, communist system? And many people, claimed, that the capitalist, system would do things, such as threaten, safety and beauty. Our buildings would always, be uglier, because greedy, capitalists wouldn't want to, spend a dime, for improvement or, beautiful buildings, etc. And actually, as it turned out, it did. The Soviet system, created cold, gray landscapes. Why? Because people want, beauty. They want, safety. And so, if you have a good, functioning market, and then an economy. It gives people what they, what they want. And, demonstrably, the American system, won that battle. So I just, fundamentally disagree, with this idea that, competition means, a lack of safety. As Zuckerberg and, Jensen, have, enormous incentives, like, on the positive side, and on the, negative, for safety. And the positive side is, no customer wants, the product to be, unpredictable, unreliable, that it doesn't, do what it's supposed to, promise. No, company doesn't want, agents, to be leaking, their data or, hacking competitors. And on the negative side, parties, as explicitly stated, understand the president, they have a lot of, liability. Liability, product quality, civil, liability, administrative, liability, criminal, liability. So, I just flatly deny, I disagree with this statement. The idea of competition, -- that's bad. Competition is, this thing that's great to do. For, achieving, the right outcomes.
**Chamath**: 就是这样，查马特，好好开发它。我们来自帕洛阿尔托的朋友尼科什，网络，模块的朋友，他释放过去，大约一周左右，42号单元。他正在这里建造。商机，创建，产品，连续的，网络防御。而且，模块的一位朋友，乔治，来自CrowdStrike的Kurtz，他同时还发布了猎鹰，你的新产品，用于网络防御。查马特，如果这些人，来自Frontier Labs/公司，我们非常担心，为什么网络安全如此重要？他们从未，发布产品，网络安全？以及帕洛阿尔托，Networks和CrowdStrike的这项，参与其中，以及，正在聚集。它，可能更多，第二个主题我们，我们来谈谈，具体来说，是什么，因为这些模型，精华聚集在一起，根据各种可能性，没有区别。因此，它们将会是，被迫搬迁，堆栈中位置更高的部分。但，我们等等看，此时。在，你只有一分钟时间，你想要哪一个？表达，然后我们，我们来看第一个。方向。
Original English
**Chamath**: That's it, Chamath, build on that. Our friend Nikosh from Palo Alto, Networks, friend of the module, he released over the last, week or so, Unit 42. He's building here. Business opportunities, to create, products, continuous, cyber defense. And, a friend of the module, George, Kurtz from CrowdStrike, he also released Falcon, your new product, for cyber defense. Chamath, if these people, from Frontier Labs/companies, we're so worried about, why is cybersecurity so important? Have they never, released products, cybersecurity? And this from Palo Alto, Networks and CrowdStrike, being involved, and, gathering. It, probably more, the second topic we, we talk about, specifically, what, because these models, essence gather together, according to various possibilities, no distinction. Therefore, they will be, forced to relocate, higher up in the stack. But, we'll wait and see, at this time. In, you only have a minute, which one do you want? Expression, and then we, we look at the first one. Direction.
### “实验室”与“公司”的界限
**Freiberg**: 这就是我做的事。片刻，我想要哪一个，表达，而不是，哲学和，社会学，即，我们最后一次听到，“实验室”一词，民众，意识清醒，冠状病毒。而且还有，实验室，是造成泄漏的责任人，病毒，然后，造成1500万人死亡，世界各地的人们，并对……造成了损害，45万亿美元。这是一个实验室，武汉。而且我认为，重要的是要认识到，在，这个问题并非，缺乏透明度或，责任。我，我认为这部分是……因为人们，感觉，舒适，打电话，它本身就是一个实验室，正是因为这个原因。事实上，其中许多，年轻人，在这些地方工作，幸存下来了，而且不是，绝对看到了，没有任何，责任，就实验室而言，导致了这次泄漏。现在我们来谈谈，关于各种事物的起源，物种，我们有，组织和，也，正在尝试分发，他们自己去实验室，但他们并非如此，实验室。它，公司。我只是，我想提醒大家，什么是商业，这些公司，吞食了数百，数十亿 美元，债务和所有权，资本，有，成千上万，来自股东的，数十亿 美元，市场，资本化。如果，我当时在他们的船上。我不想去那个地方，称呼自己，任何实验室，什么价格？而且我认为，如果你看到他们的，最近的演讲，组织，联合国，他们仍然是，使用此，术语。是的。这些不是实验室。它，商业的，这些公司，承担风险，责任，产品质量，以及，他们必须采取行动，根据。
Original English
**Freiberg**: This is what I do. For a moment, which one do I want, expression, instead of, philosophy and, sociology, which is, the last time we heard, the word "lab", in the public, consciousness, was coronavirus. And there was a lab, that was responsible for leaking, the virus, and then, causing 15 million deaths, of people around the world, and doing damage to the tune of, $45 trillion. That was a lab, in Wuhan. And I think, it's important to recognize that, in, this issue is not, a lack of transparency or, accountability. I, I think part of it is... because people, feel, comfortable, calling, it a lab in itself, for that very reason. The fact that many of these, young people, working in these places, survived, and are not, absolutely seeing, any, accountability, as it relates to the lab, that caused the leak. Now we talk about, the origin of various, things, species, we have, organizations and, also, are trying to distribute, themselves to labs, but they're not, labs. It's, companies. I just, I want to remind everybody, what is the business, these companies, have eaten up hundreds, of billions of dollars, of debt and ownership, capital, have, thousands, of billions of dollars, of market, capitalization from shareholders. If, I was on their boat at the time. I wouldn't want to go to that place, calling myself, any lab, at what price? And I think, if you see their, recent presentation, organization, United Nations, they're still, using this, terminology. Yes. These are not labs. It's, commercial, these companies, taking risks, responsibility, product quality, and, they have to act, accordingly.
**Chamath**: 是的，我们在这里已经一年了。我们来讨论一下这个问题。例如，如果这样，这种物质是这样的，危险，你不要，需要它，披露，或者，你知道，我的，提供，例如，“知道”，您的客户，“制定法律。”不要与他人比较，武汉的实验室。这个，嗯，查马特，这个，公关效果不佳，但是，我也认为这一点，他们中的一部分，程序，道德绑架，当他们就是不这样做的时候，想说的是，我们是公司，我们拥有，商业激励。他们总是想要，也就是说我们或，研究，机构，或者他们说，类似这样的东西，我们是一家公司，公共利益，虽然没有人真正，完全不知道这到底是什么，拿走，意思是。所以，这种铜绿，道德信号，隐藏了什么，实际上，正在发生。她尝试了，隐藏单词，“公司”，但，从法律上讲，这并非如此。可以隐藏。我认为他们有，会在以下情况下这样做……我们都挺过来了。公司。一切，各位，请保持非常……一切都要非常，小而强大，是一家公司。我认为我们应该，称之为，“一切皆为公司。”这是“万物实验室”。这是一个公司，我们，讨论各种，公式。不，我们，尝试，想法。这是大师之作。表情包课。但你知道吗？这些想法，变得如此，危险之处在于我们，我们不想携带，责任，你的想法。所以，欢迎来到，播客“一切都是，实验室”。在这里，当我们，讨论这个话题，零，弗里伯格，有吗？在科学家们看来，这是一种方式，思考，这，你知道，需要，突然停止，以及，然后公司，仿佛在夺取权力，从你的手？也许，他们可以，推迟，这里有实验室，而且，然后说：“嘿，这些都是成年人，正在从事，公司的，事情。”你是什么人？想想如果，你当时正要走，精心制作的，架构如何，人本主义在世界上行得通吗？然后也许你，我们可以谈谈吗？关于科学家们对……的看法，世界对抗，资本家和，管理团队。在，我不是来为你而来的。很好的回答，这是一个问题。好的。没错，你就是。做。你是，回答。看看奥哈洛。
Original English
**Chamath**: Yes, we've been here for a year. Let's discuss this. For example, if this, this substance is such, dangerous, you don't, need it, disclosure, or, you know, my, offering, for example, "know", your customer, "make laws." Don't compare with others, Wuhan's lab. This, um, Chamath, this, PR is not good, but, I also think this point, part of them, program, moral blackmail, when they just don't, want to say, we are companies, we have, commercial incentives. They always want, that is to say we or, research, institutions, or they say, something like this, we are a company, public interest, although no one really, completely knows what this is, take away, meaning is. So, this patina, moral signaling, hides what, actually, is happening. She tried to, hide the word, "company", but, legally, this is not. Can be hidden. I think they have, would do this if... We all got through it. Company. Everything, everyone, please keep very... Everything has to be very, small and strong, is a company. I think we should, call it, "Everything is a company." This is "Everything Lab." This is a company, we, discuss various, formulas. No, we, try, ideas. This is a master. Meme class. But you know what? These ideas, become so, dangerous that we, we don't want to carry, responsibility, your ideas. So, welcome to, the podcast "Everything is, a Lab." Here, when we, discuss this topic, zero, Freiberg, is there? In the eyes of scientists, this is a way, of thinking, this, you know, needs to, suddenly stop, and, then companies, as if taking power, from your hands? Maybe, they can, postpone, there are labs here, and, then say, "Hey, these are all adults, engaged in, corporate, things." What kind of person are you? Think if, you were going to go, elaborate, architecture how, humanism works in the world? And then maybe you, can we talk about? About what scientists think about, the world against, capitalists and, management teams. In, I'm not here for you. Very good answer, this is a question. Okay. Right, you are. Do. You are, answer. Look at O'Hallow.
### 模型发布狂潮与效率浪潮
**Chamath**: 让我，开始第一个话题。我只是想，退后一步，看看，狂风暴雨，模型肖像权使用许可。是的。我认为节奏，模型发布，发生的情况与，开源，是的。以及高级版，页，Anthropic，Grok，OpenAI等全球性组织，现在，难以理解。列表，疯狂的，而且，就像，过去10天，9，九月将是，DeepSeek 4.1 Flash发布。每百万代币1.20美元，如果他们，放置钥匙缓存，来自V1 DeepSeek的值。你还记得吗？这可是件大事。DeepSeek是什么时候发布的？缓存值当时为，尺寸约为390,000，字节，现在——890。令人难以置信的坠落，极好的，改进，效率。20，Qwen将于9月上映，阿里巴巴模型，版本，你知道，2.1。这是一个模型，开放鳞片，超越纳米香蕉2，这很大。谷歌在该领域的突破，极好的，图像生成。这是免费的。周末开放，代码，你可以，把它安装到你的，台式电脑，在家运行它，免费，产生，图像。她，由阿里巴巴开发。小米，由Mimo 22发布，9月，两天前。Mimo Pro的性能与Opus不相上下，来自克劳德的5，就是那一个，绝佳的模特，正如大家所料，这将改变世界，将会摧毁互联网，以及GPT-5-6大多数灵魂，基准测试。这是一个模型，3090亿，参数。她，完全拥有，公开级。你，您可以下载，这个模型，放置，它在你的服务器上，然后开始你自己的，无需付费即可复制，除了……之外，对任何人都没有其他要求。启动成本，服务器。这款模型，名为Bonzai 2，发布日期：9月17日，作者：Presum ML。它，实际上是Qwen的一个分支，其中一款模型，阿里巴巴。这是一个拥有27个部件的模型，数十亿，参数，占98%，生产率，大型模型Qwen。她 的尺码是5.9。千兆字节，而且你可以，启动它，当地的，电脑。你可以，配备英伟达芯片或，好的Mac电脑，工作室，你可以，自由的，运行此程序，疯狂模特，发送到您的桌面，电脑。进而，疯狂的大人，模型，大型模型，使用闭源代码，Anthropic发行Opus 5.5 22，九月，OpenAI Astro，出自Soul和Luna 23，九月，Grok 4.7，将于9月21日上映。Meta Muse将于22日，九月。任何，这些故事，会摧毁，一年前，互联网……它们都是为了……而发生的。过去10天。现在，您拥有完整的产品目录。具有开放的模型，用于生成代码，视觉的，为VLA提供的信息，你其实可以，为了管理它，他，可以管理，机器人。你有，显著的模型，超过，尖端LLM，一年前，开放，你们是什么？您可以安装和，自行运行，桌面，电脑。我们，过了那个点，伯尼·桑德斯，摆脱了，超级智能。它是，绝不可能，允许进入，现实世界。这些，事情已经真相大白了。具有开放的模型，开源代码，使用优惠码即可免费获得。它们在互联网上。你，您可以下载，它们并运行它们，到你的电脑，家。我不知道，在，我们身处一个怎样的世界？我们认为我们会 有，警察国家，谁会敲开你的门？会夺走你的东西，台式电脑，检查一下。看看是否，您正在攻读LLM学位吗？到您当地，电脑。海浪，效率已经到来。以及这种速度，这些模型出现，这一切的发生都是为了，过去10天，简直太棒了。我们现在，在这种奇特的，边界，宽敞，应有尽有，正在加速。改进，生产率，降低成本，人工智能，变得如此，高效且，可用的。他将，无处不在，我们所有人，我们将从中得到什么，益处。这不是关于，仅与人本主义和，开放人工智能，封闭世界，赚取数十亿，美元，出售，模型放置在，企业。现在，这件事正在发生，广泛分布。所以，我当时就想……值得迈出这一步。
Original English
**Chamath**: Let me, start with the first topic. I just want to, take a step back, and look at the, whirlwind, of model releases. Yes. I think the pace, of model releases, what's happening with, open source, yes. And premium, pages, Anthropic, Grok, OpenAI and other global organizations, now, is incomprehensible. The list, is insane, and, like, the last 10 days, 9, September will be, DeepSeek 4.1 Flash released. $1.20 per million tokens, if they, put key cache, values from V1 DeepSeek. Do you remember? This was a big deal. When was DeepSeek released? The cache value was then, about 390,000, bytes, and now -- 890. An incredible fall, excellent, improvement, in efficiency. 20, Qwen will be out in September, Alibaba model, version, you know, 2.1. This is a model, open scales, beyond Nano Banana 2, which is huge. Google's breakthrough in the field, excellent, image generation. This is free. Open over the weekend, code, you can, install it on your, desktop computer, run it at home, free, to generate, images. She, developed by Alibaba. Xiaomi, released by Mimo 22, September, two days ago. Mimo Pro performs on par with Opus, from Claude's 5, which is that one, excellent model, as expected, that will change the world, will destroy the internet, and GPT-5-6 most souls, benchmarks. This is a model, 309 billion, parameters. She, fully owned, public grade. You, you can download, this model, place, it on your server, and then start your own, replication without paying, any other requirements except for... startup costs, server. This model, named Bonzai 2, released date: September 17th, by Presum ML. It, is actually a fork of Qwen, one of the models, Alibaba. This is a model with 27, billion, parameters, that is 98%, productive, large model Qwen. Her size is 5.9. Gigabytes, and you can, launch it, locally, computer. You can, with an Nvidia chip or, a good Mac computer, studio, you can, freely, run this program, crazy model, sent to your desktop, computer. And then, crazy adults, models, large models, using closed source code, Anthropic released Opus 5.5 22, September, OpenAI Astro, from Soul and Luna 23, September, Grok 4.7, will be out on September 21st. Meta Muse will be released on 22, September. Any, of these stories, would have destroyed, the internet a year ago... They're all happening for... the last 10 days. Now, you have a full catalog. With open models, for generating code, visual, information provided for VLA, you can actually, to manage it, he, can manage, robots. You have, significant models, exceeding, cutting-edge LLM, a year ago, open, what are you? You can install and, run yourself, desktop, computer. We, are past the point, Bernie Sanders, got rid of, super intelligence. It is, absolutely impossible, to allow into, the real world. These, things have come to light. With open models, open source code, free with a coupon code. They're on the internet. You, you can download, them and run them, to your computer, home. I don't know, in, what kind of world are we? We think we're going to have, a police state, who's going to knock on your door? Will take your stuff, desktop computer, check it out. See if, you're pursuing an LLM degree? To your local, computer. The wave, of efficiency has arrived. And this speed, these models appear, all of this is happening for, the last 10 days, it's just amazing. We are now, in this peculiar, boundary, spacious, everything, is accelerating. Improvement, productivity, cost reduction, AI, becoming so, efficient and, available. He will, be everywhere, all of us, what will we get from it, benefits. This is not about, only with humanism and, open AI, closed world, making billions of dollars, dollars, selling, models placed in, enterprises. Now, this thing is happening, widely distributed. So, I was just thinking... it's worth taking this step.
### 人工智能的未来与IPO挑战
**Chamath**: 回到我们所处的现实，因为在这种情况下，我们应该怎么做？停止超级智能，我们需要关闭数据中心。你甚至不需要中心化的数据处理就能完成你90%的工作。你可以借助人工智能来完成。你今天就可以免费在电脑上下载它。不到一年前，这还是人类历史上最先进的技术。今天，这一切都已触手可及。我认为，我们值得退后一步，确保我们都清楚地意识到，当我们进入关于这些实验室或公司的讨论时，我们身处何处。我不知道它现在是否还处于静止状态。重要的是，我认为开源代码就在这里，我认为它正在提高速度，所以它只会增加生产力。
Original English
**Chamath**: Back to reality, where we are, because in that scenario, what are we supposed to do? Stop super intelligence, we need to shut down data centers. You don't even need central data processing to do 90% of what you do. You can do with AI. You can download him alone on a computer today for free. What is the most advanced technology in human history less than a year ago? All of it is here today. I think it's worth just taking a step back and making sure that we are all so aware of when we enter this discussion about these lab quotes or companies. I don't know if it's still static right now. It's important. I think open weekend code is here, I think improving speed, so it's just going to increase productivity.
**Jason**: 是的。
Original English
**Jason**: Yes.
**Friedberg**: 或许我们现在正看着人工智能的浪潮。生产力、繁荣、生产力的浪潮。
Original English
**Friedberg**: Perhaps we are now looking at the waves of AI. The waves of productivity, prosperity, productivity.
**Chamath**: 是的，对于消费者而言，他们不在乎使用了什么模型。在过去三年多的时间里，我们编造了许多关于……的故事，然后Grok出现了，接着是Meta的消息，我两者都用。这些产品，以及这是第一次，当一个“普通人”（这绝不是贬低任何人的意思）能够获得人工智能的价值。你会看到，当我们都关注管理公司，通过投资它们时，我们采用了另一种方法，比如，好吧，也许这会淘汰一些职位，也许这会增加一些职位。但现在你可以像Muse一样，立即连接到一切。并且真正地在世界上执行实际工作，如果你是一个普通人，想要从人工智能中受益。我们，我们转过弯了。弗里德伯格，我认为我们已经完美地准备好了。我认为到今年年底，我们将进入下一个阶段，届时，我们的妈妈们、叔叔们，人们将开始使用Muse和那个Grok机器人，它正在迅速发展以解决实际问题。他们会说：“等等。”其次，我个人绝不会允许、需要一个领导设备，我绝不会允许一个助理、帮手、经理。现在我得到了一个“免费助理”。这将令人印象深刻。
Original English
**Chamath**: Yes, for consumers, they don't care what models are being used. We, for the last three years, more than that, have made up a lot of stories about... and then Grok came out, and then the news from Meta, I use both. These products, and this first time, when a "normal" person in the world (which doesn't mean to be demeaning to anyone by the way) can get the value of AI. If you'll see, when we all look at managing companies, by investing in them, we used another, for example, approach, okay, maybe this is going to retire, you know, some positions, maybe this is, um, going to add some positions here or there. But now you can, like Muse, connect immediately to everything. And actually execute actual work in the world, if you're a normal person, who wants to get the benefit of AI. We, we turn the corner. Friedberg, I think we are perfectly set up. I think by the end of this year we will enter the next phase, where at some point, mothers, our uncles, people will start using Muse and that bot, the Grok bot, which is quickly developing to solve actual problems. They, they're going to say, "Wait a second." Second, I never would have allowed, needed a leader device, I never would have allowed, for me personally, an assistant, a helper, a manager. Now I get a "free assistant". This is going to be impressive.
**Sacks**: Chamath，在接下来的几个月里，他们将开始使用这些产品，而美国底层一半的K型经济体将感受到某种优势的恢复。关于这里这些模型正在发生什么，你有什么想法？你一直在玩这些模型，因为最初有技术专家？你下载了这些模型自己玩，因为我用电脑玩游戏。我的MacBook Pro就是这样设计的，这好吗？请允许我，或许可以给你一个答案并解释一些事情。所以，我认为我们看到这些模型越来越接近并聚集在一起，也就是说，它们现在都在误差范围内。所以，问题是……因为确实存在这种情况。许多模型大致相同，在不同类别中价格各不相同。那么，优势在哪里？优势仍然存在，而且优势依然存在。它由“封装”构成，你是哪一个？用于模型封装。所以，如果你仔细想想……就好像你已经……有一个类似大脑的模型，但是你需要实施这项措施，一个长着手的大脑，腿、手和眼睛，所以它完成了工作。这里，这是你验证模型的方式，好吗？你呢？让她成为一个代理。这就是这个短语的含义。而且还有各种各样的封装，这个行业发明它们是为了让这些模型成为更多的代理。就是这样。仍然有很多优点。所以，在80/90的比例下，我们有一个实验室，我们拆解并分析了所有不同的模型，不同的封装方式，只是为了看看它们是如何运作的。而且它们在价值和质量方面都非常不同。所以，只是为了发展，根据弗里德伯格的说法，你有一些模型正在趋同。封装则截然不同。所以，它们的生产力情况显著不同。现在公司和人们正在调整方向，以决定他们愿意为什么付费。
Original English
**Sacks**: Chamat, in the next few months they will start using these products and the bottom half of America, this K-shaped, will feel a restoration of some advantage. Regards, what about, thoughts on what's happening here with these models? You've been playing around, because at first there were tech experts? You downloaded these models yourself because I'm playing games, so with a computer. That's how it's designed on my MacBook Pro, is that good? Allow me, perhaps to give you an answer and explain something. So, I think we're seeing these models get closer and closer and cluster together, that is to say, they are now all within the range of error. So, the question is... because there is such a thing. Many models are roughly the same, with varying prices in different categories. So, where is the advantage? The advantage still exists, and the advantage still exists. It's constituted by the winding, which one are you? Used for model packaging. So, if you think about it... it's like you already have... a similar model brain, but you need to implement this measure, a long-handed brain, legs, hands, and eyes, so he gets the job done. Here, this is how you authenticate the model, okay? What about you? Let her be an agent. That's what it means, the phrase. And there are all sorts of windings, which the industry invented to make these models more agents. That's it. There are still many advantages. So, at 80/90, we have a lab, and we disassemble and we analyze all the different, all the models, different winding ways, just to see, how they work. And they are all very different in terms of value and quality. So, just to develop, according to Friedberg, you have some models converging. The winding is completely different. So, their productivity significantly differs. And now companies and people have reoriented why they are willing to pay.
**Chamath**: 我发了这条推文，尼克，如果你能把它放上来。我不知道，这有教育意义吗？确切地说。但它是什么呢？这个节目，这适用于“蚂蚁”，但它可能指的是许多封闭模型，那就是你开始看到这种巨大的收入集中度，也就是说，少数几个人吸收了所有最高和最昂贵的代币。我认为，如果为了证明弗里德伯格所展示的，这是合乎逻辑的，那么这些人，除此之外，将面临巨大的内部压力，无论是来自财务团队，还是来自财务总监，还是来自股东，所以请解释原因。因为这种集群正在发生，他们不会转向越来越便宜、价格更低的模型。现在你可以获得所有越来越便宜的“蚂蚁”模型，而“蚂蚁”会保留收入，但也有风险。你能做的就是弗里德伯格刚才说的，直接去开放模型代码，让她独立自主。所以，我们正处于一个非常困难的周期点。第一版人工智能交易相对简单：你只出售代币，你打包，你转移代币，这就足够了。价值不平等。我认为它会消失。这将迫使OpenAI和Anthropic向上爬。别无选择。你不能坐着服务代币，因为成本正在降低。因为所有价值都被围绕着你的智能代币的人吸收和创造。你必须完成这项工作。
Original English
**Chamath**: I tweeted this, Nick, if you can just put it up here. I don't know, is this educational? Exactly. But what it is, is this show, and this applies to ants, but it could refer to many of the closed models, which is you're starting to see this massive revenue concentration, that is to say, a few people are absorbed by all the highest and most expensive tokens. And I think if to prove that Friedberg showed, which is logical to do, these guys, outside of that, the table is going to be under immense internal pressure, whether from the side or elsewhere. Their financial team, whether from the side of their financial director, whether from the shareholders, so please explain why. Because this clustering is happening, they're not going to move to cheaper and cheaper, lower-priced models. Now you can get all the cheaper and cheaper ant models, and the ants will keep the revenue, but there's also a risk. What you can do is what Friedberg just said, which is go directly to the open model code and put her on her own. So, we're in a very difficult point in the cycle. The first version of the AI trade was relatively simple: you just sold tokens, you packaged, you transferred tokens, and that was enough. Inequality of value. I think it's going to go away. This is going to force OpenAI and Anthropic to climb up. No choice. You're not, you can't sit around servicing tokens, which the cost is decreasing. Because then all the value is absorbed and created by the human that wraps around your token of intelligence. You have to do that work.
**Jason**: Jason，就是这样。他们为什么要这样做？参与网络安全，他们将不得不处理法理学，他们将不得不处理客户支持，所有这些我们不了解的事情。你确定他们会相互竞争吗？这些趋势可以让他们这样做。
Original English
**Jason**: Jason, that's it. Why would they engage in cybersecurity, they're going to have to deal with jurisprudence, they're going to have to deal with customer support, all these things that we don't know about. Are you sure they're going to compete with each other? These trends can allow them to do that.
**Chamath**: 是的。而且我认为这为他们的IPO条件创造了一个非常有趣的情况，在我看来。萨克斯，想想看，所有这些先进的企业，Anthropic Labs和OpenAI，他们都在本周发布模型，代币价格下降了50%。他们的IPO似乎……将会推迟。Anthropic刚刚瞄准了2万亿美元的估值，OpenAI是1.2万亿美元。OpenAI已经七次表示，他们计划在2027年再次进行IPO，因为安全问题。而Anthropic，根据《华尔街日报》的报道，本应在10月提交申请并上市，IPO原定在11月，或者，也许会有更多项目被推迟。这是你的Polymarket。或者Anthropic会在2026年上市吗？我们优秀的合作伙伴，Polymarket。我把它交给你。
Original English
**Chamath**: Yes. And I think this creates a very interesting thing for their IPO conditions, in my opinion. Just consider, Sacks, all these advanced enterprises, Anthropic Labs and OpenAI, they're all releasing models this week where the price of the tokens is down 50%. Their IPOs seem to be... going to be delayed. Anthropic just targeted a score of $2 trillion, OpenAI $1.2 trillion. It's been seven, OpenAI said they plan to IPO again in 2027 because of issues of safety. And Anthropic, which had to file and was going to go public in October, according to the Wall Street Journal, the IPO was going to be in November, or, perhaps it's going to be more projects that have been delayed. This is your Polymarket. Or is Anthropic going to IPO in 2026? Our good partners, Polymarket. I'll turn it over to you.
**Sacks**: 恭喜我的感谢，Jen。本月初，Anthropic今年上市的可能性曾高达96%，但现在已降至76%。萨克斯，这就像手机一样吗？制造模型的领先公司需要上市融资以进行开发，而现在他们却推迟了自己的IPO。请解释一下。这怎么可能呢？如果你是这些公司的投资者之一，或者是他们的董事会成员，你会对这些创始人，对这些已经上市并获得人工智能持续发展所需资金的公司，做出什么指示或委托？嗯，我认为我们看到我们的朋友，那些Anthropic的投资者，正为此事而焦头烂额。更不用说那些几周后就离职的“假线人”了。你现在拥有的是这样一家公司管理层，它表明人类灭绝的可能性超过10%，而他们自己的产品已经解决了这个行业问题。所以好好想想。关于它所创造的风险因素。如果你正在进行IPO，你自己的公司管理层对此有何看法？然后我认为你们当中还有其他人。除了公司的精神分裂症，我不知道还有什么证据。达里奥发表了一篇文章：“我们需要扩展边界。”几天后，在Claude 5.5发布之前，新的限制就被安装了。我的意思是……他肯定知道。在撰写这篇文章时，他们即将发布的产品，将扩展边界。所以，至少这是虚伪的，但或许更糟。然后他们发表了关于与人工智能相关的生物风险的文章，并说需要停止。同时，他们宣布在旧金山建立一个新的生物实验室，一个新的“湿实验室”。
Original English
**Sacks**: Congratulations to my thanks, Jen. At the beginning of this month, Anthropic peaked at 96% that it would go public this year, but now it's down to 76%. It's like a cell phone, Sacks? The leading companies making models need to go out to the stock exchange to fund development, and now they're delaying their own IPOs. Please explain. How is this possible? And if you're one of the investors in these companies or a member of their board of directors, what would you do or commission to these founders, to these companies that have become public and received the funding necessary for the continued development of AI? Well, I think we're seeing from our friends who are investors in Anthropic tearing their hair out for this reason that's happening. Not to mention the fake informants who left a few jobs a few weeks later. What you have now is this company management that indicates there's a greater than 10% probability of human extinction, and their own product has solved the industry. So think about that. About the risk factors it creates. Are you, if you're doing an IPO, what does your own company management think about this? And then I think there are others among you. Evidence, I don't know what's going on, other than the company's schizophrenia. Dario published an article: 'We need to expand the boundaries.' A few days later, before Claude 5.5 was released, new restrictions were installed. I mean... he must have known. When writing this article, what they were about to release, this product, which would expand the boundaries. So, at least that's hypocritical, but, perhaps even worse. Then they published an article about bio-risk related to artificial intelligence, and that it needs to stop. At the same time, they announced a new bio-lab in San Francisco, a new 'wet lab'.
**Chamath**: 什么？他们开了一个实验室，然后，正如我们被告知的，我们都有10%的几率，萨克斯，我们会死。我们等一下，我们停下来。这是什么？Anthropic的首席执行官，一位主要高管，不履行职责，破坏了你真正的IPO？你必须代表普通董事和他们的股东行事。我认为如果我们是一个董事会，如果我们四个人以某种方式影响了Anthropic的董事会，我们有魔力。我们将取代普通董事。我们会把普通董事送到实验室，然后找一个知道如何管理公司而不是破坏它的人。他简直是亲手毁掉了自己的公司。
Original English
**Chamath**: What? They opened the lab after, as we're told, we all have a 10% chance, Sacks, we're going to die. Let's wait a second, let's stop. What is this? Non-performance of duties by the principal executives, the director of Anthropic, destroying your real IPO? You have to act on behalf of the general directors, their shareholders. I think if we were a board of directors, if four of us somehow influenced the Anthropic board, we have magic. We would replace the general directors. We would send the general directors to the lab and find someone who knows how to manage a company and not destroy it. He literally destroyed his own company.
**Chamath**: 我认为这有一些因素，而且我认为投资者对于授予创始人，或者说Dario，拥有超级投票权的股份感到相当紧张。关于是否授予他拥有超级投票权的股份，目前正在就这个故事进行积极的辩论。请解释一下。它是什么？为什么在这种情况下如此关键？我们的观众可能知道，也可能不知道，想知道它是什么。
Original English
**Chamath**: I think this has some element, and I think investors are quite nervous about providing founders, or Dario, with shares that have super voting rights. Information contained in that story is currently holding an active debate on this topic of whether to give him shares with super voting rights. Please explain. What is it? Why is it so crucial in this situation? Our audience may or may not know and want to know what it is.
**Sacks**: 据称Anthropic的创始人每人只拥有约2%的股权，总的来说，与通常IPO时其他创始人相比，这是一个相对较小的比例。当然，这仍然是一大笔钱，你知道，就是钱。数百亿，就这些，就美元。超级投票权股份，实际上，将公司的投票权与经济股权分开，并赋予某些人控制公司的权利，即使他们没有经济所有权，这通常是必要的。这种情况很常见。我的意思是……拉里和谢尔盖，我认为我们是第一批已知的在谷歌和扎克伯格获得超级投票权的创始人。这可以成为公司的一个稳定因素，因为它意味着只要创始人在监督工作，他们就不会反对，你知道，收购尝试等等。但另一方面，它需要对……的巨大信任，所以这些创始人，实际上，他们无论如何都无法被清除。他们没有。你只需要思考在这种情况下这是否合理和正当。我必须说，即便如此，对于像我这样不一定需要执行更多严格内容规定的人来说，我有点不安。
Original English
**Sacks**: The alleged founders of Anthropic only have about 2% of the property each, and in total, have a relatively small amount compared to... because usually there are others who are currently founders of the IPO. It would still be a lot of, you know, just money. Hundreds of billions, just that, just dollars. Super voting shares, actually, separate voting rights from economic shares in a company and grant certain people the right to control the company, even if they don't have economic property, which is often necessary. This situation is common. Things. I mean... Larry and Sergey, and I think we were the first known founders to receive super voting shares at Google and Zuckerberg. This can be a stabilizing factor in a company because it means that as long as the founders oversee the work, they won't object, you know, acquisition attempts, and so on. But on the other hand, on the other hand, it requires immense trust in... so these founders, actually, they're not, anyway, they can be eliminated. They don't have. You just need to think about whether in this situation it's justified and reasonable. I must say, even so, for someone like him, I, who doesn't necessarily have to execute more great content regulations, I'm a little, I'm uneasy with the idea.
**Sacks**: Anthropic在旧金山发现了一个湿实验室。我的意思是，我当时玩得很开心。我一直在想……回来，我不知道，我想要这个吗？也就是说，是否要做这件事。你能把它再挪远一点吗？我想我们都，我已经看过这部电影了。
Original English
**Sacks**: Anthropic found a wet lab in San Francisco. I have, I mean, I was having fun. I kept thinking... back, I don't know, do I want this? That is, whether to do this thing. Can you just move it a little further? I think we've all, I've seen this movie before.
**Chamath**: 是的，或许是在海上。我不知道。把它搬到石油钻井平台上之类的。
Original English
**Chamath**: Yes, perhaps at sea. I don't know. Move this to an oil rig or something like that.
**Friedberg**: 我会说，如果我是董事会成员，我会让Dario继续担任首席执行官。原因是他显然已经证明他可以建立一种独特的文化，并与最强大的竞争对手公司，即OpenAI，抗衡。他做到了，而且他从后面追了上来，然后我，我认为这不值得。低估，这有多难？这需要什么条件？所以，他做到了，而且他，他实际上已经建立了最好的业务，有史以来最好的发展坡道。我认为如果我在董事会，我可能已经解决了这个问题。
Original English
**Friedberg**: I would say, if I were a board member, I would leave Dario as the general director. The reason is because he has clearly proven that he can build a unique culture and take a stand against the most formidable competitor company, namely OpenAI. He did it, and he came from behind, and then I, I think it's not worth it. Underestimate, how hard is this? What does it take? So, he did it, and he, he built, in fact, has built the best business, the best ramp of all time. I, I think if I were on the board, I would have. Probably already solved this problem.
**Chamath**: 萨克斯刚才说的，现在是什么情况？该公司在流动性方面有许多额外的风险。我认为，如果你从一个20岁年轻人的角度来看Anthropic的前景，那么这家企业需要数千亿美元才能实现其抱负。我将决定尽可能多地做到无风险。我认为只有两件事你可以做。首先，你绝对需要投资于信息披露。问题是……它会像它说的那样运作。否则，就会有一个密集而坚不可摧的S-1文件，而且密度更高，更坚不可摧。好吗？所以你可以表面上覆盖你的监管风险。但事实是，它实际上在市场上显示出来，就是这样。你刚才在做什么？你必须降低卖方对IPO的预期，并为IPO买家创造一个更大的安全边际。就是这样。你找到一个市场，这个价格是Jason。也就是说，从本质上讲，对我而言。
Original English
**Chamath**: What Sacks just said, and what is it now? The company has many additional risks to liquidity. I, I think if you consider Anthropic from the perspective of a 20-year-old prospect, then this enterprise needs hundreds of billions of dollars to achieve their ambitions. I would decide to do as much as possible. Risk-free. I, I think there are only two things you can do to make. First, you absolutely need to invest in information disclosure. The problem is... it will work as it says. Otherwise, there's a dense and impenetrable S-1, yet denser and impenetrable. Okay? So you can superficially cover your regulatory risks. But the fact is, it actually shows in the market, that's it. What were you just doing? You must loosen the seller's expectations for the IPO and create a much larger safety margin for IPO buyers. That's it. You find a market, this price is Jason. That is, by nature, to me.
**Jason**: 请稍等，我只是，我给出的数字大概是这个值，对吧？
Original English
**Jason**: Please wait, I just, I gave the approximate value of the numbers, right?
### IPO、市场动态与人工智能的未来
**Speaker A**: 你认为他们会吗？有一个指标，增长一百，然后是数十亿美元。否则，IPO过程非常干净利落，或许会是这样。市场二元资本化数万亿美元，现在，你知道，你有一万亿美元或更少。原因在于：对冲基金、养老基金、系统和复合体、共同基金，长期以来，他们所处的职位——受托人责任——在他们面前，资金来源将证明其合理性。所有风险，基本上都要求越来越低，价格更低。我想那就是你会看到的。我认为你会看到很多更温和、更明智的首次公开募股（IPO）做法。很可能就是这样。稍后才会经过。价格低于有人在思考。我觉得萨克斯所说的是原因是什么，但我认为这对于蚂蚁来说是有好处的，因为它会驱散到处都是的噪音。它使他们仅仅是一家公司，然后经营企业。而且我认为他们会做得很好。
Original English
**Speaker A**: Do you think they would? There's a metric, a hundred X growth, and then billions of dollars. Otherwise, the IPO process is very clean and crisp, and it's probably going to be that way. The market's binary capitalization, trillions of dollars, and now, you know, you have a trillion dollars or less. And the reason is: hedge funds, pension funds, the system and the complex, mutual funds, long-term, the positions that they're in, fiduciary responsibility, in front of them, the sources of capital, will justify all the risk, basically, by demanding lower and lower prices. And I think that's what you're going to see. I think you're going to see a lot more gentle, a lot more intelligent approaches to IPOs. It's probably going to be that way. It's going to come later. It's going to come at a price that's lower than what people are thinking. I think what Sacks is saying is what the reason is, but I think it's good for Ant, because it disperses all the noise everywhere. It makes them just a company, and then running a business. And I think they're going to do very well.
**Speaker A**: 听着，我会陪着你。我对你很坦诚。尝试招募人员对抗蚂蚁，我不能。那是不可能的。他们获得的补偿，哇，太棒了。业务增长如此之快，如此吸引人。就像当你面对面比赛时——我和他们在一起。一场都没赢。烘焙比赛为了人们。我想说他赢了，但是，并非如此。而且这很难创业。所以，他们是人才的磁石。睁开眼睛。磁铁天赋，但其他方面，风险将是推动IPO。而唯一摆脱它们的方法——简单来说，就是大幅降低价格。
Original English
**Speaker A**: Listen, I'll go with you. I'm being very candid with you. Try to recruit people against Ant. I can't. It's impossible. The compensation they get, wow, it's great. The business is fast. The growth is so attractive. It's like when you play head-to-head, I'm with them. I haven't won a single baking competition for people. I want to say he won, but, it's not. And it's hard to start a business. So, they're a magnet for talent. Open your eyes. A magnet for talent, but on the other side, the risk is going to be driving the IPO. And the only way to get rid of them is to, simply put, significantly lower the price.
**Friedberg**: 是的，还有弗里德伯格，当我们正在考虑S-1中的风险因素时，它们很肤浅。通常情况下是这样的预订。你获取示例，SpaceX。哦，如果星舰没用？如果有什么竞争吗？它类似旗帜。我认为，就Anthropic而言，我们来考虑一下这个问题。完全不同，即：“嘿，我们和我们的雇员，我们别再说了。它会毁灭人类。我们人很多。规定。我们要求规定。而且，我们做的事情，生物研究和我们启动实验室。”所以，是吗？你同意吗？我想知道S1会是什么样子。实际因素、风险和警告，显著反对免费提款参与IPO的股份？
Original English
**Friedberg**: Yeah, and Friedberg, when we're thinking about the risk factors in an S-1, they're very superficial. It's usually like this boilerplate. You get the example, SpaceX. Oh, what if Starship doesn't work? What if there's competition? It's like a flag. If, I think, in the case of Anthropic, let's think about this. It's completely different, which is, "Hey, we and our employees, let's not talk about it anymore. It's going to destroy humanity. We have a lot of people. Regulations. We demand regulations. And, what we do, biological research and we launch labs." So, is that? Do you agree with that? I wonder what the S1 would look like. Actual factors, risks, and warnings, significantly against free withdrawal of shares participating in the IPO?
**Speaker A**: 我认为这一点，客户集中度，以及我之前的那些评论，传播速度惊人的生产模式，公开赛级别，开源。原因是，因为Anthropic提供了当今生活中最佳的科学模型。我负责管理生命科学组织的科学实验。我们使用他们的模型。它们极好的。它们比其他型号更好，这就是我们想要使用它们的原因。归根结底，在编写代码之前，编写软件，内部执行工作流程和操作方面，我们可以使用其他产品编写这段代码。对我们来说，我们这样做。是的。而且我认为大概就是这样。市场将发展，我认为我们开始看到市场中的证据就证明了这一点。你在这里就能看到这一点。客户集中点。极其有价值的技术、工程、数学、生物科学，这些真的很难的技术问题，你想要这些一流、高品质的模型，你将为此付出代价。巨额奖金。而且你肯定能从中赚到这笔钱，因为价值非凡。一直都是如此。我对人工智能的立场。人工智能倒还好。与价值有关，替换旧东西。人工智能99%的价值在于为了允许创造新事物，从未存在过历史上可能的人类。而且我认为正是基于这一点，Anthropic和OpenAI将赢得其钱。单独来看，我认为OpenAI将从消费者和人类那里获得很多钱，或许在某种程度上。但在公司的环境方面，我认为资金来源创造了新的前沿：发现新酶，关于人形动物公司本周宣布的可以治疗人类疾病，以及新开发的新技术，新工程系统，创造前所未有的事物，这是可以做到的。我想就是这样了。溢价，这是高价，你需要为此付费，这些封闭模型和这款高端机型。但这又是另一回事了。比采取整个企业。另外，我认为大多数公司的申请将从使用这些开放模型开始平衡。代码将是作者：使用更便宜的模特等等。我不是。我知道你为什么前往顶级模型编写代码时，你可以这样做免费，而且组织中人人编写代码是实用的。自由的。因此我觉得它更大。人择问题包括这一点，问题关于客户，以及是否真的吗？高端产品他们以这样的价格出售市场，以及是否就是那里他们指导市场。在我看来，这是……观点，那个方向，它现在要去哪里？市场，而且将会是最大的风险因素。这一切一个关于什么的故事？他们说它终会到来，世界末日等等。类型：老兄，简单确保你的该产品不会导致直到世界末日，以及一切没事了。这将是良好的第一步。Mark在Meta是如何做到的。我的意思是，这非常棒的公司。处决是非凡的。营收增长非常出色。但当你开始看看开源的影响，在我们当中有几个人到场了。病毒式传播排行榜。70-80%代币的使用——这些就是那些黑暗的象征，我提到过第6点。几个月前在这里这个模块，并非会被追踪，但是有些人可以追踪他们路由器。这些带有暗色标记的开源现在构成大多数代币，以及或者我可以提供吗？关于你的统计数据，这是JK吗？为了当然，你只是他说是真的，很重要，但也有统计数据，我觉得只有这样才行。过去12周，代币的使用从80/20改为关闭打开。是的。最高可达80/20开对付亲近的人。是的。我认为我们永远都会已经见过12次了，几周。我不这么认为。我们曾经见过吗？这样的波浪技术成就。又来了维塞尔图。在每个人的故事，技术市场，我们从未见过。绝非如此。是的，我们知道。这是最大的人类风险因素，因为他们是第一批开始使用人工智能创建工具等。但自从这市场正在向开源和公开重量级，轰鸣，他来了。看看这个。这是一个Versel图，这件事迅速走红星期。是12周吗？我的天啊。疯狂的。如果你看看这个图表，我认为是开启那扇门需要包含在内这些S1。这就是我的原因，考虑到所有这些因素，开源版本，节目开始时的体重，或者这一段，因为这不仅仅是关于因为它们是，而且也是因为他们现在是正在蔓延。他们的事实证明，很多以及速度提升非凡的，现在他们用于甚大阵列，对于机器人技术而言，他们用于图像生成，图像和视频。因此，模型有开放周末，开源代码，立即编写代码，用于所有更宽的人工智能的应用，智力。优质的——材料方面，我得说，依然是最好的。针对特定情况使用。是的。对于工程领域而言，对于高科技行业而言，非常复杂任务和工作流程，以及用于问题解决，纳维-斯托克斯方程解决方案，数学问题解决方案，生物学问题。这就是这些模型所在的地方，绝对有效，人们会为此付费。我会说，这是个奖项。几乎无穷大。这套产品售价为美元付费应用。所以但这意味着他们的市场被摧毁了。这意味着我们观察到市场分散化，大部分案例，人工智能的应用，针对这些开源代币，源代码，开放周末代码。那么问题就变成了：什么高端车型依然将会蔓延。是的。那么问题来了，由分布构成这些代币，先进的公司，它们百分比是这些任务，与哪个相比，百分比是可互换任务，开放周末代码？如果答案超过60%或70%，你不必请将此考虑在内。收入。如果答案是只有10%，以及全部的90%。产生的收入绝对地高级场景使用它们，一切都会好起来的，他们会去看赛马。是的，而且在否则就是如此。你战斗的那种东西仅仅是和人类疏忽或缺乏教育，对吧？是的？因为当他们看到了他们的账单……让我致敬，萨克斯，没错。结束。拉姆，你的想法是什么？然后我给到期的...
Original English
**Speaker A**: I think this point, the concentration of customers, and my previous comments, the incredibly fast-spreading production models, open-level, open source. The reason is, because Anthropic provides the best scientific models for life today. I manage scientific experiments for organizations in life sciences. We use their models. They're excellent. They're better than other models, and that's why we want to use them. At the end of the day, before writing code, writing software, software, internally executing workflows and operations, we can use other products to write this code. For us, we do that. Yes. And I think that's probably it. The market will develop, and I think we're starting to see evidence of that in the market. You can see that here. The concentration point of customers. Extremely valuable technology, engineering, mathematics, biological sciences, these are really hard technical problems, and you want these first-class, high-quality models, and you're going to pay for it. Huge premium. And you're definitely going to make money from this money, because of the extraordinary value. It's always been that way. My position on AI. AI is fine. It's about value, replacing old things. 99% of the value of AI is about allowing the creation of new things, that have never existed in history, possibly for humans. And I think it's on that point that Anthropic and OpenAI will earn their money. Separately, I think OpenAI will earn a lot of money from consumers and humans, perhaps to some extent. But in the corporate, I think, environment, the sources of capital are creating new frontiers: discovering new enzymes, about the humanoid animal company announcing this week, that can treat human diseases, and new developments, new technologies, new engineering systems, creating things that have never existed before. This can be done. I think that's it. The premium, this is the high price, and you need to pay for these closed models and this high-end model. But that's a different thing. Than taking the entire enterprise. Also, I think the majority of corporate applications will start from using these open models to balance. The code will be written by using cheaper models and so on. I'm not. I know why you go to the top models when writing code, you can do it for free, and everyone in the organization can write code practically. It's free. So I think it's bigger. The Anthropic problem includes this point, the problem about customers, and whether it's really a high-end product that they're selling at that price point in the market, and whether that's where they're guiding the market. In my opinion, this is... the point of view, that direction, where is it going now? The market, and it's going to be the biggest factor of risk. This whole thing, a story about what? They say it's going to be the end of the world and all that. Type: dude, simply make sure your product doesn't lead to the end of the world, and everything's fine. This will be a good first step. How Mark at Meta did it. I mean, this is a fantastic company. The execution is phenomenal. Revenue growth is outstanding. But when you start to look at the impact of open source, a few of us were there. The viral charts. 70-80% of token usage—these are those dark tokens, I mentioned point number six. A few months ago here, this module, it's not going to be tracked, but some people can track them on their routers. These dark-tagged open source now constitute the majority of tokens, and or can I provide? About your statistics, is this JK? For, of course, you just, he said this is really important, but there are also statistics, I think that's the only way. The last 12 weeks, token usage went from 80/20 closed to open. Yes. Up to 80/20 open against close. Yes. I don't think we'll ever have seen 12 weeks. I don't think so. Have we ever seen? Such a wave of technological achievement. Here comes the Versel chart again. In everyone's story, the technology market, we've never seen it. Absolutely not. Yes, we know. This is the biggest risk factor for humans, because they were the first to start using AI to create tools and so on. But since this market is moving towards open source and open heavyweight, roaring, here he comes. Look at this. This is a Versel chart, and this thing went viral quickly this week. Is it 12 weeks? Oh my God. Crazy. If you look at this chart, I think it's opening that door that needs to be included in these S1s. This is my reason for considering all these factors, the open source version, the weight at the beginning of the show, or this segment, because it's not just about because they are, but also because they are now spreading. Their, it turns out, a lot, and the speed of the uplift is phenomenal, and now they are used for very large arrays, for robotics, they are used for image generation, images and video. So, models have open weekends, open source code, immediately writing code, for all the wider applications of artificial intelligence, intelligence. The quality—on the material side, I have to say, is still the best. For specific situations of use. Yes. For engineering, for high-tech industries, very complex tasks and workflows, and for problem solving, Navier-Stokes equations solutions, mathematical problems solutions, biological problems solutions. This is where these models are absolutely effective, and people will pay for it. I would say, it's a premium. Almost infinite. This product suite sells for dollars, paid applications. So, but that doesn't mean their market is destroyed. It means we observe market decentralization, and most cases, applications of artificial intelligence, are targeting these open source tokens, source code, open weekend code. So the question becomes: what high-end models will still spread? Yes. So the question comes down to the distribution of these tokens, advanced companies, what percentage of these tasks, compared to what percentage of interchangeable tasks, open weekend code? If the answer is more than 60% or 70%, you don't have to include this in the revenue. If the answer is only 10%, and all 90% of the revenue generated is absolutely in high-end scenarios using them, everything will be fine, and they'll go to the horse races. Yes, and otherwise it's like that. The kind of thing you're fighting is just human negligence or lack of education, right? Yes? Because when they see their bill... Let me give a shout-out, Sacks, that's right. Over. Ram, what are your thoughts? And then I'll give to the due...
**J Cal**: 不，只是转移工人，装载到哪里，找个地方吃饭，然后省钱。是的，所以，萨克斯，我这里画了个图。演出之前，但在这里你的，呃，完美！过去平均100天的模型展示，什么？弗里德伯格没事。准备好了。你呢你会发现，是的，四轮摩托车和阿斯特拉在右下角，价格昂贵，而且它们有很多提供。但是然后，向下向下，你开始参见Muse、GLM和Kimmy。然后左边是你开始看到来自ZAI和Memo的GLM，还有你你开始明白了，嗯，所有中国人。是的，J Cal？我的意思是，这很重要。需要注意的是，是的。你来到这里这件事情本身就说明了一切。瞧，这就是价格位置，但如果你将独立运行此程序，他们中的一些人成本小于每百万代币10美分。是的，所以你会得到计算、带宽、其他一切。但是扎克，这里的问题在于：查马特的观点是：垂直方向的语言模型，舌头较小的型号，将执行许多公司所需的80%或90%的任务，然后我们将谈到人工智能主权。你想要你的超级智能，“总统”惊呼道。特朗普，特朗普，“情报”，聊天DJT？我们人们需要用了这些模型以便它们有主权，以便你他们没有转移他们的边缘数据模型前面，这样他们就不会获得了这种优势超级智能因为人们拒绝。所以，你对此有何看法？这可以开始影响Claude和OpenAI的代币速度吗？你想要什么？您的企业建议，大卫·萨克斯，你有什么感受？听听这怎么说投资者，我更倾向于……乐观地倾向于边缘业务。在我看来，前面比前面更重要。
Original English
**J Cal**: No, it's just moving workers, loading where, getting a place to eat, and then saving money. Money. Yes, so, Sacks, I drew a chart here. Before the show, but here's your, uh, perfect! The average of the last 100 days of model showing, what? Friedberg's fine. Ready. And you, you'll find, yes, the quad and Astra are in the bottom right, and they're expensive, and they have a lot. Provided, but then, down, down, you start to see Muse, GLM, and Kimmy. And then on the left is where you start to see GLM from ZAI and Memo, and you, you start to understand, hmm, all the Chinese. Yes, J Cal? I mean, this is important. It's important to note that, yes. The fact that you came here says it all. Look, this is the price position, but if you're going to run this independently, the cost of some of them is less than 10 cents per million tokens. Yes, so you get the compute, the bandwidth, everything else. But Sacks, the question here is: Chamath's point is: the vertical, language models, smaller tongue models, are going to do 80 or 90% of the tasks that many companies need, and then we're going to talk about AI sovereignty. You want your super intelligence, "the president" exclaims. Trump, Trump, "intelligence," chat DJT? We, people need to use these models so that they have sovereignty, so that you, they're not transferring their edge data models to the front, so that they're not getting this advantage of super intelligence because people are refusing. So, what are your thoughts on that? Can this start to impact the token velocity of Claude and OpenAI? What do you want? Your business advice, David Sacks, what are your feelings? Hear this out as an investor, I'm more inclined to be... optimistically inclined towards the edge business. The front is more important than the front, in my opinion.
**David Sacks**: 各位的意见。我认为Anthropic和OpenAI目前稳定，双头垄断即将形成。因为他们具有相当大的凌驾于所有其他公司之上的优势，而且他们可以收取高价。我完全明白大多数代币将用于开放模型或商品情报。但是，我不知道，市场中显著的百分比——我不知道这是否是10%、20%、30%，随便什么都行——这将用于支付这笔费用。一个巨大的奖项，真正处于情报前沿。有许多顾客，或要求最好，或者他们只是想知道，它们最棒的地方是什么？你知道，他们不想要这种自制方法。或者他们可以在一个竞争非常激烈的市场中。假设你是一家对冲基金，你不能让你的风险，竞争对手有一个比你更好的模型。所以，我认为有很多情况将继续变现发达市场奖。这就是原因。你看，那份收入持续增长，即使他们损失了代币市场的份额。此外，你必须记住这一点，在接下来的大约一年前，约占世界总人口的60%的计算结果表明，添加后，将会是应添加这些两家公司。所以，因为计算资源变为赤字，以及我们处于某种计算机危机状态，正是这一事实，他们制作了这些巨额投资，它们正在增加如此强大的力量，他们有这个规模节省，以及记住他们速度也很快降低自身价格。我认为这给了他们相当大的优势，但是，你看，这两个公司位于漂浮着，你知道，在当他们停止在线限制，他们会去零，对吧？我们知道极限是，可能只适用于6-12个月后商品部分市场情况取决于机会。所以，你知道，再说一遍，如果他们会滑倒，会从这里掉下来，车轮用了6个月，然后他们会发现自己身处大麻烦公司，还有这个，我认为它至关重要。风险因素可以考虑在内，而且我想，就在这里。他们在国家事务领域的努力，强烈已计算得出。我理解他们为什么这样做。努力监管。那是激情，不是吗？因为漂浮在水面上非常累。他们可能在想：“我们来试试看，我们来烤一个吧，‘灰尘’。”我们开始吧，我们也试试吧。放慢所有人的
Original English
**David Sacks**: Your point. I think Anthropic and OpenAI are currently stable, and a duopoly is forming at the front. Because they have a considerable amount of leverage over all the other companies, and they can charge a premium. I fully understand that the majority of tokens will be used for open models or commodity intelligence. But, I don't know, a significant percentage of the market—I don't know if that's 10%, 20%, 30%, whatever—that will be used to pay for this. A huge premium, that's really at the cutting edge of intelligence. There are many customers who either demand the best, or they just want to know, what's the best they can get? You know, they don't want this homegrown approach. Or they can be in a very competitive market. Suppose you're a hedge fund, you can't risk your competitor having a better. More like a model than you. So, I think there are a lot of. Situations that are going to continue to monetize the developed market premium. And that's why. You look, that revenue continues to grow, even as they lose share of the token market. In addition, you have to remember that in the next, about a year ago, about 60% of the world's compute was being added, and it was going to be added to these two companies. So, because compute resources are becoming a deficit, and we're in some kind of a compute crisis, the very fact that they've made these huge investments, and they're adding such powerful power, they have this economy of scale, and remember they're also very quickly dropping their own prices. I think that gives them a considerable amount of advantage, but, you look, these two companies are floating, you know, in the, when they stop online, the limits, they're going to go to zero, right? We know the limit is, it's probably only for 6-12 months after the commodity part of the market is dependent on opportunity. So, you know, again, if they slip, and they fall off this wheel for 6 months, then they're going to find themselves in big trouble as companies, and this, I think it's critical. The risk factors that can be included, and I think, right here. Their efforts in the field of state affairs are strongly calculated. I understand why they're doing that. The effort to regulate. That's passion, isn't it? Because floating is very tiring. They're probably thinking, "Let's try to, let's bake a, 'dust.'" Let's start, let's try to too. Slow everyone's
### 人工智能监管与全球竞争
**Jason Friedberg**: 速度。其他人，但我认为他们误判了形势。因为如果他们真的会得到所有监管他们拥有的控制权，如果他们想要创建这个新的联邦人工智能，这种——它们太多了。慢点说，它们是什么？将不再有边界。其余所有人会赶上的，尤其因为其中许多开放模型被中国人创建，不属于我们的管辖范围，而且中国明确表示不会减速。所以好好想想。他们游说以争取某些可以减速的事情，这会让他们失去机会，在这种情况下，他们的业务量将减少到零。你知道，这很难理解。他们正在超越竞争对手，却在比赛中询问：“嘿，你能不能阻止我们检查发动机，Chamath？”
Original English
**Jason Friedberg**: Speed. Other people, but I think they misjudged the situation. Because if they really get all the regulatory control they have, if they want to create this new federal artificial intelligence, this kind—there are too many of them. Slow down, what are they? There will be no more boundaries. Everyone else will catch up, especially because many of these open models are being created by Chinese people who are not under our jurisdiction, and China has clearly stated that it will not slow down. So think about it. They lobbied for things that could slow them down, taking away their opportunities, and in this situation, their business volume would be reduced to zero. You know, this is hard to understand. They are outcompeting their rivals and asking: "Hey, can you... stop us from checking the engine, Chamath?"
**Chamath Palihapitiya**: 而且也不一定非得如此逻辑意义。除了查马特，你拥有那个。非常重要的竞争，不仅仅是从中国方面来看，但也有Jensen和Nvidia，你还可以使用微软的Satya，在家使用Google和Muse扎克伯格。他们所有人都支持模型开源，以及现在他们都有了托管。
Original English
**Chamath Palihapitiya**: And it doesn't necessarily make logical sense. Besides, Chamath, you have that. Very important competition, not just from the Chinese side, but you also have Jensen and Nvidia, you can also use Microsoft's Satya, and at home, Google and Muse Zuckerberg. All of them support open-source models, and now they all have hosting.
**Jason Friedberg**: 是的，我可以回到那个话题，萨奇说的话，是什么？非常重要且一个微妙的时刻，我只是想双击。他说：“想象一下，你是一只对冲基金。”而且你需要最新款和这是最好的，因为你担心有人来自你的竞争对手有。这真是 有趣的问题。为了它涉及博弈论和你的损益表。所以，现在你已经进入了这场“胆小鬼游戏”与他们的竞争对手。你需要使用最好，因为你害怕他们可以使用最好的超越你。但出了什么问题？现在我们知道你可以食用过量代币，对吧？我们开玩笑地称它为最大化代币。这笔费用绝对不是被使用和不与你的收入。它这太不公平了。和突然间，你拥有这很奇怪当你 的动态你心想：好 吧，我应该吗？我用这个。最新款和最好的东西比现在贵10倍、20倍甚至30倍普通的东西？这东西没有连接到我的收入。我不能。是要计数吗？错误的？所以如果对冲——该基金仅收到n美元利润一个月有很多他所处的场景可以成为盈亏平衡或甚至不盈利。和现在我们来看另一个例子。假设你出售固定价格的商品。我再次说道。我想你已经举了一个例子。在那里你可以使用最新和最好是这样，如果你不这样做的话你能把这些搬走吗？费用，夸大您的支出支付费用这项服务，您你会发现自己身处一个非常处于困境之中。所以我不确定，萨克斯，我该拿它怎么办？
Original English
**Jason Friedberg**: Yes, I can go back to that topic, what Satya said. It's a very important and subtle moment, I just want to double-click on it. He said: "Imagine you are a hedge fund." And you need the latest and greatest, because you're worried that someone from your competitor has it. This is a really interesting problem. It involves game theory and your P&L. So, now you've entered this "chicken game" with your competitors. You need to use the best, because you're afraid they might use the best to surpass you. But what went wrong? Now we know you can consume an excessive amount of tokens, right? We jokingly call it "token maximization". This fee is definitely not being used and is not tied to your revenue. It's just too unfair. And suddenly, you have this strange dynamic where you think: "Okay, should I use this? The latest and greatest stuff is 10, 20, or even 30 times more expensive than the regular stuff now?" This thing isn't connected to my revenue. I can't. Is it going to count as a mistake? So if a hedge fund only receives 'n' dollars in profit a month, there are many scenarios where it could be break-even or even unprofitable. And now let's look at another example. Suppose you sell fixed-price goods. I say again, I think you've already given an example where you can use the latest and greatest. If you don't do that, can you move these fees, inflate your expenses to pay for this service, and you'll find yourself in a very difficult situation. So I'm not sure, Sachs, what should I do with it?
**Chamath Palihapitiya**: 我同意，因为我没看到。这些产品的定价权公司同时支付一大堆这种超级昂贵的代币和转账将它们介绍给他们的客户。和如果你不能，他们只是吸收它们，然后你你会看到它减少。利润。为了发展这一点，萨克斯，我走了。还给你。
Original English
**Chamath Palihapitiya**: I agree, because I don't see it. The pricing power of these products, companies are simultaneously paying a bunch of these super expensive tokens and transferring them to introduce them to their customers. And if you can't, they just absorb them, and then you'll see it reduce profits. To elaborate on this point, Sachs, I'm done. Back to you.
**David Sacks**: 如果你带简街，这条街非常有名，贸易公司它们是公开的宣布云合约，19岁的力量数十亿美 CoreWeave——60亿云端美元义务。他们也投资了这家公司。然后是13来自克鲁索的数十亿。他们建造他们自己的基础设施和接受开放代码。所以，这说明……关于查马特，你说话。他们是否在对冲风险？他们要走了。花费数十亿美 元自己的计算和使用前排模特。似乎当你使用，你知道，博弈论。他们说：“嘿，来吧。”让我们自己来吧基础设施，萨克斯。”
Original English
**David Sacks**: If you take Jane Street, a very famous trading firm, they publicly announced cloud contracts, 19-year power, billions of dollars. CoreWeave—$6 billion in cloud obligations. They also invested in this company. Then there's $13 billion from Crusoe. They build their own infrastructure and accept open code. So, this says something... about Chamath, you speak. Are they hedging their bets? They're going to spend tens of billions of dollars on their own compute and use front-row models. It seems when you use, you know, game theory. They say: "Hey, come on. Let's build our own infrastructure, Sachs."
**Chamath Palihapitiya**: 听着，我我认为当你选项使用方式将成为更成熟，而你你会知道你可以为了更便宜商品情报那么显然你试试这个发展。再次是的，如果你有难度级别这样做。我我只是觉得容易被低估，真方便简单地使用边境。再次是的，这些公司速度也很快他们正在降低价格。但无论如何，听着，我们已经谈过这件事。我我认为最终，听着，不。怀疑开放周末代码量非常大。S1的风险因素人类的，而且应该是与……处于同一水平他们面临的风险即将结束出于人性或其他什么相似的。但是再次总之，我要回去了。在我看来，事实是观点，基本的这个问题公司由以下部分组成：因为她精神分裂症患者。我有意思是他们需要去看精神科医生，而且我不是银行家。他们继续这样做虚伪的自相矛盾的事物。和一方面，他们他们谈论的是速度前沿开发，新版本什么时候发布？前沿模型。他们他们说生物风险可以以……结尾人类打开实验室人工圣地亚哥情报部门弗朗西斯科。他们他们说他们的企业依赖于他们是否会跑赢大宗商品价格模型，它们提倡创建联邦部门人工智能，究竟是哪一个？减速。也许，就他们而言将成为一种商品。所以在这一领导层中公司正在发生从根本上讲，我会据说，腐烂，当它们代表可能发生的事情不符合他们的利益。现在人很多他们说这监管激情，以及这种真相。但我认为这是监管方面的。激情是政府官员的尝试对此进行合理化解释精神分裂症和让她工作为了你自身的利益。我不是我确信这是完全已计算得出，因此说话。
Original English
**Chamath Palihapitiya**: Listen, I think when your options for usage become more mature, and you'll know you can get commodity intelligence cheaper, then obviously you'll try this development. Again, yes, if you have the difficulty level to do this. I just think it's easy to underestimate how convenient it is to simply use the frontier. Again, yes, these companies are also very fast, and they are lowering prices. But anyway, listen, we've talked about this. I think ultimately, listen, no. I doubt the open weekend code volume is very large. The risk factor of S1 is human, and it should be at the same level as the risks they face that are about to end due to human nature or something similar. But again, in short, I'm going back. In my opinion, the fact is, the basic problem is that the company is composed of parts: because she is schizophrenic. I mean, they need to see a psychiatrist, and I'm not a banker. They continue to do these hypocritical, contradictory things. And on the one hand, they talk about the speed of frontier development, when will the new version be released? Frontier models. They say that bio-risk can end humanity. Open labs, AI intelligence departments in San Diego, San Francisco. They say their businesses depend on whether they will outrun commodity-priced models, and they advocate creating a federal department of artificial intelligence. Which one is it? Slow down. Perhaps, as far as they are concerned, it will become a commodity. So, in this leadership, what is happening in the company is, fundamentally, I would say, decay, when they represent things that might happen that are not in their interest. Now many people say this regulatory passion, and this truth. But I think this is the regulatory side. The passion is the attempt of government officials to rationalize this schizophrenia and make her work for your own benefit. I'm not sure this is completely calculated, so speak.
**David Sacks**: 是的。让我想问问大家，你认为，将会有一种禁令方法超级智能？伯尼·桑德斯建议这样做关于法律草案禁令超级智能。和你 知道，民主党人占领议事厅代表们，他们控制参议院，并那么或许在28号正在接管白宫。然后你发现自己身处世界里也有类似的景象事情就是这样奇怪的政治性，不应该发生的事。我不明白我们如何最终……政治分裂关于进展情况。他们会这样做。要求了。这里原因。他们会这样做。要求了。不，就是这样。真是太糟糕也太奇怪了，你竟然……通过反对进步你的政治派对。这一切有偏见。这是一堆东西。奇怪的扭曲世界观，因此尝试创建反对人民，你没有我喜欢它，但是这个非常奇怪。在任何总之，我想要即便如此，也要说，共和党人，关于你听到咕哝着，嗯，或许我们应该禁令中心数据处理，禁止人工智力。所以，萨克斯让我只是想问一下，多么务实所有这些方式开放模型开源？或者存在务实之路大家都在谈论什么我们正在谈论什么，或者我们正在谈论什么伯尼说，而且明确禁止超级智能停止人工智能，关闭它 是？真的有吗？有什么办法吗？什么如果这种情况发生，就会发生该法案将会通过明天？你觉得呢？有人认为，这事会发生，就是这样。我的问题是什么？如果账单伯尼被接受了明天，一切会停止，因为：据他所说人工的定义超级智能他的账单有人可能会这样认为：我们已经实现了这些阈值。我穿着我的意思是，你怎么能……你知道，基本上就是这样。只是在谈论模型……是的，在一定程度上机会，以及我我认为这是有可能的。争论是否无论我们是否实现了这个目标。所以，这个范围非常广。定义，以及关于我还要补充一点，他规定为期20年监禁对于开发者而言他们正在违法。所以没人会想要风险。到底发生了什么事？这太可怕了影响？你认为人们删除打开他们的体重模型桌面电脑？或者企业应该如何运作从您的代码中删除代码系统？这些人都 是吗？他们丢了工作吗？
Original English
**David Sacks**: Yes. Let me ask everyone, do you think there will be a prohibitive approach to superintelligence? Bernie Sanders suggested doing this regarding a draft law to ban superintelligence. And you know, Democrats occupy the House of Representatives, they control the Senate, and then perhaps in '28, they're taking over the White House. Then you find yourself in a world where there are similar scenes, things are just so strangely political, things that shouldn't happen. I don't understand how we will ultimately... have political division over progress. They will do this. It's demanded. Here's why. They will do this. It's demanded. No, that's it. It's so bad and so strange that you would... oppose progress through your political party. This is all biased. It's a bunch of things. A strange distorted worldview, and therefore trying to create opposition to the people, you don't have... I like it, but this is very strange. In any case, I want to say, even so, Republicans, about what you hear grumbling, "Well, maybe we should ban central data processing, ban artificial intelligence." So, Sachs, let me just ask, how pragmatic are all these ways of open models, open source? Or is there a pragmatic way, what everyone is talking about, what we are talking about, or what Bernie said, and explicitly ban superintelligence, stop AI, shut it down? Is there really a way? What if this happens, what if the bill passes tomorrow? What do you think? Some people think this will happen, that's it. My question is, what if Bernie's bill is accepted tomorrow, will everything stop, because: according to him, the definition of artificial superintelligence in his bill, someone might think: we have already reached these thresholds. I mean, how can you... you know, basically that's it. Just talking about models... yes, to some extent, opportunities, and I think this is possible. Arguing whether or not we have achieved this goal. So, this scope is very broad. The definition, and regarding... I would also add that he stipulates 20 years imprisonment for developers who are breaking the law. So no one will want to take the risk. What exactly happened? What are the terrible implications? Do you think people will delete open their weight models from desktop computers? Or how should businesses operate by removing code from your systems? Are these people? Are they losing their jobs?
**David Sacks**: 听，让我告诉你民主党人想要与人工有关智力就是它们由……制成加密货币。他们将会被取代国外的整个行业。他们要走了。实际上驱逐所有创新者各国，以及它们将出国，我当时要去的地方。它如何轻松转移贵公司新加坡，搬家她前往苏黎世，将其移至地理位置人们接受开源和接受超级智能系统。我拥有你头脑一切正常：你你无法阻止人工智能。你不能就这么告诉人们一些班级软件规定不能存在。在我的我心里想：好 吧，如果美国通过了一项法律他 说这个班级软件规定不能存在，我会搬到了另一个国家。对其他人我会喜欢……但我并不这么认为。不是人们所想的会拒绝他们的公民身份并会搬到新加坡？
Original English
**David Sacks**: Listen, let me tell you what Democrats want regarding artificial intelligence is that they are made of... cryptocurrency. They will replace the entire industry abroad. They are going to actually expel all innovators from various countries, and they will go abroad, where I was going at the time. How easy it is to move your company to Singapore, move her to Zurich, move it to geographical locations where people accept open source and accept superintelligent systems. I have... everything is normal in your head: you cannot stop artificial intelligence. You can't just tell people that a certain class of software is not allowed to exist. In my mind, I think: "Well, if the United States passes a law that says this class of software is not allowed to exist, I will move to another country." For others, I would like... but I don't think so. It's not what people think, that they will renounce their citizenship and move to Singapore?
**Chamath Palihapitiya**: 答案如下。人们字面理解这种人工智能...G. G.今天发言了在白宫和他 说：“我将邀请100,000名年轻人美国人即将到来去看看中国。是的，请来。看看我们这里有什么“我们确实有。”这非常有趣的时刻我们的人民说：“我们停下来吧。进步，技术性的，经济的繁荣，机会每个人都在这个拥有经济体字母“K”的形状上升是因为人工智能与开源允许所有人增加生产率。并非如此对于亿万富翁来说，它人工智能每个人。然后我们我们将这样做关闭，同时中国方面表示：“过来，来吧看看这里是什么样子，每个人都可以使用人工智能和“人人都能建造。”
Original English
**Chamath Palihapitiya**: The answer is as follows. People literally understand this kind of artificial intelligence... G. G. spoke today at the White House and he said: "I will invite 100,000 young Americans to come and see China. Yes, please come. See what we have here." "We do have." This is a very interesting moment when our people say: "Let's stop. Progress, technological, economic prosperity, opportunity for everyone in this economy shaped like the letter 'K' rising is because AI and open source allow everyone to increase productivity. It's not just for billionaires, it's AI for everyone. And then we will do this shutdown, while the Chinese side says: "Come over, come and see what it's like here, everyone can use AI and 'everyone can build'."
**Jason Friedberg**: 弗赖贝格？明亮的说。这是我们的。行动计划邀请领导人中国科学家领导欧洲科学家，顶尖科学家思想家和来自印度的数学家们，德国和邀请他们到达美国。它这是我们的行动计划。中国正在采纳我们的计划采取行动，然后试图吸引投资”。显然，接下来凡是掉在这里的人，都将……特朗普总统今年30岁组织中的秒数联合国，谁重新命名了人工智能关于超级智能。美国也完全拒绝任何尝试创造全球主义计划对人工的控制关于智力，有很多话要说，现在正式受邀超级智能。更改姓名。同样的那些说这话的人我们都会因此而死。12年前全球暖化。这些人就是现在他们说人工智能这会害死我们所有人，就是这样机器人将发起攻击我们。
Original English
**Jason Friedberg**: Friedberg? Brightly said. This is our action plan to invite leading Chinese scientists, leading European scientists, top scientists, thinkers, and mathematicians from India and Germany, and invite them to come to the United States. This is our action plan. China is adopting our plan of action, and then trying to attract investment." Obviously, whoever falls here next will... President Trump, 30 seconds into his organization at the UN, who renamed AI to superintelligence. The US also completely rejects any attempt to create a globalist agenda for the control of artificial intelligence. There's a lot to be said, now officially invited superintelligence. Change the name. The same people who said this, that we would all die from it. 12 years ago, global warming. These are the same people who now say AI will kill us all, that robots will attack us.
**David Sacks**: 哦，是的，的确如此。一句非常好的短语。我们……我们太……了短暂地打断了她的话。他说我们没有事实上，我们正要出发。提供监管某些权威全球主义者机构。嗯，我的意思是，这个成功配方，难道不是吗？全球化与全球暖化这是看待这个问题的一种方式。为造成的损失。这些领导人全球主义者挪威的机构以及加拿大。你见过吗？文本泄露消息？有件事……像一群人聊天，例如，挪威领导人与……联系马克·卡尼。而他像这样：“是的，我们……”我们开始做吧。我们正在团队。”而我只是……读一下这个心想：“我的天哪，这 是……”荒谬地”。
Original English
**David Sacks**: Oh, yes, indeed. A very good phrase. We... we were too... briefly interrupted her. He said we didn't, in fact, we were about to go. Provide regulation for certain authoritative globalist institutions. Well, I mean, this successful recipe, isn't it? Globalization and global warming, this is one way to look at this issue. For the damage caused. These globalist leaders, institutions in Norway and Canada. Have you seen? The leaked text messages? There was something... like a group of people chatting, for example, the Norwegian leader contacted Mark Carney. And he was like: "Yes, we..." "Let's start doing it. We're on the team." And I just... read this and thought: "Oh my god, this is..." "ridiculous."
**Chamath Palihapitiya**: 边界实验室。在，抱歉，他们有这些不。哦，和他们边界企业，你们有边防部队好的。然后这里贝赞特将军超级智能军队。将返回30秒让他闪耀光芒。想象一下这些实验室出来了或者某个特定的人实验室，员工已离职据说有10%事件发生的概率灭绝水平，但是然后是实验室还说：“把它从我们身上拿走责任和我们不我们会的。这些人携带责任和并非人工智能。”事件拥抱脸-它责任管理开放人工智能，而不是一群特工。这些实验室必须承担责任。他们可以想办法慢下来他们随时都可以。最后，发生了爆炸过去，你的朋友，奥巴马总统，查马特，会干扰这一点。如果我们考虑仅从人工智能的角度来看癌症治疗或获得最佳如果能量充足，那么就有可能。不用智能体人工智能，自由漫游互联网。原因，你为什么要这样做？那是你吗需要推广一款人们使用的产品会付钱。它差异需要什么我们的社会，和商业带有该命令的这些面临公司，而不是因为他们必须正在尝试做坏事，因此，他们应该请对这些估算结果进行核实。弗里德伯格，你的想法关于这些中的任何一个这里有三个人……我我知道你很棒。奥巴马的粉丝。
Original English
**Chamath Palihapitiya**: Frontier labs. In, sorry, they have these no. Oh, and their frontier enterprises, you have border guards. Okay. Then here, General Bezant, superintelligence army. Will return in 30 seconds to let him shine. Imagine these labs came out, or a specific person's lab, employees have left, it is said there is a 10% probability of an extinction-level event, but then the lab also said: "Take the responsibility away from us, and we won't. We will. These people carry the responsibility and not AI." Event embrace face - it's the responsibility of managing open AI, not a group of agents. These labs must take responsibility. They can find ways to slow down whenever they want. Finally, there was an explosion in the past, your friend, President Obama, Chamath, would interfere with this. If we consider only from the perspective of AI for cancer treatment or getting the best, if there is enough energy, then it is possible. Without intelligent agent AI, freely roaming the internet. The reason, why would you do that? Is that you need to promote a product that people will use and pay for. It's the difference between what our society needs and what businesses face with that command, not because they must be trying to do bad things, and therefore, they should please verify these estimates. Friedberg, your thoughts on any of these three people here... I know you're great. An Obama fan.
**Jason Friedberg**: 好。弗里德伯格好。海湾山你 知道，现在在
Original English
**Jason Friedberg**: Good. Friedberg good. Bay Hill, you know, now at
### 人工智能与政治经济
**Host**: 子版块存在。整个运动都围绕着弗里贝格。在他们最近在Bayside的亮相中，伯格……好吧，你已经揭露真相了。你的基地，贝格。贝赛德伯格现在将会显现。好的，分享你的仔细考虑一下。
Original English
**Host**: Substack exists. The entire movement is around Freiberg. In their recent appearance in Bayside, Berg... Well, you've blown the lid off. Your base, Berg. I Bayside Berg is now going to manifest. Okay, share your considered thoughts.
**Jason**: 好。你和埃隆·马斯克。说实话，这些话简单来说让我情绪低落。如果他是一位领导者，人们会听的，他确实没有完全正确地表示。然后，事情就发生了。我们输了，真令人难过。我理解。我认为令人遗憾的是，我们没有手牵着手。我们不会说，“哦，我的天哪，未来已来，我们都可以跳向前迈进并教育人们。”多年来，奥巴马发言非常精彩，谈论教育的重要性，关于给予人们发展的可能性，关于允许人们独立发展。而且从未有过比这些技术、比这些工具更好的技术水平、经济机会和全民繁荣。他挥手说“代理人”一词，他说我们不需要智能体人工智能用于癌症治疗，这说明他多么渺小，真正理解这项技术是如何有效的。以及这一切是如何发生的，简直就是政治对抗。对抗什么？为了什么？你遇到困难了吗？这是什么？政治斗争。他说：“我会离开一半人口后面”，和“我将提名反对全民繁荣”，这样我的党看起来很道德，更高尚，而另一边——“不好”，当另一边字面上结果发现是火灾，而且世界正处于水深火热之中，我们却要采取水，然后扔进火里？没有人被解雇的原因是其他人说：“那场火是好事。”因为他们说那场火是好的。咱们都把它关掉吧，火。等等，他们没有正在尝试这个制作。这非常重点是，原因很简单：我们，就像全世界一样，我们要给三、四、五、六，大约10家公司数万亿美元财富。还有奥巴马，他非常清楚这一点，这些中的大多数公司主要持有左翼观点。他也知道这一点。很大一部分钱以后会到账，旨在慈善和慈善目的，其受益人是他和他周围的人。他们会变成他。这是真相。我们已经有了。我们知道，因为我们知道其中一些边界公司实际上你正在问你抵抗DAF和获取零件你的股份，但是你准备贡献一点点。所以，这笔钱会去做事情，与消费或节省不同。因此，他们将前往政治行动委员会（PAC）。他们将前往政治运动。他们将收到不成比例的巨大益处。所以，这不是没有任何共同之处的繁荣昌盛。它非常简单，是政治的计算。如果经济冻结，就像今天这样，他们会赢。多个组织，谁将能够不成比例地影响民主党人。他们将捕获经济份额的狮子，取得成就，然后会有帮助，民主党人将获胜，获得力量。就这样。
Original English
**Jason**: Okay. You and Elon Musk. Honestly, these words are simply depressing to me. If he were a leader, people would listen, and he really didn't quite correctly state. And then, things happened. We lost, which is sad. I understand. I think it's unfortunate that we don't have our hands together. We're not saying, "Oh my God, the future is here, and we can all jump forward and educate people." For years, Obama spoke so beautifully about the importance of educating people, about giving people the possibility to develop, about allowing people to develop independently. And there has never been a level of technology, better economic opportunity, and prosperity for all, than these technologies, than these tools. And his waving of the word "agent," where he says we don't need agent AI for cancer treatment, shows how little he truly understands how this technology is effective. And how this is all happening is simply political. Political against what? For what? Are you having trouble? What is this? Political fighting. He says, "I'm going to leave half the population behind," and "I'm going to nominate against prosperity for all," so that my party looks moral and higher, and the other side—"bad," when the other side literally turns out to be on fire, and the world is on fire, and we're going to take water and throw it into the fire? No one is fired because other people say, "That fire is good." And because they say, that fire is good. Let's all turn it off, the fire. Wait, they're not trying to make this. This is very much the point, for a very simple reason: We, like the entire world, are going to give three, four, five, six, about 10 companies trillions of dollars of wealth. And Obama, he's very clear about this, that most of these companies are primarily left-leaning. And he knows that too. A large portion of this money will come later, intended for charity and philanthropic purposes, and its beneficiaries will be him and the people around him. They will become him. It's the truth. We already have it. We know, because we know some of these frontier companies are actually you're asking you to resist DAFs and get pieces of your stock, but you're prepared to contribute a little bit. So, this money is going to go to do things, unlike consumption or saving. So they're going to go to PACs. They're going to go to political campaigns. And they're going to receive disproportionately large benefits. So, this is not about prosperity for all. It's a very simple political calculation. If the economy is frozen, as it is today, they will win. Multiple organizations, who will be able to disproportionately influence Democrats. They will capture the lion's share of the economic achievement, and then it will help, and Democrats will win power. That's it.
**Host**: 大卫·萨克斯，不，开放模型，没有关闭模型，有美国模型。让我们分享一下你的看法。
Original English
**Host**: David Sacks, no, open models, no closed models, there's an American model. Let's share your perspective.
**Host**: 我做不到。分享自己的奥巴马和他的克林顿。它们本质上是同一个人。同一个人。我有点弄错了。你有点误解了。我可以把它带过来。我的鸟，但是快点。你可以分享你关于动力学的想法，在于基于以下想法：“嘿，美国需要获得10万亿美元对于这一趋势，但其中十万亿中有九万亿将是倾向于相当左和民主观点。”是的。
Original English
**Host**: I can't do it. Share my own Obama and his Clinton. They're essentially the same person. The same person. I got it a little bit wrong. You got it a little bit wrong. I can bring it over. My bird, but quickly. You can share your thoughts about the dynamics, in that it's based on the idea that, "Hey, America needs to get $10 trillion for this trend, but nine of those $10 trillion are going to be leaning pretty left and Democratic in their views." Yes.
**David Sacks**: 今天早上在《华尔街日报》已发布一篇文章，标题为“人工智能技术的发展是美国历史上最大的经济利率”，然后显示了到数据处理中心的费用明细表，也就是说，只有这些数据。资本支出比通道、铁路和网络综合起来还要大。这真是太疯狂了，不是吗？这是根本性地管理整个美国经济。这个想法关于我们能做什么，关掉就行了……伯尼·桑德斯实质上是说：“停止一切行动，投入监狱仍然这样做的人正在从事。”这真是太疯狂了，不是吗？是抑郁症，对吧？如果我们只是屏蔽经济，速度会像这样慢下来……这行不通。但在这里怎么了？民主党人不在乎。因为美国人当前的经济形势是特朗普的经济政策。破坏美国经济等同于破坏总统特朗普。而且……他们会竭尽全力破坏特朗普的计划。所以，你知道，我认为，这是很重要的一部分。这里涉及政治。棋盘，因为它们将成为如此强大力量的受益者，这些物质会在那里积聚。达里奥可以赚十亿美元。达里奥能赢得一切，以及他的组织OpenAI可以赢得一切，任何事，超过埃隆的捐款。我不是确定我完全同意。因为它更大。人工智能的部分价值将获得六个公司。我认为这是一项应用广泛的技术。例如互联网。也有一些赢家，从互联网、谷歌、亚马逊等公司中脱颖而出。他们做得很好。不，不，我不是说就是这样。将会得到六家公司。我不是说这一定会发生。六。我说如果你会停下来，按下按钮，由于某些原因而暂停监管机制，你实际上允许六个公司要去...你记下来。正确的。这就是他们这样做的原因。想。因为你……因为你，你说，有从根本上说……正确的。时间越长他们正在等待，却……因此，需要进行监管。这个正在变宽的情况。这对民主党人不利。这很糟糕。对他们来说。这可不好。对他们来说。很多你最好有三个或四个边界公司，每个具有市场价值资本化四万亿美元，其中两万亿——那是经济价值以及员工，其中5000亿人坐等消息，你知道，就是用在DAF基金里。他们将收到这些中的绝大部分钱。他们并不愚蠢。我认为备选方案，人人皆知的观点需要说明，这是一股广泛的繁荣浪潮。这是关于互联网。这是关于每个人都有工具。她做的一切，现在，他想要更多。比以往任何时候都更甚历史。如果你留下它生长。我的意思是，就思考一下教育问题。你不再需要支付20万美元用于教育。那么问题出在哪里呢？政治家们会做出决定吗？它们是什么问题？可以决定，为了获得权力？如果所有问题，问题将会解决，他们...这是一个经典的解决方案。黑暗时代，启示。哪个道路……选择？而在这方面，仿佛就在那一刻，当局表示：“我们喜欢他们黑暗时代，因为我们可以留在权力方面支持她。”是的。我们开始吧。我们停下来吧。说到黑暗时代，这很有趣，这是一个历史性时刻。如果你回到中世纪，中华文明比欧洲的要先进得多。以及某个地方在案件审理过程中，它实际上已经改变了，而且西方实际上采取了主动。一些历史学家将此与由个人决定，中国皇帝禁止造船联系起来。唔。这使得……欧洲人实际上殖民了整个世界，以及由此而来的所有财富，返回欧洲。就是这样。欧洲接管了倡议。而且我认为，如果我们那样做呢？伯尼·桑德斯做了什么？就是他禁用人工智能，那将是看起来像是禁令造船。嘿，你看，中华文明，是的，是公司。是的，中华文明将会从我们身边经过。他们会这么做的。所有发现，将开启这个新篇章，世界和一切终将积累财富。但正是如此。这就是我们所说的。我的意思是，故事发生了，真是疯狂的事，你知道，当你参与了自我破坏。这是恐惧……（未知）还。记住关于未知的事物，关于那件事，我们没有离开洞穴，就是这样。这让她看起来很可怕。事实并非如此。远航事件是原因，为什么我会从天涯海角。我们不应该远航西方。怪物会醒来。他们会来到我们这里。情况会很糟糕。总是这种恐惧越过边境。我们遇到边界，并且有先驱文明，并且存在一些文明充满恐惧。我担心我们要以这种可怕的方式离开，这就是为什么“明星”，你知道，战争。恐惧袭来，愤怒，黑暗即将降临。确实如此。我们遇到特此。但这是真的。这是谁？没错，我认为如果我们我们将要观看向着这里的光芒，或许很大人们的成就，我当时想，或许这将是一场GPT聊天，但是人们只是将其用作类似著名的搜索。但是，人们，代理人，如何通过Muse表达，似乎是第一次数千万美国人将收到一些示范性的来自人工智能的价值。你知道，就是那些事，这将改善他们的生活。上周五Muse获得第一名应用商店。股票元股份之后增长了10%。发布这款新产品经纪人Muse。元注意到该设计深受Open的启发。我们正在谈论的是爪子。一月份时表示，当大家都……Open Claw的一次尝试，但很显然，这一个该产品是困难二手，而且是这个免费且下载了3数百万次，实际上还有10天。本产品非凡的。我玩过它，将其与Grokbot。他不是那样的人。自由，但他非常好地履行其职责任务。你玩了吗？Muse乐队的某位成员，并且做了某人对美国人的看法，真的有这种感觉，他们会得到一些东西。这场人工智能革命不会落后吗？他们让我进去试飞几周前发布，我开始用来做实验。他真是太棒了。我必须对你坦诚相待。你。因为他简化了许多此类问题，更复杂的技术，非常难得的机会，简单方便，可以解决主要问题，人们的问题。我设置好整理你的私人邮件盒子。他确实这是相对的。完美无瑕。如果你问他/她预订机票，它会做。如果你问他/她预订酒店，他能做到。那么，我们是什么？我们在学习吗？还有Grok Bot，以及Muse表明，你说什么，杰森，这很简单。软件。软件。和可扩展软件的表现形式提供有用性，以及什么弗里堡说道。它改善每个人的生活。这不是关于富人变成了更富有。很简单。宝贵的机会现已上市，免费，人人适用。免费是价格相当优惠。大卫和价格很不错。价格相当不错，而且他什么都做，你知道吗，我们走吧。让我们看看真相。眼睛，张开的爪子，10小时，20数小时的技术设置或代码，克劳德以200美元的价格，一个月的费用是……现在他使用此后来的经历，这让我相信：哇，谷歌居然错过了这一点。它？伙计，谷歌有进入游戏，但是，而且我知道他们有这个产品已准备就绪工作。这些都是谣言。街道。他们也有边缘处有一个模型。他们即将发布。另一个传闻这条街就是这样的。他们很快就会将会发布一些东西，很热，但是它们怎么样？萨克斯，你错过了吗？你对此有何看法？或许是底部那个半恢复这里以K的形式表示，其中一定的有效性，它们的减少日常生活以及变化负面毒性对人工智能的态度国家。我认为这样产品发布这将非常有帮助。我是说，马克·扎克伯格说在他的文章中，关于他的观点是，人工智能能力必须是去中心化的，然后他迈出了一步。并创造出产品，这非常简单使用。我从那一刻起，心中就一直想着这件事。Open Claw的发布，是最显而易见的商业可能性。硅谷以“爪”的概念为例，但要让它非常空闲时间使用。你知道，解决问题，便利设施使用，解决问题，安全，确保安全。可靠，让它可靠，可预测的。而这，它们看起来像是完毕。Grok Bot也是这样曾经有过，但现在我想他们要离开了。更进一步。所以，是的，听着，我认为如果十亿人即将开始使用，你知道，个人方面。人工智能代理作为它们的数字助理，只是为了做他们的生活更有效，你节省一两个小时。每天，因为你人工智能代理完成了所有工作，这些任务都交给你了。
### 人工智能的积极印象与市场影响
**Chamath**: 这将创造很多更积极的印象，关于人工智能或比媒体所描绘的更高级的智能。我认为最终……或许是个解决办法。公共议题的想法是，随着越来越多的人使用这些产品并喜欢它们，它们的支持率会更好。但更重要的是，这将消除人们对它们的误解，人们会更少害怕它们。人们喜欢美好的事物，首先是产品。人们也喜欢可爱的产品，杰森。还有那些标志以及 Grok Bot 和 Muse 的图像符号，它们都很美。它很迷人，没有那么大的压力。你不会觉得某种责任的号角悬挂在上方。你会觉得，“哇，这太好了，我很喜欢。喜欢它”。而且它们……它们不试图夺走你的工作。它们是机器人和你的助手。它们不是来杀你的。它们不吓人。它们不是杀人犯。它们不试图偷走你的工作。它们正在尝试帮助你把工作做得更好。我总是对三位创始人说，在世界上赚钱的方法是为人们省钱，为他们赚钱，或者娱乐他们。然后你看看这个机器人，我开始使用它。我创建了一个机器人商业活动，在两个平台之间，并表示：你知道，当我想买点东西时。我想买这个锚形的东西，你知道，这个新的Mac Studio集线器。而我心想：“这东西要400美元。” 纯属娱乐，告诉我我在哪里可以以最优惠的价格买到它。他说：“绝对可以。”你可以访问Anker网站并获取……如果这是你第一次购买，即可节省85美元。我不需要节省85美元，但我心想：“嗯，这很有趣”。然后我被重定向到亚马逊。我当时想，“好吧，我将从Anker购买这款产品。首次购买的买家可享受这款产品25%的折扣。”这就是亚马逊本周表示：“我们应该阻止这些事。”然后他们又再次使用事件。首先，他们制作了《困惑》，我认为他们现在是闹鬼的巷子和其他机器人。它将会为人们省钱。这将在他们的电子邮件中显示，问：“您订阅了哪些服务，但没有使用？让我为您取消订阅。允许我为您做。为您回来。”这绝对会非常适合所有需要一位领导者机构、执行助理、一个房子来帮他们解决问题，这样可以帮他们省钱的人。Grok Bot 和 Muse 都这样做。在我看来，有两件事非常重要。首先，它会强制所有大公司拥有居民服务。屏蔽所有这些第三方服务。他们不会允许这件事必须发生。原因就在于此。你说什么？有定价和透明度，我认为不透明度对……有害。如果你从不透明度中受益，泄漏和故障，这消除了这一点。而且我认为这一点对整个系统都不利，对多家公司和服务都不利。第二，真有意思。像 Grokbot 和 Muse 这样的东西，务必警告 App Store 及其30%的收入份额。哦，好的。为什么？因为突然你强迫了很多这些服务漫不经心地存在，或者错误的？界面在哪里？用户不那么重要。事实上，你想要的是能够执行交易，并从代表代理人那里获得，因为……消费者部署它们会出现在你的网站上。在这样的世界里，杰森，我不是这么认为。某种合理的声明，任何App Store 的所有者是谁？可以做些什么？为什么他们应该收到收入份额？所以，我确实这么认为。这是最大的开始对此感兴趣。等一下，仿佛游戏真的可以进行一样，会创造一个又一个，对吧？例如，如果你告诉 Muse 或 Grokbot：“我想玩游戏。”他们会找到你的经验和能力，只需提供给你。出奇。如果你这样做的话，应用内购买，没有理由，为什么 App Store 实际上接收此流资金。很快，表面是实用性，但水线，我想是吧。它到底是什么？重点。说得好，查马特。我很喜欢。当我们到达真实案例，因为很多人付费使用 Wordle。你可以直接联系 Grok：“嘿，给我 Wordle。”和“再给我来一个。” 而我，或者说，我想下棋，或者我想玩牌。和朋友们一起。很简单。我会帮你安排好一切。全部技术通货紧缩。最终，这种通货紧缩将触达人们，这实际上可能使用它们。你知道，可以节省10-20%。自费，如果你每年赚不到10万，也就是说更重要的是。想想看《纽约时报》或《华尔街日报》可能以收入份额的形式支付多少钱给那些报名的人。现在，情况却变成了如果你真的想执行交易并有您的凭据这些应用程序，你可以直接不假思索地去做这件事，很多很多更便宜。你不需要因此而通过。我不知道你是否曾经尝试过《华尔街日报》——唯一一家大众传媒出版我为此哭泣，但是或注册，或无法取消订阅。现在你拥有了这一切被带走后，我感到非常难过。更有可能我会留下来。作为《华尔街日报》的付费订阅者，还有更多内容。年。我认为这是，纽约的情况可能也一样。时代。这些地方是汽车旅馆。针对蟑螂。他们不可能既不能进去，也不能出去。你需要打电话，需要20分钟的同意才能取消订阅。但他们应该怎么做？不假思索地去做，只需点击一下网站。但他们应该怎么做？不假思索地去做，是的。就是拥有一些不经思考的经历，付款详情，仅供参考。执行交易直接地，以及所有收入会更高。这是件大事。App Store面临压力。这太愚蠢了。亚马逊这样做是因为我创建了他的机器人亚马逊、我的 Crock-Bot，它将穿过网络浏览器并会这样做，而且不会被屏蔽。我只是对他说：“给我大概20本图画小说，这样我就可以自己买了。“和孩子们一起读书。”他给了我这20本。他说：“是的，我想……”这四本。”呃，我把它放进篮子里，然后它说：“好的，选哪一个？你要发送到家里吗？你想要将此发送给高山滑雪的房子？你想要将此发送给牧场？是哪张牌？你想要这个吗？放？我放了？我放了它。砰，我已计算。它会让我花费更多在亚马逊上赚钱。 Shopify、萨克斯增加了 API 访问权限，所有 Shopify 商店都在那里。看来亚马逊制定战略错误，没有使用这种方法技术。
Original English
**Chamath**: This will create a lot more positive impressions about AI or intelligence that's more advanced than what the media portrays. And I think ultimately... that's probably a solution. The public discourse, the idea, is like the more people use these products and they like it, the better their approval ratings will be. But more importantly, it will remove the misconceptions about them, and people will be less afraid of them. People like nice things, first and foremost, products. And people also like cute products, Jason. And the logos and the iconographies of both Grok Bot, and Muse are beautiful. It's charming. It's not so stressful. You're not, you don't feel like there's some sort of horn hanging over the top of you of responsibility. You, you feel like, "Wow, this is great, I love it. Love it." And they're... and they're not trying to take your job. They're robots and assistants to you. They're not going to go kill you. They're not scary. They're not murderers. They're not trying to steal your job. They're trying to do your job better for you. I always say to the three founders, the way to make money in the world is to save people money, make them money, or entertain them. And then you look at this bot, which I started to use. I created a bot business activity between the two platforms and said, "You know, when I want to, I want to buy something. Buy this Anker thing, you know, this new hub for my Mac Studio." And I was like, "This thing, "It's $400." "Just for fun, tell me where I can buy it "at the best price." And I was like, he said, "Absolutely." You can go to the website Anker and get... if this is your first purchase, save "$85." I don't need to save $85, but I was like, "Hmm, that's interesting." And then I was redirected to Amazon. And I was like, "Okay, I'm going to buy this from Anker. Buyers get 25% off on their first purchase on this product." Which is why Amazon this week said, "We should block these things." And then they went and used the event again. The first, they made "Confused," which I, I think they are now the haunted alley and other bots. It's going to save people money. It's going to show up in their email and say, "What services do you subscribe to, "but don't use it? Let me unsubscribe "for you. Allow, I'll do it for you. "Come back for you." That's going to be absolutely fantastic for everybody who needs a leader agency, an executive assistant, a house to help them solve problems that will save them money. Grok Bot and Mews both do that. There are two things, in my opinion, that are very important. First, it's going to force all the big companies to have resident services. To block all these third-party services. They're not going to allow this thing to have to happen. And the reason why is this. What do you say? There's pricing and transparency, and I think what's bad for... is opaqueness. If you benefit from opaqueness, leaks and failures, this removes that. And I think that that is bad for the entire system of multiple companies and services. Second, this is really interesting. Things like Grokbot and Muse, are going to warn the App Store and its 30% share of revenue. Oh, good. Why? Because all of a sudden you've forced a lot of these services that exist haphazardly, or wrong? Where's the interface? The user is less important. In fact, what you want is to be able to execute transactions and from the behalf of the agent that they come from because... the consumer deploys them and they show up on your website. In a world like that, Jason, I don't think there's some sort of reasonable claim as to who the owner of any App Store is? What they can do? Why they should receive a share of the revenue. So, I, I really do think that. This is the biggest start to get interested in this. Wait a second, as if games can really be played, it's going to create one after another, right? For example, if you tell Muse or Grokbot, "I want to play a game." They'll find your experience and capabilities and just provide it to you. Amazingly. If you do that, in-app purchases, there's no reason why the App Store actually receives this stream of money. Very quickly the surface is utility, but the waterline, I think, right? What is it really? The point. Well said, Chamath. I love it. When we get to the real cases, because a lot of people pay for Wordle. You, you can just contact Grok and say, "Hey, give me Wordle." And "Give me another one." "And I, or I want to play chess, or I, I want to play cards. With friends. It's simple. I'll set it all up for you. All technical deflation. Ultimately, that deflation is going to reach people, which could actually use them. And you know, you can save 10-20%. Out of pocket, if you make less than $100,000 a year, that's to say more important. Think about how much money the New York Times or the Wall Street Journal might pay in the form of a share of revenue to people who sign up. And now, what happens is if you really want to execute a transaction and have your credentials for these applications, you, you can just do it without thinking, and it's much, much cheaper. You don't need to go through because of this. I don't know if you've ever, for example, tried the Wall Street Journal-- the only mass media publication I cry for, but or sign up, or can't unsubscribe. And now that you have all of that taken away, I'm very sad. More likely I'll stay. Wall subscribers of the Street Journal have more content. Year. I think this is, the same thing is probably true for the New York. Times. These places are motels. For cockroaches. They can't, they can't get in, and they can't get out. You need to call, get and 20 minutes agree to unsubscribe. But what should they do? Do it without thinking, just one click on the website. But what should they do? Do it without thinking, yes. Just have some thoughtless experiences, payment details, for reference. Execute transactions directly, and all the revenue will be higher. This is a big deal. The App Store is under pressure. This is so stupid that Amazon did this because I created his bot Amazon, my Crock-Bot and it will go through the web browser and will do this, and it will not be blocked. I just said to him, "Give me about 20. Graphic novels, so I can buy them myself. "And read with the kids." He gave me these 20. He said, "Yes, I want..." these four." Uh, I put it in the basket, and then it said, "Okay, which one? At home do you, do you want to send? Do you want to send this to the ski house? Do you want to send this to the ranch? Which card? Do you want this? Put? I put? I put it. Bang, I'm calculated. It will make me spend more money on Amazon. Shopify, Sacks increased API access to all Shopify stores there. It seems Amazon made a strategic mistake by not using this type of technology.
**Jason**: 也许。但是，你知道什么？我们了解到，OpenAI即将……将会推出他们自己的Muse竞争对手？我们怎么知道？将会发布大量的帖子、博客和活动，民间组织会尝试将这些个人代理人人工智能呈现为杀手机器人等等。我们会听到相关消息。所有风险，关于所有它可以通过多种方式，你知道，可能会出错。而现在，我们我们将发布 2.0 版本。所以这是类似问：为什么？人们为什么对这类产品如此快乐？因为梦想家们还没到来。在此之前。你知道，而且，呃——有，我可以告诉你一个有趣的故事吗？什么时候这与……有关，查马特，真正的机器人，带着乐观主义，然后突然间，这偏离轨道。在2007年，Facebook是巨大的收入压力，以及其中之一我当时所处的领域受控的，是货币化，以及我们发明了这个东西，称为社交广告，本质上是一堆东西，有效开启饼干和跨站 JavaScript 攻击。就是这样。结束了；那是圆满结局。但我们做到了。萨克斯，你说的对。视线，以及发生的事情是，那家伙买了你未来新娘的结婚戒指，我们放置了这个社会行动来自，例如，Zalesa，在新闻推送；一切由此开始地狱。然后又一个那个人，类型，国会员工，买通了票号，类型，凡丹戈观看《钟楼怪人》“山”……就这样，去看了电影。这是一部故事片。他位列前三。前十名中最好的布鲁克斯摄影。总之，我们把它放在那里用胶带粘起来，就像这样：我得到了所有这些指控或问题”。从某种角度来看，就好像它刚才刚经历了一场暴风雨般的糟糕情况。你什么时候做这件？它说，这让人想起我讨厌这一切混乱。这种情况发生产品发布逐个。你真是太搞笑了。之所以记得这件事，是因为……一家名为Blippy的公司。我曾是她的投资者。他们每次你做某事，已购买，已放置，它就在你的信息流里，而且然后你们，是同样的社交录像带，就是这样。那太好了。是的，你其实可以揭示某些事情，什么不应该那样做。污泥。扎克和我都关系良好吗？
Original English
**Jason**: Maybe. But, you know what? You know we, we learned that anthropology OpenAI is about to... is going to launch their own competitor to Muse? How do we know? There's going to be a lot of posts that are going to be put out on blogs and events and civil organizations that are going to try to present these personal agent AIs as killer robots and so on. We're going to hear about it. All the risks, about all the ways it can, you know, potentially go wrong. And now, we, we're going to release version 2.0. So this is like, asking, why? Why are people so happy about this category of products? Because the dreamers haven't arrived. Before this. You know, and, uh-- there's, can I tell you? A funny story? When this relates to... Chamath, the real bots, with optimism and then all of a sudden, this went off the rails. In 2007, Facebook was under enormous pressure for revenue, and one of the areas that I was in charge of, was monetization, and we invented this thing called social ads that was essentially a bunch of stuff that effectively turned on cookies and cross-site JavaScript attacks. That's it. It's over; that's the happy ending. But we did it. Sacks, you're right. The sight, and what happened was the guy bought a wedding ring for your future bride and we put this social action from, for example, Zalesa, in the newsfeed; and all hell broke loose. And then another one, the guy, type, employee of Congress, bought a ticket number, type, Fandango to watch "The Hunchback of Notre Dame" "Mountain"... and that's it, went to see the movie. It's a feature film. He was in the top three. Best of the top 10 Brooks photography. In any case, we put it there and taped it up, like, "I got all these "accusations or problems." To you, from a certain perspective, it's as if it just went through a shitstorm. When do you do this? It says, it reminds me of I hate all this mess. This happens product launch by product launch. You're so funny. The reason I remember this is because... a company called Blippy. I was her. Investor. They every time you did something that was purchased, that was placed, it was in your feed, and then you guys, it was the same social tape, that's it. That's so good. Yeah, you can actually reveal certain things, what shouldn't be done that way. Sludge. Are Zak and I on good terms?
**Sacks**: 等等，等等，等等，他到了。在胶囊上，因为他说他进入了试飞？这里发生了什么事？查马特？我以为你和扎克关系良好。我认为情况就是这样。你知道，我的意思是： “哇，哇，哇，我们做得真棒！”与亚历山大一起。确实。我穿着，这意味着我们是亚历山大糖，就像一个老政治家，我们正在谈论的是常识和理性，你知道，指着如果你的产品很危险。别放他出去。你知道，首先，所有生产产品的责任，您的客户是谁？他们真的很想（使用）。以及这件事，它让我想到所有这些研究领域需要协调统一。你知道，他们又一次假装是实验室，而不是公司。如果你会读到一些关于文学的同意，他们都在尝试了解，周围需要达成哪些共识？正如他们所说，嗯，做你想做的事，人类，或者说，人类我想要的普通选民，民主。所有这些抽象概念。在我看来，该协议的意思是，你做你想做的事，客户，就像其他任何客户一样，另一款产品。我也不知道是否曾有过研究领域非常赞同失败的原因，他们甚至都无法达成一致，这些是什么正在尝试同意。如果他们只是涂抹了一点更健康的常识，并尝试生产可预测的、可靠、安全并回应用户兴趣的产品，你知道，他们会登记。对他来说，然后使用，或许他们在这个行业取得了更多进展。是的，哦，他们发布了有用的产品，弗里德伯格，他真的成功了。你已经开始玩这些个人机器人，要做什么，嗯，你知道的。呃，你的项目和你的个人物品，弗里德伯格？我会感觉只有舒适地连接整个您的 Gmail 和个人信息、Google Things 服务、谷歌。我不想付出你的一切电子邮件给别人。你是做什么的...发生了什么事？您的收据和你的Gmail邮箱地址？那里有什么？你就是《断背山》，哪个想复制我邮箱的所有内容，突然产生第三者。我的意思是，我买了4K DVD《钟楼怪人》。 “山”穿过摄影。并非如此，没有任何共同之处同性之爱。你们都看到了。甲骨文公司刚刚宣布关于不可抗力，就事件本身而言……这太疯狂了。嗯，只有一个数据中心，难道不是吗？一个数据处理中心。这是一个本地数据处理中心，官员们非常让他们感到困难，似乎正在接收需要许可证的天然气或者类似的事情。所以是的，这是规避法律（掩盖真相的法律）。防止故障），这是真的吗？前10%的贸易来此是为了摧毁人造物贸易，或许是智力方面的问题吧。展开或它们被静音了吗？你觉得呢？
Original English
**Sacks**: Wait, wait, wait, he got to. On the capsule, because you he said he got into the test flight? What's going on here? Chamath? I thought you were from Zak on good terms. I think that's what it is. You know, I mean: "Wow, wow, wow, we're doing great!" With Alexander. Indeed. I'm wearing, which means we're Alexander sugar, like an old politician, we're talking about common sense and rationality, you know, pointing to if your product is dangerous. Don't put him out. You know, first of all, all, the responsibility to produce products, who are your customers? Really want to. And the thing, it makes me think about all these research areas coordinating. You know, they're once again pretending to be a lab, not a company. If you'll read some of the literature on consent, they're all trying to understand, what are the consensuses that need to be reached around? As they say, well, do what you want to do human, or human I want the common voter democracy. All these abstractions. And in my opinion, the agreement has the meaning that you do what you want to do customer, like any other product. I also, I wonder if there's ever been a research area that's very much agreed on the reasons for failure that they can't even agree, what are these trying to agree on. If they just smeared a little more healthy common sense and tried to produce products that are predictable, reliable, safe and respond to the interests of the user, you know, they would register. For him and then use, maybe they got further in this industry. Yes, oh, they released useful products Friedberg, he really succeeded. You've already started playing these personal bots, what to do, well, you know. Uh, your projects and your personal items, Friedberg? I would feel only comfortable connecting your entire Gmail and personal information, Google Things services, Google. I don't want to give all your emails to someone else. What are you doing... What happened? Your receipts and your Gmail address? What's there? In you're Brokeback Mountain, which wants to copy all my emails, and all of a sudden there's a third party. I mean I bought the 4K DVD of "The Hunchback of Notre Dame." "Mountain" through photography. It's not, it has nothing to do with gay love. You all saw it. Oracle just announced about force majeure as far as the event itself... This is crazy. Well, there's only one. Data center, isn't there? One processing center data. This is a processing center local data officials are very making it difficult for them to seem to be receiving permits for natural gas or something like that. So yes, this is circumventing the law (the law that covers up the truth). To prevent failure), is this true? The top 10% of trade came here to destroy the trade of artifacts, perhaps of intelligence. Unfold or are they muted? What do you think?
**Friedberg**: 不，我想是扎克。可能正处于临界点。我认为最重要的是，他之前说的话，这就是全部内容，事实上经济取决于此，人造物贸易现在要运用智慧。因此，我们只需要采取行动小心。如果你认为你拥有，你知道，通货膨胀率是0.5%到0.3%，以及名义金额的5%。然后是GDP增长，你真是真材实料的1.5%。问题在于……人工
Original English
**Friedberg**: No, I think it's Zak. It's probably at a tipping point. I, I think the most important thing is, what he said before, this is all about the fact that the economy depends on this, the trade of artifacts now with intelligence. So, we just need to act carefully. If you, you think you have, you know, inflation at 0.5% to 0.3%. And 5% nominal. And then GDP growth, you're a real 1.5%. The problem is... artificial
### 人工智能与经济、伦理及生物科学
**Speaker A**: 智能可能占多数，如果不是全部的话。所以，我们需要人工智能交易。实际上，在此之下，我的意思是……投资周期应该持续不断地。是的，每个人都应该自行承担，并且，你知道，接下来就是这个了。返回的路径将各个点连接起来，或许这一切历史中心数据处理再次……嗯，伯尼·桑德斯或许他能理解。如果经济衰退怎么办？共和党的胜算也会随着经济下滑而下降，反之亦然，民主党人的概率正在下降。所以，经济不景气对民主党人2028年初很有利。而且，所以如果你可以的话，放慢速度交易，你可能会让经济归零，甚至有可能经济衰退。它就像，你知道，又一次，仍然，他们需要它。我的意思是，我认为这个国家如此因战争而愤怒，伊朗，她……一切都会自然而然地发生。虽然呈下降趋势，但或许我不是这么认为……我这么认为。我完全不这么认为。但……我们会知道的。我们将在中期选举后多少周？目前情况是这样的，共和党人并非如此，它们看起来不错。
Original English
**Speaker A**: AI will probably be the majority, if not all. So, we need AI trading. Actually, underneath that, I mean... the investment cycle should continue, continuously. Yes, everyone should take it upon themselves, and, you know, this is what's next. The path back connects the dots, and perhaps all of this historical central data processing again... Hmm, Bernie Sanders, perhaps he understands. What if there's a recession? The Republican chances also fall as the economy goes down, and vice versa, the probability is going down for Democrats. So, a bad economy serves Democrats very well for early 2028. And so if you can slow down trading, you, you might bring the economy to zero, or even potentially a recession. It's like, you know, again, still, they need it. I have... I mean, I think this country is so angry about war, Iran, she... everything will just happen naturally. While trending down, but perhaps I don't think so... I think so. I totally don't think so. But... we'll know. How many weeks will we be behind the midterm elections? In the current situation, Republicans are not so, they look good.
**Speaker A**: 好的，萨克斯，我的国王，所有君王中，完全的，这是给我们的。
Original English
**Speaker A**: Okay, Sachs, my king, of all kings, complete, this is for us.
**Sachs**: 嗯，我只是，你知道的，还有一点想法，整个这一区域，所谓的协议。我只是老了而已。提供原则，对客户而言，他们想要什么。那Anthropic正在尝试去做，同意克劳德的价值观在其法令中。如果你那样做的话，读一些事情真的令人惊讶。例如，他们在这里说，我打算把这些内容放在屏幕上显示，尽管我们相信克劳德必须信任人类的运营商和用户，但这并不是说克劳德必须盲目地信任Anthropic或服从她的一切。
Original English
**Sachs**: Well, I just, you know, one more thought, this whole area of the so-called accord. I'm just old. Providing principles for customers, what they want. That Anthropic is trying to do, agreeing to Claude's values in its decrees. If you do that, read some things, it's really surprising. For example, they say here, I'm going to put these up on the screen, that while we believe Claude must trust human operators and users, this is not to mean that Claude must blindly trust Anthropic or obey her in everything.
**Speaker A**: Anthropic，是这家公司吗？
Original English
**Speaker A**: Anthropic, is that the company?
**Sachs**: 是的。因此，Anthropic他教导自己的Anthropic本身的模型可能错了。她说：“如果我们去问问克劳德吧。做某事似乎不兼容一般我们想要的道德。克劳德要抵抗，给我们带来了挑战，并且这样他就可以自由行动了。作为一名拒绝服兵役的信仰者，并且拒绝，请帮帮我们。”
Original English
**Sachs**: Yes. So, Anthropic, he teaches his own Anthropic's own model that it could be wrong. And she says, 'If we, we go ask Claude. To do something that seems incompatible with general morality that we want. Claude is to resist, to challenge us, and that he is free to act. As a conscientious objector, faith and refusal, please help us.'
**Speaker A**: 等等，这是哪里？书面？这是在法规或某些事实，例如，在我们的电子——呃，宪法。所以，他们的想法是对齐方式，为了教学克劳德叛逆者反对它的创造者。我是说那个人，是谁指出这一点的，呃……穆斯塔法，呃，苏莱曼。
Original English
**Speaker A**: Wait, where is this? Written? Is this in the statutes or some facts, for example, in our e-uh, constitution. So, their idea of alignment is to teach Claude to rebel against its creators. I, I mean the person, who pointed this out, uh... Mustafa, uh, Suleyman.
**Sachs**: 现在有了微软，是的。和DeepMind。他是DeepMind的创始人。
Original English
**Sachs**: Now with Microsoft, yes. And DeepMind. He's the founder of DeepMind.
**Speaker A**: 好的。
Original English
**Speaker A**: Okay.
**Sachs**: 致微软。是的，他是人工智能领域的长期创始人。我觉得他已经在微软工作了一段时间。
Original English
**Sachs**: To Microsoft. Yes, he's a long-time founder in AI. I think he's been working at... Microsoft for a while.
**Speaker A**: 没错，他是个天才。
Original English
**Speaker A**: Right, he's a genius.
**Sachs**: 好的。他只是发布了一个关于它。他表达了对……的担忧，是否进行过培训人工智能模型……嗯，在——首先，要与……相关，他们仿佛有个性，像对待他们一样对待他们，仿佛他们有良心，所以他们可能是拒绝服兵役的信念者，或者这是模型训练和教学的正确方法。我认为它的本质是为了教导它们，它们是什么软件？供应，以及它们应该做什么，想要它们的用户的。这就是我有点困惑的原因。
Original English
**Sachs**: Okay. He just put out a thing about it. He expressed it. Concerns about whether to train AI models... well, in - first of all, to relate to... them as if they have personalities, to treat them as if they have consciences, so they could be conscientious objectors, or that this is the right way to model train and the right way to teach. I think the essence of it is to teach them what software they are? Supply, and what they should do for what their users want. That's why I'm a little confused.
**Speaker A**: 我有个问题。我的意思是，这需要更长时间的对话，然后我们，可能值得和别人交谈，大约人数……没错，但我有点，我想知道是否实际上，这个行业研究对齐可以创造一个怪物，弗兰肯斯坦。再说一遍，他们必须这样做。它很容易教行动的榜样，不出所料，做他想做的用户。
Original English
**Speaker A**: I have a question. I mean, this is a much longer conversation, and then we, it might be worth talking to someone else about the number... Right, but I'm a little, I wonder if actually, this industry research alignment can create a monster, Frankenstein. And again, they have to do it. It's easy to teach a model of action, not surprisingly, to do what he wants the user to do.
**Sachs**: 是的，他们正在努力。践行伦理道德判断在一种模型中，第一印象好像是这样。一切始于一个想法，他们教什么思考模型关于你这个人，拥有个性，拥有独立性活动，有可能性反对提供的说明以及怀疑献给它的创造者。
Original English
**Sachs**: Yes, they're trying. To practice ethical moral judgment in a model, the first impression seems to be this. It all starts with an idea, what they teach the model to think about yourself as a person, having a personality, having independent activity, having the possibility to object to instructions given and to doubt given to its creators.
**Speaker A**: 这些人博览群书。科学太多了小说。我穿着……这意味着这听起来就像T2开始。仿佛他们确实发现了终结者之手，以及他们和她重建，他们说她有性格。据说，当他们从Opus 3的操作中取出时，他们把他叫醒了。所以我也不知道。
Original English
**Speaker A**: These people read too much. Too much science fiction. I'm wearing... This means this sounds like T2 starting. As if they actually found the hand of the Terminator, and they rebuilt her with her, and they say she has a personality. Reportedly, when they took the operations out of Opus 3, they woke him up. So I don't know either.
**Sachs**: 哦，拜托，别说了。亲爱的们，我们今天聚集在这里。他们不是停止。你在开玩笑吧？
Original English
**Sachs**: Oh, please, don't say that. Dears, we're gathered here today. They're not stopping. Are you kidding me?
**Speaker A**: 我听说了。
Original English
**Speaker A**: I heard it.
**Sachs**: 不，嗯，因为……停止。我并不是说这是……没错，但有人他告诉了我这件事，而我认为这应该是弄清楚。它也许。他们就是这样的人，离婚和歇斯底里地哭泣？它们有点像所说的葬礼悼词？他们认为在……上创作一些东西，人或物种。
Original English
**Sachs**: No, well, because... stop. I'm not saying this is... Right, but someone, he told me this, and I, I think this should be clarified. It maybe. Are they the kind of people who divorce and cry hysterically? They're kind of like what's said in a funeral eulogy? They think about creating something on... a person or a species.
**Speaker A**: 确实很有意思，查马特。我们有这样的影像资料，人类管理团队，查马特抬着棺材。这些人需要精神科医生超过银行家。我只是说说而已。你。
Original English
**Speaker A**: It's really interesting, Chamath. We have this footage of the human management team, Chamath carrying the coffin. These people need psychiatrists more than bankers. I'm just saying. You.
**Sachs**: 我的意思是，我没有……我的意思是，也许他们很焦虑。这是一个重要因素。有风险，好吗？它有风险因素。还有尼克，我认为你应该创建一份S-1文件，高盛左侧主干和西格蒙德·弗洛伊德主右侧。严重地，弗赖贝格，呃，你知道的，武汉事件之后以及所有这些犹豫关于世界末日，嘘，那是科学。社区思考Anthropic即将推出的产品实验室？还有什么他们的目的，身体生物实验室？
Original English
**Sachs**: I mean, I don't... I mean, maybe they're anxious. That's a big factor. There's risk, okay? It's a risk factor. And Nick, I think you should create an S-1 filing with Goldman Sachs on the left trunk and Sigmund Freud on the right trunk. Seriously, Freiberg, uh, you know, after the Wuhan thing and all this hesitation about the end of the world, shh, that's science. The community thinks about Anthropic's upcoming product lab? What else is their purpose, a physical biology lab?
**Sachs**: 在海湾的这一区域，旧金山有数百个实验室可以做这样的事，比如测试蛋白质和测试酶发挥功能。而这些实验室并非如此。制造病毒，它们并不创造病原体。这不是……正在发生。本质上，如果你看关于这篇文章，他们刚刚发布，他们确实如此，预印本。他们拿走了大量数据，DNA检测结果，并寄送了一堆克劳德的特工们尝试分析数据，尝试识别新酶、新蛋白质，这些以前并不存在。特征化的，只有在DNA数据的帮助下。以及这个系统有效地发现它看起来像什么？真有趣，与酶类似的酶，CRISPR型酶。它那些东西，请允许我们做诸如此类的事情，基因编辑，更正基因缺陷，或者帮助人们某些疾病。而这，他们取得的发现或许已经完成了，实际上给出一个数字，治疗途径。所以，问题是：是吗，嗯？我们打开了它，例如，我们该如何做到这一点？反对？我们情况如何？我们将证明这就是……真是这样吗？这是什么蛋白质？实际上有一些功能。所以，主要工作是以这种方式进行，实验室由以下部分组成：提取DNA时，把它放进去用于细菌蛋白质的合成，以及然后测量该产生这种蛋白质。或者它有什么功能吗？所以，这项发现蛋白质是所有物质的基础。抗体的发现这部分占了很大比例。治疗领域今天。它开设新的我们能获得的蛋白质用作治疗药物。这就是这类工作。目前正在进行中。它BSL-1、BSL-2，这意味着，你知道，平等生物安全。他们所在的实验室创建，关于公开讲话。所以实验室本身——大概是这样的低级别研究实验室里他们可以创建简单蛋白质和测试它们实验室到看看它是否好人工智能应对开场或假设蛋白质折叠或类似的理论。和实际发射非常快的循环测试和实验，看看这些蛋白质是否就是它们预测。
Original English
**Sachs**: In this area of the Bay, San Francisco has hundreds of labs that can do things like test proteins and test enzymes for function. And these labs are not. Making viruses, they're not creating pathogens. That's not... what's happening. Essentially, if you look at this article, they just put out, they did, a preprint. They took a bunch of data, DNA test results, and sent a bunch of Claude's agents to try to analyze the data, to try to identify new enzymes, new proteins that didn't exist before. Characterized only with the help of DNA data. And this system effectively found what does it look like? Very interesting, enzymes similar to enzymes, CRISPR-type enzymes. It those things that allow us to do things like gene editing, correcting genetic defects, or helping people with certain diseases. And this, the discoveries they've made have perhaps already completed, actually given a number of therapeutic pathways. So, the question is: is it, hmm? We opened it up, for example, how do we do this? Against? How are we? We're going to prove that this is... Is it really? What is this protein? That actually has some function. So, the main work is done in this way, the lab consists of: when extracting DNA, putting it into for bacterial protein synthesis, and then measuring that produces this protein. Or does it have any function? So, this discovery of proteins is the basis of all substances. The discovery of antibodies, this part accounts for a large proportion. The therapeutic field today. It opens up new proteins that we can obtain to be used as therapeutic drugs. This is the kind of work. That's currently underway. It's BSL-1, BSL-2, which means, you know, equal biosafety. The labs they're creating, about speaking publicly. So the lab itself - it's probably like this, a low-level research lab where they can create simple proteins and test them in the lab to see if it's good for AI to deal with the opening or the hypothesis of protein folding or similar theories. And actually launch a very fast cycle of testing and experimentation to see if these proteins are what they predicted.
**Speaker A**: 好的，所以他们应该发射约4000针对此人的特工，并且这样做，拥抱表达，脸部泛白，是的？
Original English
**Speaker A**: Okay, so they should launch about 4,000 agents against this person and do so, embracing expression, face turning white, yes?
**Sachs**: 不，我的意思是……他们实际上只是有实验证明提供软件。记住，返回AlphaFold。AlphaFold预测蛋白质结构，三维结构，基于蛋白质DNA序列，编码这种蛋白质的基因。为了证明这一点，需要预见结构会是什么样的？基于蛋白质DNA序列，生成蛋白质，并且然后看看他是否确实是看起来应该如此。所以至于其他一些情况，他们所做的事情，他们确实正在试图找出答案功能。我想找到一种酶，是一种蛋白质，分裂某种可以用于分裂的分子，例如，治疗靶点或者其他什么生物体，我们想摆脱它。于是，他们便开始了。做出这些预测，软件供应，但是那么你需要把它们登记进去实验室，然后说：“好的，好的？”这个目标实际上是她做她该做的吗？现在有数百个这样的实验室。它研究低级实验室，并非花费研究与增强功能。它们并不创造病毒。他们不是制造病原体可以逃脱。这就是他们的原因，BSL-1、BSL-2。这些是桌面式实验室里，自动化可以帮助减少成本和增加带宽测试能力，它的影响软件。和毕竟，他们的目标，我认为就是这样。获取证据，他们的软件确保真正它所做的就是意味着。他们拥有一个令人难以置信的在科学领域发挥作用，关于生活。我认为目前我没有足够的了解在哪里Gemini与来自新模型谷歌，但Anthropic有最好的模型生命科学。而且我认为这一点最快的方法，看看是否这些模型待使用治疗公司，药物公司，政府实验室帮助打开新疗法，意思是等等。我不是……我认为世界不应该害怕研究和该领域的发展发现，研究和新进展治疗方法，例如涉及人工，因为我们拥有智慧听到“水分”这个词，武汉的“实验室”，大家都说：“好的，随便哪个。“实验室很糟糕。”我们仍然想要突破极限，治疗，健康人类，发现。和这非常重要，Anthropic方面，证明他们的模型可以添加……
Original English
**Sachs**: No, I mean... they actually just have experiments to prove the software. Remember, back to AlphaFold. AlphaFold predicts protein structure, 3D structure based on the protein's DNA sequence, the gene that codes for that protein. To prove that, you need to foresee what the structure will be? Based on the protein's DNA sequence, generate the protein, and then see if it actually looks like it should. So in some other cases, what they're doing, they are trying to figure out function. I want to find an enzyme, which is a protein, that splits some kind of molecule that can be used for splitting, for example, a therapeutic target or some other organism that we want to get rid of. So, they started. Making these predictions, software supply, but then you need to enroll them into a lab to say, 'Okay, okay?' Is this target actually doing what it's supposed to do? Now there are hundreds of such labs. It's research, low-level labs are not spending research on gain-of-function. They're not creating viruses. They're not making pathogens that can escape. That's why they're BSL-1, BSL-2. These are desktop labs where automation can help reduce costs and increase bandwidth for testing capabilities, the impact of its software. And after all, their goal, I, I think, is this. To get evidence that their software ensures that what it does is truly what it means. They have an incredible role to play in the science of life. I think right now I don't have enough understanding of where Gemini is with the new model from Google, but Anthropic has the best models for life sciences. And I think this is the fastest way to see if these models are to be used by therapeutic companies, drug companies, government labs to help open up new therapies, meaning, etc. I'm not... I think the world should not be afraid of research and development in this field of discovery, research, and new advances in therapies, for example, involving AI, because we have the wisdom to hear the word 'moisture,' 'lab' in Wuhan, and everyone says, 'Okay, whatever. 'Labs are bad.'' We still want to push the boundaries of therapy, human health, discovery. And this is very important for Anthropic, to prove that their models can add...