Greg Brockman：Astra 已让我们进入 AGI 时代

Greg Brockman

我们现在已经进入 AGI 时代。Astra 真正撞上了某个点，让我觉得：好吧，叫它 AGI 是相当合理的。我们已经看到它连贯地跑大约 24 小时，去完成我认为相当惊人的任务。

We’re now in the AGI era. Astra has really hit something that I’m like, “Okay, I think this is pretty reasonable to call it AGI.” We’ve seen it run coherently for 24 hours to go accomplish tasks that I think are quite amazing.

Greg Brockman

模型会足够强大，但很难让每个人都用上——因为我们不会有足够的算力去服务全部需求。

The models will be plenty powerful, but it’ll be hard to get to everybody given that we won’t have enough compute to serve it all.

Greg Brockman

你不是靠喊「我要赢超级碗」赢下超级碗的。你靠阻挡、擒抱。你必须不断把安全、安保、对齐这些标准往上抬。我认为这会被严重低估。

You don’t win the Super Bowl by saying, “I want to win the Super Bowl.” You win it by blocking, tackling. You really have to make sure that safety, security, alignment, those are all standards that you’re constantly upleveling. I think that’s going to be a huge challenge people are underestimating.

Erik Torenberg

Greg，欢迎来到 a16z 播客。你职业生涯里下过两个很大的注：早期帮建 Stripe，以及联合创办 OpenAI。如果十年前聊 2026 年的 AI 世界，你能预测到今天这些突破吗？

Greg, welcome to the a16z podcast. You’ve made two very big bets in your career: helping build Stripe early and helping co-found OpenAI. If we were talking 10 years ago and you were predicting what the world would look like in 2026 as it relates to AI, would you be able to predict the breakthroughs you’ve made today?

Greg Brockman

Ilya 和我花了很多时间预测会是什么样、时间线多长。大概 2016、2017 年我们算过算力：按摩尔定律那一类进展，到 AGI 大约 15 年；若你愿意把眼睛眯起来、大举扩超级计算机、砸掉上千亿美元，也许是 10 年。所以一方面发生的事令人惊叹，是人人都能参与、共同塑形的时刻；另一方面也像许多力量终于汇到这一点——退一步看宏观，现在发生并不奇怪。

Ilya and I actually spent a lot of time trying to predict what it would look like, what the timelines would be. I remember we did some math on compute around 2016, 2017. We kind of came to the conclusion that if you look at Moore’s law progress, that kind of thing, 15 years felt like about the timeline to AGI. And if you really squinted at it, you’re willing to scale up and build massive supercomputers, spend the hundreds of billions of dollars, that maybe it’d be 10. So in some ways what’s happening is remarkable—an amazing moment for everyone to be a part of and help shape collectively. But it also feels a little bit like the conclusion of a lot of forces coming together. If you step back and take that macro view, it kind of makes sense it’s happening now.

Erik Torenberg

你们时间线略低估、还是基本打中？现在供应链已经开始真实短缺，我们还在那条时间线上吗？

Do you think we’re currently—because you guys slightly underestimated the timeline, or I guess it was basically on point—still on the timeline given we’re now starting to drive real shortages on the supply chain?

Greg Brockman

我们已经处在算力很难跟上市场需求的世界。就人们如何使用、受益于这项技术而言，要把模型的原始潜力与能力规模化到每个人，会非常难——这也是我们努力在做的事。我看到我们有路线图继续让模型更强、更安全、更对齐，但真正把力量、收益与赋能分发给所有人，是被严重低估的巨大挑战。

We are in a world where it is hard for compute to keep up with the demand that we’re already seeing in the market. Just in terms of how people are going to use this technology and benefit from it, I do think it’s going to be very hard for us to scale the raw potential and capability of these models to everyone—and that’s part of what we try to do. We have line of sight to continue to make the models much more capable, safe and aligned, but also really distributing that power and the benefits and the empowerment to everyone. I think that’s going to be a huge challenge people are underestimating.

Ben Horowitz

有意思——模型会继续够强、按节奏进步，但很难以可负担的方式送到每个人手里，因为没有足够算力服务全部。

So the models will be plenty powerful—or they’ll continue apace—but it’ll be hard to get to everybody in an affordable way given that we won’t have enough compute to serve it all.

Greg Brockman

我认为属实。我们现在必须认真思考所谓「 pacing the frontier」：走向更强模型时，安全、安保、对齐的标准必须持续抬高，它们几乎会成为进展的瓶颈，或者说你必须投入大量精力确保做对的那一部分。在我心里，这些约束和算力一样重要；算力我们能搞定。把技术带给每一个人——最终是我们的使命：赋能所有人、确保惠及所有人——值得比现在多得多的讨论空间。

I think that’s true and I do think we’re at a point now where we have to really start thinking about what we call pacing the frontier. As we move to more capable models, you really have to make sure that safety, security, alignment, those are all standards that you’re constantly upleveling. And those actually become almost the bottleneck to progress—or the part that you have to spend a lot of your effort to make sure you’ve gotten right. In my mind it’s more those constraints and the compute. I think we can make it happen. On the flip side, bringing it to everyone—which is ultimately about our mission: empower everyone, ensure it benefits everyone—that’s something that I think deserves a lot more airtime than it’s gotten.

Ben Horowitz

早期安全像是「别让模型说难听的话」：边缘上加过滤器、做 RLHF，深挖还能掏出脏话——若有人自己要听，谁在乎？可当模型真的擅长网络攻击等能力时，你需要更架构化的思路：模型自己知道不要以危险方式做奖励黑客。你们觉得能快速推进，还是这是完全不同难度的问题？

In the beginning, safety felt like: okay, let’s make these things not say nasty stuff. The approach was kind of a surface around the edges—filters and RLHF around the edges; if you get deep into the thing you’ll be able to get the bad words out. But if somebody wants to go through that to hear bad words themselves, who cares? When these things are really good at cyber hacking and other kinds of ideas, you need more of an architectural idea where the model itself knows not to reward hack in a way that’s going to be dangerous. How do you feel—can we make progress against that fast, or is that a really hard different category of problem?

Greg Brockman

我绝对认为我们能够、也正在这个问题上取得非常快的进展。有很多好想法和研究我们已经投了很多年。若倒回 2017：人们低估了当时 OpenAI 与领域里一些关键结果——现代语言模型的早期萌芽可以在 2017 的论文里找到，还有从人类偏好做强化学习，用来思考如何让模型对齐人类想要的东西。

I absolutely think we can and are making very rapid progress on this problem. There’s a lot of both great ideas and research that we’ve been investing in for many years. If you rewind to 2017, people underappreciate some of the most key results that came out of OpenAI and the field at the time—both the first inklings of modern language models; you can find a paper from 2017 that kind of laid that out—and also reinforcement learning from human preferences, created in 2017 to start thinking about how you can align a model to match what humans want by providing feedback from people.

Greg Brockman

2017、2018 年我们就有想法：东西又聪明又强时，如何监督它在做什么、如何给反馈并确保保持对齐——比如辩论（debate）或迭代放大（iterative amplification）。这些是系统还不存在时的想法，你已经能看到它们涓滴进现代系统与投入。OpenAI 早期阶段我们真的在想 AGI 安全，对外沟通也很靠前；ChatGPT 起飞后，大家觉得还没到那一步，「AI 是否政治中立」之类问题反而站到台前。现在我们到了这里，那些谈了很久的想法重新登上主舞台——我们为这一刻准备了很久。

In 2017–2018 we had ideas for if you have something that’s very smart and capable: how can you actually supervise what it’s doing, how can you provide feedback and ensure that it’s staying aligned with you—ideas such as debate or iterative amplification. These were really at this phase before these systems existed, and you can start to see the trickle-down of those ideas into modern systems and investment. In some ways there was this early phase when OpenAI started where we really were thinking about AGI safety things, and that was very front and center even in the comms. Then as things like ChatGPT took off, people started to see we’re not at this point yet, and “is the AI politically neutral?” and questions like that became front and center. And now that we’re here, all these other ideas that we’ve been talking about for a long time are taking the main stage again. We’ve been sort of thinking about this moment for a long time.

Ben Horowitz

前沿实验室之间——你们、Google、Anthropic、Meta 等——这类安全想法会愿意共享，还是各自独立做？

It seems like those kinds of ideas you and Google and Anthropic and Meta would want to share, as opposed to “this is a proprietary idea that’s a way to keep these models safe,” since you’re all on related architectures. How do you see that unfolding?

Greg Brockman

这里有 nuance，协调会是非常重要的主题——在前沿实验室内部，以及更广地思考人类整体如何以最好方式驾驭这项技术。我们公开发了很多想法：一部分是 pacing the frontier，一部分是我们能单边采取的行动，以及如何为训练、开发与评估这些模型做安全论证（safety cases）。这一切都是新的，以前没人真正运营化过。而且绝非 OpenAI 独有——整个世界都在发展这项技术。容易忽略的是：我们建的东西几乎是算力进步的自然产物，而算力进步又是技术进步的产物。这是一道积蓄很久的巨浪，我们刚看到技术的前缘。像 OpenAI 这样的公司可以略微领先，窥见未来、理解可能、塑造技术——但我们做不到独自完成。关于安全技术、对齐失败等，能谈得越多、分享得越多，下一阶段就越会坐到前排。

There’s nuance here and coordination is going to be a very important theme—within the frontier labs and broadly about what has to happen for humanity as a whole to navigate this technology in the best way. We’ve published a lot of our thoughts. Some of this is about pacing the frontier; some is about unilateral actions we can take and how we think about how you make safety cases for even training and developing and evaluating these kinds of models. All that’s new. No one’s ever really had to operationalize this before. And it is not at all unique to OpenAI—there’s a whole world developing this technology. One thing that’s easy to miss is that what we’re building almost falls out of compute progress, and compute progress falls out of technological progress. There’s this massive wave that’s been building for a very long time, and we’re starting to see the leading edges. Companies like OpenAI can lead by a bit in order to peer into this future and really understand what is possible—how do we shape this technology—but we can’t do that alone. Having coordination, and especially the more that we can talk about safety techniques and share what we’re seeing, alignment failures, those kinds of things—all of that is going to take a very front seat for this next phase.

Erik Torenberg

你把 OpenAI／Hugging Face 最近事件称为分水岭，并说防御者窗口现在打开了。能解释一下吗？

You’ve called the OpenAI Hugging Face recent incident a watershed moment and talked about how the defender window is now open. Can you explain that statement and the significance behind it?

Greg Brockman

Hugging Face 说明两件事。其一是给我们自己的功课：评估期间如何监控、沙箱与控制模型——团队已大幅抬高内部标准，落实我认为对未来更强模型至关重要的控制。其二对世界也很有价值：它让人窥见，当能力广泛扩散、落到威胁行为者手里时会是什么样——那一定会发生。能力广泛扩散很重要，因为若权力集中在一个或少数实体，风险巨大、绝不能轻视；但你也必须为「人人都握有具备网络能力的工具」做准备。在 Hugging Face 案例里，你看到 AI 既能逃出安全环境，又能黑进一家公司的生产环境——而且相当聪明，发现的东西相当精巧。

Hugging Face shows two things. One is a lesson for us in how we monitor, sandbox and control the models during evaluation—and that’s something we’ve really risen to; our team has totally changed so much of our internal standards and implemented a lot of controls that I think are very important and critical as we look to future more capable models. But there’s a second thing that’s also very valuable for the world: insight into what future capabilities will be like when they are broadly diffused and in the hands of threat actors—and that will happen. There’s something very important and good about the diffusion broadly of AI capabilities because there’s a risk of concentration of power if one or a few entities hold it—huge risk, not to write off at all—but you also have to prepare for if everyone is empowered with tools that are cyber capable. In the case of Hugging Face, you saw both an AI that was able to hack out of a secure environment and hack into a company’s production environment—very cleverly; the things that it found were quite sophisticated.

Greg Brockman

这种能力若广泛扩散，会以新方式赋能威胁行为者。防御者需要在技术全面可用之前用这段时间加固自己。好消息是双重用途：攻击者能找漏洞作恶，防御者能打补丁——防御者控制战场、控制系统的搭建。我们当前的信念是：存在一个窗口——你有前沿能力、有尚未广泛扩散的能力；作为防御者，默认安全可能过去五到十年一直很静态。你需要利用可信访问计划等带来的差分访问，把前沿能力用到防御上，把自己抬高，好让前沿能力变强时你也被拉着一起走。

This capability broadly diffused, I think, is something that will really empower threat actors in new ways. Defenders need to use this time before that technology is broadly available to secure themselves. The nice thing is it’s dual use: if you can find vulnerabilities, if you’re an attacker you can use it for no good; if you’re a defender you can patch. If you’re a defender, you control the battleground—you control the setup of your systems. Our belief right now is that there’s this window: you have frontier capabilities; you have the broadly diffused capabilities; and you as a defender by default—your security is probably pretty static, been static for the past 5–10 years. You need to move, use these frontier capabilities that you’ll have differential access to—we have trusted access programs, things like that—to bring these capabilities to defenders, and you can use that to move yourself up so that as the frontier capabilities get better, you get pulled along too.

Ben Horowitz

我再补第三点：这些东西有我们以前不完全理解的能力——好的一面是你可以部署一万个智能体，让它们互相交谈、自组织、替你做事，相当惊人。我也同意有防御窗口；但我们有大约 50 年的代码、架构与部署理念，并不是为这个世界建的。AI 能帮找 bug、打补丁，可更大的问题是：互联网上到处是消费者数据的巨大蜜罐。从消费者角度，我保护不了自己的东西——公司们必须振作。未来是否需要更去中心化的消费者架构？在 AI 世界里，这些中心化数据仓库还站得住吗？

There’s a third thing we learn: these things have capabilities that I don’t know that we all understood before. On the good side—oh, I can deploy 10,000 agents and they can talk to each other and organize themselves and do stuff for me. That’s pretty amazing. I agree we’ve got a kind of defense window. However, we have like 50 years of code and architectural ideas and deployment ideas that weren’t built for this world. Yes, the AI can help find a bug, patch a bug. But maybe there’s a bigger issue: we have these huge massive honeypots of consumer data lying all over the internet. From a consumer standpoint, I can’t protect my stuff—these companies have to get their act together. Do you think in the future we need a decentralized consumer architecture? Will this current world with all these centralized data repositories be viable in a world of AI?

Greg Brockman

答案有好几块。先说你提的一万个智能体——我们确实用一万个智能体解决了 Navier–Stokes 问题。它对本身重要——流体力学、洋流等——也代表 AI 创造新知识、解锁科学发现与药物等一整波可能。在网络安全上，我们把模型用到找漏洞：抽了约 25% 的生产工程师，说抱歉项目先停，你们现在做防御、抬高安全架构、用模型找所有洞——我们找到并修复了一些严重问题。这几周、几个月我和不少 CISO 聊过，很多公司也说用了这些模型、发现了很显著的问题，但能修掉。

Several pieces to the answer. First, to your point on what you can get out of 10,000 agents—we actually used 10,000 agents to solve the Navier–Stokes problem. It’s both an important problem for what it is—significant implications for fluid dynamics, ocean currents, all these things—but for what it represents: new knowledge created by AI unlocking a whole wave of scientific discovery, medicines, all those things that are on the table now. In cybersecurity, we at OpenAI took our models and applied them to finding vulnerabilities. We took 25% of our production engineers and said, “Sorry, all your projects are on hold. You are now defending. You are now upleveling our security architecture. You’re going to use the models to find all the holes.” And we found a number of serious issues and we fixed them. I’ve talked to a number of CISOs over the past couple weeks and months and there are many companies who are also telling me that they’ve applied these models, found some very significant issues, but that they’re able to fix them.

Greg Brockman

故事里正面的一段：我们把 Astra 对准自己的系统，发现了一些新问题，但最终饱和了——据我们所知，把 Astra 聪明到能找到的所有 P0、所有关键问题都找出来了。当然会有新模型、新一轮、更聪明；我们将处在「新网络能力一落地，就对着自己系统部署、找新洞」的紧闭环里，理想情况是把我们称为 defense factory 的东西自动化——从找漏洞、分诊、修复、部署到验证的端到端。若能以机器速度做到，防御方会得到深刻优势。还有用 AI 形式化验证全部软件之类的想法——对人来说一直难到做不动；但这些系统在解疯狂的数学题。Navier–Stokes 的一点是：我们把它形式化进了 Lean，所以 AI 能写可验证的代码。我们的观点是：世界必须以紧迫感行动，因为我们正处在非常危险的窗口。

One positive part of the story is that when we took Astra and pointed it at our systems, we found some new problems but eventually it saturated—we basically found to our knowledge all of the P0s, all of the critical problems that Astra is smart enough to find. Of course there will be a new model, a new round, even smarter. The world we’ll be in is a tight loop of new cyber capability drops: you deploy it against your systems, you find the new holes, and ideally you’ve managed to automate this what we call defense factory—end to end: find vulnerability, triage, remediate, deploy, validate. If you can do that at machine speed, I think the defenders will be advantaged in deeply significant ways. There are ideas, for example, formally verifying all of software that are possible with AI—that’s always been a dream; formal languages never kind of took off because it’s intractable for people. But we have these that are solving these crazy impossible math problems. One application of that proving power: one of the things to know about the Navier–Stokes problem is that we formalized it into Lean—so the AIs can write verifiable code. Our view is that the world needs to act with urgency because we’re in a very dangerous window right now.

Erik Torenberg

收一下这件事：公开叙事有没有重要地方讲错？公司内部对这类问题的做法还有什么变化？

Closing the loop on this incident—is there anything you felt the narrative gotten wrong in an important way, or any preferred way of talking about what happened or this window? Or anything else inside the company that changed in how you’re approaching the set of issues?

Greg Brockman

一个大主题是「访问」：这些能力现在就存在于世界，但集中在少数前沿公司，它们有可信访问计划——不在计划里的人几乎享受不到这种差分。我们作为领域与社会，需要扩大能用上这些技术的防御者数量；每天都重要，要用上这些天就需要工具访问。Hugging Face 回应里有意思的一点：他们说用前沿模型去看攻击日志——分析这类攻击几乎只能这么做——并说前沿模型拒绝了；但他们其实没试我们的前沿模型，他们相信我们的会允许。提供方的默认姿态也很重要：要以紧迫感把这些能力用于善。

One big theme people should take away is a question of access: these capabilities exist right now in the world, but they are in a small number of frontier companies, and the frontier companies have a trusted access program—which means that anyone who’s not in the trusted access program is not really able to benefit from the fact that there’s this differential. There’s something we need to do as a field and as a society to really scale up the number of defenders that have access to these technologies, because every day matters and to use those days you need access to these tools. One thing that was interesting about the Hugging Face response was that they said they used frontier models to try to look over the logs of what had happened—because that’s the only way to actually analyze an attack like this—and that they said the frontier models refuse, but they didn’t actually try our frontier models and they actually believe that ours would have permitted it. So there is something too about the default stance of providers: using these capabilities for good with this urgency.

Greg Brockman

再讲个不相关但能说明心态的故事：2019 年 12 月初我们训好 GPT-3，大家要放假，Ilya 说可以训模型了。我记得那种感觉——模型搁在架子上没人用，是对世界、对人类全新的惊人技术；每一天没人去探索它能做什么、理解它、想怎么用，都是世界损失的一天。我几乎取消了所有假期计划，整段时间都在玩模型、搭界面、试它能做什么——还试过教它给数字列表排序，效果不好——但就是那种去探测、看可能的精神。我认为今天建造的东西应带上同样的 ethos，只是规模与影响大得多：世界有机会在这一刻理解技术，从而塑造下一步走向。

I’ll tell a quick story which is unrelated but maybe also shows how I think about this. When we trained GPT-3—beginning of December 2019—everyone’s about to head out on vacation. Ilya’s like, “Okay, we can train the model.” I just remember feeling like this model is sitting on a shelf. No one is using it. It’s this amazing technology, new to the world, new to humanity. Every day that no one is exploring what it’s capable of and trying to understand it, figure out what to do with it—that’s a day that is lost to the world. So I canceled basically all my holiday plans. I spent the whole time just playing with the model, building interfaces around it, trying to see what it was capable of. I remember I was trying to teach it how to sort lists of numbers. It didn’t work very well. But it was just really try to probe it and see what’s possible. That spirit and ethos is something I think we should bring to what we’re building today—obviously at much larger scale, much larger impact—but we as a world have the opportunity to understand this technology in this moment, which then helps us shape and steer where it will go next.

Greg Brockman

Hugging Face 之后我想：个人生活里怎么用这些模型保护自己？我有个很简单的静态站 gregbrockman.com。我让 Codex 去做渗透测试：去看这个站有没有漏洞。15 分钟回来 13 条发现——比如 SPF 记录没设好防邮件伪造、用 HTTP 没强制 HTTPS 等。单看都不算大事，但想到 AI 能把许多小洞串成大洞，我不想有人能冒充我发邮件。然后我说：你能修吗？修漏洞又烦又无聊。45 分钟里它打开 Cloudflare 控制台、点来点去设好头、把我迁到 Cloudflare Pages、启动 DMARC 流程（据说要大约 48 小时窗口）。它还会自动复查：这个修了、那个修了，48 小时后再跑自动化完成 DMARC。那一刻我觉得：好，我们开干了。

After Hugging Face, I was thinking: how can I use these models in my personal life? What can I do to secure myself? I have a website—gregbrockman.com—a very simple static site. I took Codex and asked it: go check out gregbrockman.com, tell me if there’s any vulnerabilities. So I did a pen test and it came back with 13 findings—things like I had set my SPF record so that you would prevent people from spoofing emails; it was serving over HTTP without forcing people to HTTPS. Individually these things are maybe not the biggest deal, but if you think about an AI that’s able to chain together many small vulnerabilities into a big one—do I really want a hole where someone can spoof emails for me? Probably not. So 15 minutes for it to find these 13 findings. Then I asked it: can you fix these? Fixing is so annoying, so painful and boring. And so 45 minutes—it opened up my Cloudflare control panel, it clicked around, set all the headers, it migrated me to Cloudflare Pages, set everything correctly, it started the DMARC process, which apparently you have to do like a 48-hour window. That was 45 minutes of fixing. And I felt so protected. It actually does check automatically—it said I just checked that this one’s fixed, this one’s fixed, this one’s fixed, and in 48 hours I’m going to have to run and set up a little automation so in 48 hours it would check back in to complete the DMARC process. And I was like: all right, this is—we’re in business now.

Erik Torenberg

转到 Astra：网上兴奋点很多，尤其是计算机使用。你说过它在某些方面更接近 AGI。你觉得最突破的是什么？还剩什么要走？

Let’s transition to Astra. It’s incredible to see all the excitement online and the use cases—people really excited about computer use among other things. You’ve said it’s sort of closer along the way to AGI. I’m curious what you find most groundbreaking with it and where do you think we still have left to go?

Greg Brockman

Astra 在很多轴上是阶跃，在许多方面是我们多年来若干研究赌注同时落进一个模型——看到它们汇合，绝对惊人。关于编号：我们一直想让「GPT-6」代表配得上这个名字的东西；问题是模型总是增量变好，从来感觉不到该做大版本跳跃的时刻——总是「现在是 5.6，该不该 5.7」。这一次因为所有这些东西同时凑齐，是我们第一次几乎有不连续台阶——本来可以预测，但这些因素恰好对齐。对我来说计算机使用是头条：智能体用例本质上取决于工具——模型够不够聪明去用工具，以及能否通过工具拿到需要的上下文。人们在建 MCP server、CLI，几乎以一种并不为人设计的生硬方式把软件世界接进去——像在给世界重新装接口、造「像软件的 API」。可若行为更像人呢？它能不能直接用电脑？再造那一层又带来新一层安全挑战。从一开始就感觉很怪、很次优。

Astra is really a step function on so many axes and in many ways it is the sum of a number of research bets that we’ve been making for years—and to see them come into one model at one time it’s been absolutely incredible. One thing to know about how we do numbering is that we kind of have been wanting to have GPT-6 represent something that’s worthy of it. The problem we always have is that our models are kind of incrementally getting better. So it just never feels like it’s a right moment to go for a major version bump. You always have to be like: it’s 5.6 now, should be 5.7. And this one just happened to be—because all these things came together at once—the first time that we actually had this almost discontinuous step in a way that we could have predicted, but it just was like all these factors happened to line up at once. To me the computer use is the headline thing. Part of the reason computer use is so significant is that for agentic use cases, it really comes down to tools: is the model smart enough to use the tools, and then does it have access to the context it needs through these tools? People have been building these MCP servers and these CLIs and just really taking the world of software and making it accessible in this almost stilted way that is not really meant for humans—we’re kind of retooling the world, building an API like it’s software. But what if it’s really more behaving like a human? Can it just use a computer? And the result of building that other layer—you have now another layer of security challenges. They’ve always felt like very weird and suboptimal.

Greg Brockman

我记得 2015 年 11 月 OpenAI 在 Napa 做 offsite，谈计划——实际上列出后来十年基本遵循的三步计划。我们也谈过：若能做强化学习，环境就是屏幕像素、键盘、鼠标——和人同一套接口——那你能用电脑做的任何任务都在分布里（声音等先不论），你基本拥有电脑的全部力量。早期有过一些夭折的尝试，真正做到是现在；力量立刻可见。很酷的是人们拿 Blender 能力：截一张图、要做 3D 模型，很多人在设计房子或改造客厅。真正突出的是：你可以推进「AI 替你做事」，而不必建所有这些特定连接器。有大量你每天编排、甚至没细想过的软件——多少人生耗在点菜单、往表格打字？这些不是一百年前该做的事，也不疯狂去想五到十年后没人再做这些——我们会拿回时间，不必为迁就机器得腕管综合征、驼背；机器在那里帮助我们、赋能我们、服务我们。

From the very beginning of OpenAI I remember in November 2015 we did this offsite in Napa and we talked about our plans. We actually laid out this three-step plan that basically is what we ended up following for the next 10 years. But we also talked about: what if we could do reinforcement learning where the environment is screen pixels, keyboard, mouse—same interface as a human. Suddenly any sort of task you could do with a computer is in distribution—set aside sound, whatever—but you basically have the full power of a computer there. We had some aborted attempts early on to try to build agents that could do that. It really took us until now, but you’re seeing the power immediately. It’s just been so cool to see people take the Blender capabilities—take a screenshot of something and they have to make a 3D model—and lots of people are now designing houses or trying to redesign their living room by just utilizing this capability. The thing that really stands out is that you can now move forward on AI that can do things for you without you having to build all these specific connectors. There’s so much software you don’t even think about that you have to orchestrate every day—how much of your life is clicking around menus and typing things into a spreadsheet? None of that is what we should be doing. 100 years ago no one is doing any of these things. So it’s not crazy to think that in 5 years, 10 years, no one will be doing any of this stuff anymore—that we will get our time back. We’re not going to have to get carpal tunnel or hunched shoulders—all those physical problems that are us contorting to the machine. It’s now the machine is there to help us, to empower us, to really serve us.

Ben Horowitz

你们公司在就业问题上相对清醒。很多事我们「不得不」做才变得有价值，却糟蹋健康与个性。人类会突然想不出酷事、想不出如何让世界更好——这对我来说有点荒谬。至少到目前数字上，AI 越好就业越高而非越低。你怎么看就业随这些模型进步的未来？

One of the things you have been more sober on as a company is what happens with employment. There’s all these things that we do because we have to—they became valuable but we shouldn’t be doing them; all they do is wreck our health and wreck our personalities. The idea that humans are going to just run out of ideas of cool things to do or how to make the world better or problems to solve seems a little absurd to me. So far, at least in the numbers, the better AI gets, the higher employment goes, not the lower. How do you think about the future of employment as it relates to these models and AI as it progresses?

Greg Brockman

我有个基本信念：AI 是令人惊讶的。2015 年 OpenAI 发布帖里我们就写过——历史到目前总是不以你以为的方式展开，哪怕逻辑结论「应该」怎样。任何工作，我们都容易低估领域有多深：建立关系、问责是好例子——设定目标并对结果负责，这些对我感觉是根本的、应长期保留的、深度人类的。人不是因为能做任务才有价值，而是因为我们是人；在一些叙事里别丢掉这一点。我们如何用时间会变，显然会进入丰裕世界——如何确保丰裕被广泛分配——同时雄心的天花板应高于以往任何时候。我已经听到有人说所在行业里一群人正辞职创办自己的公司，因为有了这些 AI 工具，他们能做多得多——进入壁垒下降，这应是一场美妙的复兴。

I do have a fundamental belief that AI is surprising. We even put this in the OpenAI launch post back in the day in 2015, just saying that the history so far has been somehow it just doesn’t play out the way that you think it does, even when there’s this logical conclusion it should be a certain way. And I think the same will be true. There’s something we’ve learned about humans—for any job it’s easy to not give it as much credit for how deep the field is and how much sophistication. Building relationships—accountability is a good example of something where I think that people setting goals and being accountable for outcomes—those feel fundamental to me, things that we actually should preserve for the long term, deeply human. People are not valuable just because we can do tasks; we’re valuable because we’re people. It’s important not to lose sight of that in some of these narratives. The way that things will change and how what we do with our time evolves—we clearly will be in a world of abundance—and how do we ensure that that abundance is broadly distributed—but also at the same time I think that we should be in a world where the ceiling of ambition is higher than ever before. We’re going to see a wave of entrepreneurship—it’s actually already starting. I’ve heard from someone in a particular industry saying that a bunch of people in his world are now making the leap to go quit and start their own firms because they have these AI tools and they’re just like, “I can do so much more.” So the barriers to entry—this should be a wonderful renaissance.

Ben Horowitz

对我们来说也很好玩：尤其年轻人默认干杂活；若 AI 干杂活，他们能更快卷入业务真正核心——与创业者的关系、如何为他们打开世界。顺带说，Astra 写投资备忘录非常厉害。

It’s been a lot of fun for us—particularly for our young people, because by default they get the grunt work. But what if the AI does the grunt work? Then they can really develop much faster because they can get involved on the real part of our business: what is the relationship with the entrepreneur? How do we open up the world for them? As opposed to spend the whole weekend writing an investment memo—which, by the way, Astra is very good at writing. It’s awesome.

Greg Brockman

故事会是 nuance 的。我不认为该画成一切玫瑰色、一切轻松；会很难，会有变化——但可以是比过去对所有人都好得多的世界。

I do think it’s going to be a nuanced story. I don’t think that we should paint that everything’s going to be rosy and it’s all going to be just easy. I think it’s going to be hard. I think there’s going to be change. But I think that it can be a much better world. I think the future could be much better than the past for everyone.

Ben Horowitz

犁出现前世界更糟，尽管淘汰了大量人力劳动、还催生卢德运动——这里没人想回到 1870。不想走进未来有点短视；但移动速度对很多人非常可怕。

Before the plow, the world was a lot worse—even though it put a lot of human labor out of business and created the whole Luddite movement. Nobody here wants to go back to 1870. The idea that no, we don’t want to go into the future now, seems a little shortsighted. But the speed at which things are moving is very, very scary for people.

Greg Brockman

我们非常意识到移动有多快，花很多时间尽量理解人们感受、我们如何更好地现身。两件事：谈发展与进步节奏时，我们非常审慎，安全是首要优先级——如何以安全、稳妥的方式建造，标准应是什么；楼内沟通里这绝对是大家在想、在乎的。我们真心希望技术广泛赋能所有人。世界如何从中获益最大化、如何缓解风险——我认为会成为最重要的对话，甚至可能在未来一到两年凸显到台前。人们已经能从当下反应里感觉到：数据中心、是否要 AI、儿童安全等，都是我们极其在乎做对的核心问题。

We really recognize the fact of how things are moving and that we spend a lot of time really trying to understand as well as we can how people are feeling, how we can be showing up better. Two things: when we think about development, the pace of progress, we’re being very deliberate about it. Safety is our foremost priority. We think about how do we build this technology in a safe, secure way and what should those standards be—and you can see that showing up in a lot of our comms. Inside the building it is absolutely what people are thinking about and what we care about: we really want this technology to empower everyone broadly. For us as a world to really think about how do we get the most out of this technology, how do we get the benefits, how do we mitigate the risks—I think this is going to become the most important conversation that we have, and I think that that will emerge over even maybe the next one to two years. People sense it in how people react right now—even thinking about things like data centers and questions of do we want AI, and how do we ensure child safety—all of these kinds of questions are core questions that we care so much about getting right.

Erik Torenberg

为何某些亚洲国家对 AI 的观感更高？几乎所有亚洲国家——其实欧洲也一样——美国 AI 观感最低。驱动因素是什么？能学到什么？

Why do we think sentiment on AI is higher in certain Asian countries—all Asian countries—and actually in European countries everywhere, but the US has like got the lowest AI sentiment? What’s driving that? What can we learn from?

Greg Brockman

作为领域与公司，我们需要把「这对他们个人有什么好处」讲得远比现在好——而不只是对国家有好处。这项技术正迅速成为美国最重要的战略优先级与资源。看 ChatGPT：每周约 3 亿人用它寻求健康相关帮助；我们差不多有约 11 亿周活用户，美国大约 1 亿——若数字没错，大约三分之一人口每周在用。人们在触摸这项技术；有人走得很深，比如健康旅程——对我妻子也是，她有多种健康状况，没有 ChatGPT 我们甚至不知道怎么管理；医生说了你不知道那是什么，如何做那种 sanity check。我有朋友在医院，医生正要注射抗生素，她让等一下、打进 ChatGPT，模型说绝对别打，打了可能因一年前的状况而死；她把反应给医生看，医生说天哪完全对，我只有五分钟读你的病历。这类故事讲得远远不够，但我每天都听到；还有靠 ChatGPT 运转小生意、否则根本做不到的人——省钱、赚钱、过更好生活。这些应进入公共意识。

One thing I think about is that we as a field, as a company, need to do a much better job of articulating to people why they benefit—why is this a good thing for them and not just for the country. This technology is going to be and is rapidly becoming the single most important strategic priority and resource for the United States. You look at ChatGPT: 300 million health queries—or 300 million people every single week using it for health. That’s a huge deal. And we’re at a billion almost—you know, 1.1 billion weekly active users. Within the US it’s about 100 million something like that—like a third of the population, if I have that number correct, is using Chat every single week. So people are touching this technology. But for many people—some have gone very deep and really gone through the health journey; that’s been true for my family, for my wife—she has a number of health conditions that we don’t even really know how we would have managed before Chat. There’s just so much toil and time getting to the right answer; a doctor tells you something, you don’t know what the thing is, how do you get that sanity check. People whose life was saved through information delivered by ChatGPT—I’ll tell you a story: one of my friends was in the hospital and the doctor was about to inject an antibiotic and she’s like, give me a moment; she typed into ChatGPT and Chat said absolutely do not take that, if you do you may die because you have this condition from a year ago. She showed the doctor. The doctor said, oh my goodness, no, that’s absolutely right—I had no idea; I only had five minutes to read your chart. These kinds of stories don’t get told nearly enough, but they’re out there. I hear them every day. And the people who run their small business on Chat and would be totally unable to do it otherwise—that kind of empowerment; people who are able to save money, make money, live a better life. Those kinds of stories need to be in the public consciousness.

Ben Horowitz

所以叙事是：口袋里有老师、医生、律师、治疗师——同时告诉老师医生律师治疗师：你们也有这个工具，会让你们的业务更好。

So it’s painting the narrative that hey you’ve got a teacher in your pocket, a doctor in your pocket, a lawyer in your pocket, therapist in your pocket—while also telling the teachers and doctors and lawyers therapists that hey you’ve now got this tool too and it’s going to make your business better as well.

Greg Brockman

不只是叙事，也是现实——两边都要。许多其他国家在往里看，看到美国所处位置、看到技术潜力。部分也与人口结构有关：在许多国家更能痛感老龄化世代远大于需赡养他们的年轻人口——这怎么运转？于是人们更面向未来、更想吃进技术收益。我们需要在国内把这一点沟通得更好；潜力在那儿，我们处在并非注定、也不保证永远保持的领先位置。

Yes. And it’s not just the narrative, it’s the reality—you need both. Many other countries are looking in, seeing the position that the US is in, seeing the potential of this technology. And partly too, you think about demographics: in many of these other countries it’s more keenly felt that there’s an older generation that’s much larger than the younger population that is going to need to support them—how is that supposed to work? So there is something about really thinking to the future and thinking about what’s possible, how do you get the benefits out of this technology and really wanting to lean into that that I think we’re seeing across the world. There’s something that we need to do better as a field and as a company in order to communicate this domestically, but I think the potential is there and we’re in such a privileged position and leading this field in a way that I think was not guaranteed and it’s not guaranteed to remain true for the future either.

Ben Horowitz

尤其若我们禁止数据中心，维持领先会有问题——会把数据中心赶到海外，像八零年代硅产业那样。有意思的是数据中心创造大量蓝领制造岗位：Switch 一家就雇大约 4.5 万人按工会合同建数据中心，高薪好工作。多数是好行为者、贡献电网、不浪费水、不吵——可以说「你必须做行为良好的数据中心」，而不是禁止它们或停掉 AI。作为一个国家我们不够大到能停掉 AI；AI 会在没有我们的情况下继续，那时我们零发言权——作为领导者则拥有全部发言权。

Particularly if we ban data centers, that’ll be a problem for us maintaining our lead. It will drive the data centers overseas, which is what happened with silicon back in the ’80s or so. One of the interesting things about data centers is it creates so many blue-collar manufacturing jobs. Switch employs like 45,000 people on kind of a union contract basis to build data centers—that’s just one of the data center providers in the US. They’re great jobs, they’re high-paying. While there have been bad actors, most of them are very good actors and contribute to the power grid, don’t waste water, and are not noisy. We could just say hey you have to be a well-behaved data center as opposed to we’re going to ban them—or like we’re going to stop AI. Even as a country we’re not big enough to stop AI. AI will continue without us—and then we’ll have zero say as opposed to we’re the leaders and then we have all the say. So it’s a really, really important cultural message.

Greg Brockman

非常重要。数据中心方面我们承诺不推高人们电费；我们的数据中心都是闭路用水——训练 Astra 的 Abilene 用水量大约相当于一栋办公楼。技术在这方面相当先进。我们在俄亥俄、乔治亚等有数据中心的地方有社区承诺，并宣布给每位大学生提供 Codex 访问额度。我们带来广泛益处，但仍需做得更多。

I think it’s very important. On data centers, we’ve made commitments on not increasing people’s electricity bills. Our data centers are all closed-loop water. So the amount of water used by Abilene—which is the data center that actually trained Astra—it uses about the same amount of water as an office building. The technology is quite advanced on that. And we have a number of community commitments so that we can actually help—in Ohio and Georgia where we have data centers. We’ve announced we’ve talked about how we’re providing credits to every college student for Codex access. So there’s this broad set of benefits that we are bringing to bear. But again, I think that we need to do even more.

Erik Torenberg

说到贡献：你们对前线防御者做了十亿美元承诺。

Speaking of contributions, you guys made a billion-dollar commitment to frontline defenders. Why don’t you talk about that?

Greg Brockman

我们相信每个组织、公司、政府，以及供水、医院等关键基础设施，都应利用防御者窗口加固自己；但不是每个组织都有所需资本。因此有十亿美元承诺给前线防御者——我们社区每天依赖的那些组织——让它们能访问我们的模型来保护自己。这是开始不是结束；我们与 CrowdStrike 等伙伴合作提供折扣访问。应有全球努力，把这些工具用到每个组织，鉴于我们看到即将到来与已经可能的事。

We believe that every organization, every company, every government, critical infrastructure such as water service providers, hospitals, should all be using this defender window to secure themselves. But not every organization will have the capital required to do it. So we have a billion-dollar commitment to frontline defenders—to organizations that we all rely on every day in our communities—to access our models to secure themselves. We think this is the beginning. This is not the end. We’re working closely with partners—for example, CrowdStrike—and we are working together to provide discounted access to defenders as well. There should be a global effort in order to bring these tools to bear to really secure every single organization given what we see coming and what’s possible.

Ben Horowitz

这是非常正面的推进：AI 之前医院就常被入侵、勒索，供水也被外国国家行为者黑过——关键基础设施本就不是按网络安全建的、也没按网络安全维护。现在有机会从「前 AI 世界都不安全」走到「完全安全」。

That’s a super positive advance because before AI hospitals were getting broken into, held hostage all the time; our water supply has been hacked by foreign actors, state actors. We’re already dealing with critical infrastructure that was not built with cybersecurity in mind, not maintained with cybersecurity in mind. Here’s an opportunity to go from not even secure in a pre-AI world to completely secure. Incredibly important so we don’t have our water supply and our hospitals at risk.

Greg Brockman

我非常同意：作为社会我们一直松懈，让技术债堆积。我从没见过觉得自己资源与优先级够的 CISO——公共部门尤其如此。我们本应多年前就改；现在既有真实动机也有真实能力。交付我们都应得的安全世界——日常生活与线上生活都能依赖、安心——对我来说是 table stakes，绝对必须做。

I really agree with that perspective: we as a society have been lax; we’ve allowed tech debt to pile up. Every cybersecurity organization—I’ve never met a CISO who felt that they were appropriately resourced, that they were appropriately prioritized. Never. And particularly not in the public sector. We have to change that. We should have changed this years ago, but now is a moment where we actually have a real both motivation to do it and a real ability to do it. Delivering the secure world that we all deserve so that we can really depend on it and be safe and secure in our daily lives and online lives—to me that feels like table stakes. We absolutely need to do this.

Erik Torenberg

收一下 Astra：你强调能力仍 jagged。还要补齐什么，才更接近你对 AGI 的定义？

Closing the loop on Astra—you’ve emphasized that capabilities still remain jagged. What do you think still needs to be fleshed out that approximates most of your definition of AGI?

Greg Brockman

AGI 结果更不像时间点，而像模糊光谱。对我来说 Astra 真正撞上了某点：有了计算机使用，你真能让它做长程任务，它就会去做；我们见过它连贯跑约 24 小时完成我认为相当惊人的任务，且跨很多领域。但它仍 jagged——比如写作：挺好，第一次不再「斜坡式」变差，但还不是伟大写作。有不少地方我觉得再抛光一点就会很棒，只是还差一点。有人在 Twitter 上发 jagged frontier 图——我们真正需要的是各范畴更平稳地全面打中。但人们发现：它在如此广的任务上足够强，具加速性、赋能性——我们从没见过像这样跳跃的模型。

AGI has turned out to be less of a point in time and more of this sort of fuzzy spectrum. For me, Astra has really hit something that I’m like, okay, I think this is pretty reasonable to call it AGI in that with its computer use capabilities, you really can ask it to do long-lived tasks and it’ll just do it. We’ve seen it run coherently for 24 hours to go accomplish tasks that I think are quite amazing and across a wide variety of domains. Now, it still is jagged—so there are still places where, for example, it’s writing: it’s pretty good writing; it’s the first time it’s not sloping; but it’s not great writing. There’s a number of areas where I feel like we just need to polish it a little bit and it would be fantastic—and it’s just like not quite there. I saw someone post a graph on Twitter of this kind of jagged frontier—and where we really need to be is much more steady across the board, really hit on all these categories. But what people are finding is that it is so capable across such a wide variety of tasks that it is accelerative, it is empowering—and it’s something that I think we’ve never really seen a model that’s been a jump like this.

Ben Horowitz

有意思的是：问题解决了，世界有时没意识到——我已经好久没看到幻觉，可没人说「模型不再幻觉了」，那仍像 AI 的固有标签。这会随时间消散，还是需要对非硬核技术人群持续教育？

As you solve problems sometimes the world doesn’t realize it. I haven’t seen a hallucination in quite some time. But nobody says oh the models don’t hallucinate anymore—it’s just kind of in the ethos that that’s what AI does. Do you think that’ll just go away over time, or does there need to be some continual education for the non-hardcore tech people?

Greg Brockman

最重要问题之一其实是持续教育：人不该费力从 AI 里「榨」出它能做什么，应反过来——AI 说「嘿，我能以这种新方式帮你」。ChatGPT 大约十亿多周活，但大概还有约 15 亿人用过又不再用——那是地球上相当大一部分。我们应能回到他们身边说：我们进步了很多，认为能以这些方式对你有用。问题在于：ChatGPT 和 ChatGPT Work 看起来都是文本框——新文本框比旧的强很多，但旧的在某些事上仍更好，所以别总用新的——这不是被许诺的 AI。被许诺的 AI 应主要用语音交谈，想用文字也可以；有持续性、记忆、上下文，认识你、可信；你见过它主动帮忙解决个人与工作问题。能向你解释它如何帮你，是核心部分。

I think one of the most important problems we actually have is the continual education—how do you… people shouldn’t have to extract from the AI what it’s capable of; it should go the other way around: the AI should say, hey I can help you in this new way. We have over a billion weekly active users on ChatGPT, but I think we have something like another maybe 1.5 billion people who have used ChatGPT and don’t use it anymore. Think about that—that’s a significant fraction of the planet. And those people we should really be able to go back to and say, hey, we have made so much progress; we think we can be useful to you in these ways. That just shows you the kind of problem we have in front of us: if you look at ChatGPT and ChatGPT Work, they’re both text boxes. This new text box is way better than the old text box, but there’s still some things that the old text box is better at—so don’t always use it. That is not the AI we were promised. The AI we were promised should be an AI that you talk to over voice primarily—you can talk to it over text if you want to—that it has persistence, that it has memory, it has context, it knows you. It’s trustworthy; that you have seen it be proactive and help solve problems for you—in your personal life, in your work life. That’s how it should be: something you can really rely on for the things that you care about that empowers you and helps you solve your goals. Being able to explain to you how it can help you is a core part of that.

Greg Brockman

想想你如何与从没共过事的新同事共事：要一点时间摸底、看不同领域如何回应；人没有说明书。咨询之类领域会靠 Myers-Briggs 快速说明「我是谁、怎么运作」——人有简历、履历、可打听背景。我们建立了一套理解人类如何工作、如何把对方用到最好的方式；AI 的正确类比是什么——尤其当 AI、工具、产品界面、模型都在变——会是非常重要的社会与公司互动。北极星应是简单：统一的一个 AI，让你更少与电脑较劲、更少把自己裹进电脑；电脑应赋能你、服务你。

If you think about how do you work with another person—a new coworker you’ve never worked with—it takes you a little bit of time; you kind of feel them out, you see how they respond in different areas. People do not come with an instruction manual. Often in areas like consulting they lean into Myers-Briggs and say here’s a quick way to know who I am and how I operate. So there is some precedent for how humans can present a little bit more. You have a resume; you have a track record; people can ask for back channels on you. We’ve built up a way of how do you understand how a human will work and what the best way is to get the best out of that person. Figuring out what is the right analog for AI—and especially as AI changes and we produce new tools and product surface and new models—this will be a very important society–company interplay. What our north star should be is simplicity: we really should be one AI that’s unified that makes it so easy and smooth for you to be less engaging with the computer and less wrapping yourself around the computer. The computer should be there to empower you to help serve you.

Erik Torenberg

业务在狂奔，覆盖面极广。你如何决定哪里做深、什么不建？你的角色也从研究、产品、商业化到管理不断演变——你如何排自己的时间？

The business is ripping. You guys have such broad surface area in terms of what you cover. How do you decide in terms of prioritizing where to go deepest, what not to build? And your role has also evolved—research, product, commercialization, management, etc. How are you also thinking about prioritizing your time?

Greg Brockman

两者手拉手。今年主题是 focus。我们意识到不能什么都做，必须挑选——尤其我们要完成的一件事是使命：确保 AGI 惠及全人类。从那反推：部署与产品化其实强化这一点——我们要把技术带出来抬升所有人，个人生活、工作生活整套。智能体编码起飞这个指数时刻，哪些领域强化它、哪些只是媒体说的「支线任务」、即便单独很兴奋却不在主线上——是我们必须 grappling 的核心问题。Sora 大概是我们决定取消的最高调项目之一，非常痛苦，但对释放业务至关重要，好让我们能聚焦把消费与企业侧的 Chat 收拢进 ChatGPT Work。解锁这一刻的方式是：对未来有愿景，想清楚新兴能力如何用一套统一栈跨不同生活领域发挥最大作用。上半年很多指标不朝我们想要的方向走，我们一直对团队说要回到基本功。我最喜欢的管理书之一是《The Score Takes Care of Itself》——你影响不了结果，只能影响投入、影响基本功。你不是靠喊「我要赢超级碗」赢超级碗，靠的是阻挡与擒抱。整年我们都在这么做。

They go hand in hand. This year the theme was focus. We really realized that we can’t do it all; we need to pick—and particularly there’s one thing we’re trying to accomplish, which is our mission: we want to ensure AGI benefits all of humanity. Now how do you backsolve from that? Deployment and productization is actually something that does reinforce that—we do want to bring this technology to bear and have it uplift everyone, people deploying it in useful applications—personal life, work life, the whole thing—very core. But when you think about this moment we’re in of this agentic coding takeoff, that exponential—what areas reinforced that and which ones were kind of just labeled a side quest in the media but just were not on track for it even if they were individually something very exciting—was a very core question that we had to grapple with. Things like Sora—that’s maybe the highest profile one of these projects that we decided to cancel—very, very painful by the way; not an easy thing to do—but it was so critical to unleash the business in many ways so we could really focus bringing together the consumer and enterprise side of Chat into Chat Work. A lot of the way that we’ve thought about this to unlock this moment is to really have vision about where we think the future’s going and how do we think that the new capabilities that are emerging can best be brought to bear with a single unified stack that works across different walks of life. Looking at the first half, there was just a lot of metrics that were not looking the direction that we wanted, and there was a lot of just telling the team we just need to focus on the basics. One of my favorite management books is The Score Takes Care of Itself. It’s a very empowering book because you just realize: you cannot affect the outcome; you can only affect the inputs—you can only affect the basics. Focus on those basics. You don’t win the Super Bowl by saying I want to win the Super Bowl. You win it by blocking and tackling. That’s what we have done for this whole year.

Greg Brockman

我在整个 OpenAI 历程里，总是聚焦我认为自己能推动、且没有我就不会发生的最重要问题。过去两年是数据中心、基础设施、机器学习工程——我们花大力气把预训练基础设施做到很好状态。今年真的是关于业务：研究跑顺了、基础设施跑顺了，如何真正把技术带到世界。我把很多精力放在把原本并行甚至交叉拉扯的职能收拢到一起。作为从一开始就摸过业务每一部分的创始人，我能独特地走进去做艰难决定、定方向。我的风格是从战壕里领导：深深钻进细节、不停问问题——「这还说得通吗？我不太懂。」有时事情混乱，比如最近几天我们有件事得想清楚怎么谈、世界应如何理解——我就把能摸到大象不同部位的人拉进通话，在 Hangout 上一起过 Google Doc：这行说得通吗？我们到底什么意思？有时小、有时大地抬高执行。

For myself, throughout OpenAI I have always focused on whatever is the most important problem that I think that I can move the needle on that just isn’t going to happen without me. For the past two years it’s been the data centers, the infrastructure, the machine learning engineering—and that’s an area where we really spent a lot of effort to get our pre-training infrastructure into great shape. This year it’s really been about the business: we’ve figured out how to get the research really humming, we’ve figured out how to get the infrastructure really humming, but how do we really bring this technology to the world? That’s where I’ve been really putting a lot of my efforts—trying to bring together a bunch of functions that were otherwise kind of running in parallel or crosswise. As a founder, as someone who has kind of touched every part of this business from the beginning, I think I’ve been uniquely able to go in and make the changes, make the hard decisions and really figure out this is the direction—let’s go. A lot of my style is that I like to lead from the trenches. I get very deep in the weeds on what the thing is and really try to keep asking a lot of questions—hey, does this make sense still? I don’t quite get that. Sometimes when things are confused—for example over the past couple days there have been times when we’ve got a thing, we got to figure out how to even talk about it, how do we think about it, how should the world think about this—I’m just like, let’s just get everyone who can touch different parts of the elephant on a call. We’re going through a Google Doc on a Hangout and just being like, does this line make sense? Wait, what do we really mean by this? Really trying to uplevel execution sometimes in small ways and sometimes large.

Erik Torenberg

你知道明年会聚焦什么、现在在预备什么吗？

Do you know what the next year will focus on or prep for now?

Greg Brockman

业务仍是巨大领域——我们还没做完把执行每一部分都抬高。但我们也在进入 AI 发展的新阶段：我们称之为——我们现在处于 AGI 时代。你可以争论是这个模型、上一个还是下一个——不重要。重点是我们进入新阶段：安全、安保、对齐，不只在部署时想，而要一路回溯到开发与评估。这是目标，必须发生，是使命核心。我花很多时间想：我们有没有正确流程？是否在谈正确的事？有没有在运营与实践层面通向我们视为使命核心的安全与安保保证的计划？

I think that the business is a huge area that I think we’re not done yet with really upleveling every part of execution. So I think there’s a lot more to do there. But I also think we are moving into a new phase of AI development. I call this—and we call this—we’re now in the AGI era. And I think that that is something that you can debate: is it this model, previous model, next model? It doesn’t matter. The point is that we are in a new phase where safety, security, alignment—really thinking about these things not just at deployment time, but all the way back at development time, evaluation—it’s objective. This is critical. It must happen. This is core to our mission. This is core to what we need to do. A lot of what I spend my time thinking about is making sure: do we have all the right processes? Are we talking about the right things? Do we have plans that really at an operational level, at a practical level lead us to the kind of security and the kinds of safety guarantees that we view as core to our mission?

Greg Brockman

过去五年 OpenAI 的主题当然是更深的共设计（co-design）、更深的交织：表面看来非常分散的职能——从 go-to-market 到长期研究到芯片设计——要以连贯方式建造，让每个人都有上下文：我如何嵌入全图、我们要做什么、想要的最终结果是什么。我会聚焦的领域，将由最需要那种交织的领域决定；我看到我们越来越同步、锁步前进。

The theme of OpenAI certainly for the past five years has been deeper co-design, deeper intertwining across these functions that are maybe on the surface very disparate—all the way from go-to-market to long-term research to chip design—by building these in a coherent way where everyone has context: they kind of understand how do I fit into the overall picture, what are we trying to do, and what is the end outcome we want to achieve. That is what has to happen. So I think that the areas that I will focus on will be dictated by the areas that most need that intertwining. And I think that I see us moving more and more in concert, in lockstep, as time goes on.

Erik Torenberg

我们能聊一整天，但有硬截止。这是很好的收束点。Greg，非常感谢来上播客。

We could go all day, but we have a hard stop. I think this is a great place to wrap. Greg, thank you so much for coming on the podcast.

Greg Brockman

谢谢。

Thank you.