米拉·穆拉蒂：从数学奥林匹克到 OpenAI 首席技术官

我们正在做的事情将改变一切。它将改变我们的工作方式、互动方式、思维方式，以及生活的方方面面。

We're working on something that will change everything. Will change the way that we work, the way that we interact with each other and the way that we think and everything really, all aspects of life.

大家好，欢迎收听《科技背后》。我是主持人凯文·斯科特，微软首席技术官。在这个播客中，我们将深入科技背后，与那些让现代科技世界成为可能的人对话，了解他们创造的动力。和我一起，也许能学到一些计算历史，并获得一些关于当今科技发展的幕后见解。请继续收听。今天我们有位非常令人兴奋的嘉宾——米拉·穆拉蒂。过去几年里，我有幸与米拉及其在 OpenAI 的团队密切合作。尽管有很多机会与她交流，但听到更多关于她的故事——她如何成长，如何最初对数学、物理和科学产生兴趣，以及她从小强烈的好奇心最终引领她走向何方——仍然非常有趣。我们的对话中有一些精彩的片段。我迫不及待想深入探讨，让我们开始吧。米拉·穆拉蒂是 OpenAI 的首席技术官。她曾担任工程师和产品经理，最著名的是参与开发了特斯拉 Model X。她于 2018 年加入 OpenAI，担任应用 AI 与合作伙伴关系副总裁，随后晋升为 CTO。在此期间，她帮助推出了 ChatGPT、DALL-E 和 GPT-4 等 AI 产品，并与我们在微软的团队紧密合作，将他们的技术集成到我们的产品中。今天能邀请你上节目真是太棒了，米拉，非常感谢你的到来。

Hi everyone. Welcome to Behind the Tech. I'm your host, Kevin Scott, Chief Technology Officer for Microsoft. In this podcast, we're going to get behind the tech. We'll talk with some of the people who've made our modern tech world possible and understand what motivated them to create what they did. Join me to maybe learn a little bit about the history of computing, and get a few behind the scenes insights into what's happening today. Stick around. Today we have a super-exciting guest with us, Mira Murati. I've had the pleasure of working very close with Mira and her team at OpenAI for the last several years. Even though I've had all of these opportunities to interact with her, it was so interesting to hear more about her story, like how she grew up, how she first became interested in first mathematics and then physics and science and like where, like this intense curiosity that she had from childhood eventually led her. I think there were just some amazing nuggets in our conversation. Just can't wait to dive right in so let's get at it. Mira Murati is the CTO of OpenAI. She worked as an engineer and product manager most notably helping to develop the Tesla Model X. She joined OpenAI in 2018 as the VP of Applied AI and Partnerships and has since been promoted to CTO. During that time, she's helped bring AI products like ChatGPT, DALL-E and GPT-4 public and has partnered closely with our team at Microsoft to integrate their technology into our products. It is so awesome to have you on the show today, Mira, thank you so much for joining us.

谢谢你，凯文。很高兴来到这里。

Thank you, Kevin. Excited to be here.

今天我会了解到很多关于你的事情，这让我非常兴奋。我很想知道你最初是如何对科学和技术产生兴趣的。

I'm going to learn a lot about you today that I don't know which I'm super stoked about. I would love to understand how you got interested in science and technology in the first place.

一切始于数学。小时候，我自然而然地被数学吸引，总是做习题集，后来还参加了奥赛，我热爱这些，这是一种激情。我在阿尔巴尼亚长大，那是一个欧洲小国，当时正处于从极权共产主义向自由资本主义的转型期。我两岁时，独裁政权垮台，一夜之间陷入无政府状态。但我想人们对共产主义政权的一个误解是，当一切平等时，知识和教育的竞争非常激烈，教育就是一切，这就是我成长的环境。我一直对知识充满渴望，追求知识。但在一个政权不断更迭、一切都不确定的地方，我更倾向于科学中的真理，那些稳定且能追根究底的东西。历史书或其他书籍的来源也值得怀疑，历史一直在变。我想，我对科学和数学的本能自然倾向，可能被成长环境放大了。从很小的时候起，我就对数学和物理非常感兴趣，并一直追求到大学。

It started with math. When I was a kid I just gravitated towards math and I would do problem sets all the time and then we eventually did Olympiads and I loved doing that, it was such a passion. I grew up in Albania, it's a small country in Europe and this was during the transition from totalitarian communism to this liberal capitalism. When I was two, the dictatorship regime fell and it was anarchy overnight. But I think one thing that people misunderstand about communist regimes is that, when everything is equal there is really fierce competition for knowledge and education is everything and so that's the setting that I grew up in. I was just always very hungry for knowledge and the pursuit of knowledge. But in a place where there's this constant regime change and everything is uncertain, I gravitated more towards the truth in science, something that felt steady and you get to the bottom of. Also the sources of history books or other books are questionable, history kept changing. I think maybe just intuitive and natural gravitation towards sciences and math was amplified by the circumstances in which I grew up in. From a very young age I was super interested in math and physics and continued to pursue them until university.

你的父母是数学家或科学家吗？

Were your parents mathematicians or scientists?

不，不是。他们实际上是教文学的，所以我对数学和科学的兴趣是自然而然的。

No, not really. They actually taught literature and so it was just an organic interest towards math and science.

来自西方，有一点——我比你大一点，或者大很多。我成长过程中也对数学、科学和编程很感兴趣，当时让我印象深刻的是，西方自由民主国家与俄罗斯阵营之间存在竞争，知识本身，尤其是科学、数学和技术知识，在两边都受到高度重视，因为这是我们在各种竞赛中竞争的一种方式。我不知道在阿尔巴尼亚是否也有这种感觉。

Coming from the West like one of the things that - I'm a little bit older than, or a lot older than you I think. One of the things that struck me growing up where I also was interested in math and science and programming fairly early on, was that there was this competitive nature between the liberal democracies of the West and some of the Russian coalition that knowledge itself, like particularly science and mathematics and technical knowledge, were like one of these things that were highly valued both here and there at the time, because it was a way to like just compete in whatever contest it was that we were playing. I don't know whether it felt like that in Albania or not.

是的，非常像。我就是喜欢参加各种奥赛，无论是化学、生物还是数学，小时候你不会想那么多，只是一种激情。但回想起来，我能看到环境因素，也要记住当时没有很多工具或娱乐，所以一部分也是出于无聊。实际上，我认为无聊是一个非常强大的动力，促使你去探索并追求任何领域的前沿。在我童年早期，阿尔巴尼亚非常孤立，就像今天的朝鲜。除了书籍，几乎没有娱乐或其他东西流入。书籍就是整个宇宙，那时我就在书中搜索一切。现在我们手头有这么多强大的工具，几乎可以做任何事情。

Yeah, very much like that. I just love doing all these Olympiads whether it was chemistry or biology or math and when you're a kid you don't really think about that. It was just a passion. But looking back, I can see the circumstances and also just keep in mind that there wasn't access to a lot of tools or entertainment and so a little bit was just out of boredom as well. Boredom actually I think is a very powerful motivator to go explore and really pursue frontiers of anything. In the first years of my childhood, Albania was incredibly isolated like North Korea is today. There wasn't much inflow of entertainment or anything really besides books. Books were this entire universe and back then I'd just search everything in books. Now we've got all these powerful tools at our fingertips and can do anything really.

你刚才说的，无聊是一件非常有用的东西，我完全同意。我觉得很有趣的是，我们社会似乎已经认定无聊是坏事，是需要最小化的东西。这也是我和自己孩子斗争的一点。我有两个孩子，一个 12 岁，一个 14 岁，他们没有我小时候那种无聊的能力。我不是在阿尔巴尼亚长大的，我知道这样比较可能不公平，但我在弗吉尼亚中部的农村长大。我们只有三个电视频道，我经常感到无聊，大部分时间都在读书，这非常有用，我很快就能专注于实质性的东西。

What you just said, that boredom is a very useful thing, I could not more strongly agree with. I think it's really interesting that we seem to as a society have decided that boredom is bad and it is a thing to minimize. It's one of the things that I struggle with my own children. I've got a 12 and a 14-year-old and they don't have the same capacity to be bored as I did when I was a child. I didn't grow up in Albania. I'm sure it's probably unfair to even make this comparison but I grew up in rural Central Virginia. We had three television channels and I was bored a lot and most of my life was to get in books and it was a very useful thing to, I got focused very quickly on things that were substantive.

正是如此。锻炼那种专注于某件事、反思信息或进一步提炼信息的能力，很多数学就是这样。你需要长时间面对一个问题，这锻炼了那种肌肉和信念：如果你坚持，你就会有所发现。

Yes exactly that. Exercising that ability to stay focused on something and reflect on information or distilling this information further, and a lot of math is like that. You just need to sit with a problem forever and it exercises that muscle and faith that if you sit with it, you'll discover something.

确实。我不知道你是否也有过这样的经历，我以前遇到过一些很难的数学问题，我如此着迷，以至于会梦到它们，有时甚至醒来后想，哦，终于，我梦到了这个定理的证明。

For sure. I don't know about you but I've even had hard math problems that I worked on in the past where I was so obsessed with them that I would dream about them and I sometimes would even wake up and I'm like, oh, finally, like I got the proof for this theorem that I just dreamt.

没错。

Exactly.

我很想听听，你那种兴趣——听起来像是天生的，或者只是从成长环境中自然而来的——是如何被培养的？有些东西很难，你有导师或老师吗？学校好吗？也许我应该换个方式问。当你想做有意义的事情时，事情会变得足够困难，以至于你会卡住。

I'm interested to hear how that interest that you had, which sounds like it was innate, or just sort of in the air and culture just from the circumstances of how you grew up. But how did it get nurtured? Some of this stuff is hard, and so did you have mentors or teachers? Were the schools good? Maybe I should ask a different way. At some point whenever you are trying to do something substantive, things get hard enough where you get stuck.

你是如何让自己摆脱困境的？

How did you get yourself unstuck?

对我来说，成长过程中我的老师们非常支持我，这并不常见，因为我觉得今天可能很难有这样的条件。但那时，也许他们在我身上看到了什么，真心想帮助我追求自己的兴趣。课堂上我经常做完全不同的习题集，因为我对常规课程感到无聊。我仍然和大家坐在一起，但他们非常支持我做完全不同的事情。我也很幸运，我姐姐比我大一岁半。当我感到无聊时，我会去看她的书。然后我做完她的书，又会找其他书，老师们对此也很有帮助。我认为那可能是最有帮助的事情。我总觉得还有别的东西，还有更多值得追求和学习的东西。16 岁时，我有幸获得奖学金去加拿大温哥华留学，在那里完成了高中最后两年。这是一个走出阿尔巴尼亚、在国际学校与来自不同国家的人一起学习的好机会。那对我来说是一个巨大的机遇。

For me, my teachers, when I was growing up, they were extremely supportive and it was unusual circumstances because I think today maybe less of that would be available. But back then, I don't know. Maybe they saw something in me and they really wanted to help me pursue my interests and often in class I'd do completely different problem sets because I was bored with the usual curriculum. I would still sit there with everyone, but they were very supportive of me doing something entirely different. I was also lucky that my sister is a year-and-a-half older than me. When I'd get bored with myself, I would go and look into her books. Then when I'd do her books and then I would find other books and my teachers were very helpful with that. I think that was probably the most helpful thing. Like I always knew there was something else. There was more to pursue, there was more to learn and then when I was 16, I was fortunate to get a scholarship to study abroad in Vancouver, Canada, where I did my last two years of high school. That was a big opportunity to get outside of Albania and study in an international school with people from many different countries. That was a great opportunity for me.

计算机是什么时候进入你的生活的？

Where did computers enter the picture for you?

可以说相当晚。可能是在我十几岁在阿尔巴尼亚的时候，那时互联网很慢，但我已经通过数学和解决问题思考了很多关于智能的问题，就像世界如何运作的场景，试图用数学甚至物理来解释很多事情。但我一直对大脑如何工作以及智能在理论和抽象层面感兴趣。不过，我追求的艺术更多是试图应用我的知识和技术来解决真正困难的问题，从而以某种方式改善我们的生活。上大学时，我学习工程，因为我认为这是将知识应用于解决世界实际问题的最佳途径。学习工程时，我对将可持续交通带到世界以及可持续能源非常感兴趣。我的毕业设计实际上是建造一辆混合动力赛车。这很有趣，但我们也想做一件感觉很难的事情，所以我们没有用电池，而是用了超级电容器，试图推动可能性，显然这不是可以量产的东西，但这是在推动科学，看看什么是可能的。之后我就去了特斯拉工作，我对可持续能源充满热情，希望为世界带来可持续交通尽一份力。大约 10 年前在特斯拉的那段时间非常激动人心。

It was quite late, I would say. Maybe when I was a teenager in Albania and Internet was slow, but I already thought about intelligence a lot more through math and solving problems and just like, the scenario of how the world works and trying to explain a lot of things through math or physics even. But I was always interested in how the brain works and intelligence more theoretically and at abstract levels. But I would say that the art of what I pursued was more in the theme of trying to apply my knowledge and trying to apply technology to really hard problems that in some way makes our lives better. When I was in college, I was studying engineering because I thought this was the best way to apply my knowledge to actually solving real problems in the world. When I was studying engineering, I was very interested in pursuing ways to bring sustainable transport to the world and also just sustainable energy in general. My senior project actually was building this hybrid racecar. It was fun, but also we wanted to do something that felt really hard and so instead of batteries, we used supercapacitors and really trying to push what was possible, and obviously that was not something that you could build in production, but it was pushing science and seeing what's possible. That's why thereafter I went to work at Tesla and I was really passionate about sustainable energy and doing my part in bringing sustainable transport to the world. That was a very exciting time about 10 years ago at Tesla.

太棒了。你学的是哪种工程？你是电气工程师、机械工程师，还是其他？

That's awesome. What type of engineering did you study? Were you an electrical engineer, mechanical engineer, something different?

我学的是机械工程。有很多动手实践的内容；软件，但也有动手操作。

I studied mechanical engineering. A lot of hands-on stuff; software, but also hands-on.

你最喜欢机械工程的哪一点？因为你现在做的事情非常不同，机械工程和领导软件工程团队很不一样。有趣的是，我的整个职业生涯都建立在软件工程上，但我的业余时间大部分都在做机械工程和机械设计。最初是什么吸引了你，除了可持续能源这个杠杆之外？它和你现在做的事情有什么不同？

What was your favorite thing about, because you're doing something very different now, like mechanical engineering is quite a bit different than running a software engineering team, and like, I love mechanical engineering. It's funny enough, like I built my entire career on software engineering, but most of what I do in my free time is mechanical engineering and mechanical design. What attracted you to that in the first place other than the sustainable, that it was a lever on doing something in sustainable energy, and how how was that different than what you do now?

我想那时我可能把它看作一种更切实的改变事物的方式，它不抽象。感觉很实在。你做出改变，就能看到它，看到它如何影响现实。我一直是个思考者，会探索不同的事物。这很难。机械工程很难，但也很有成就感，而且总是有软件部分，比如在混合动力汽车中，你面对的是整个系统。不仅仅是机械工程部分，还有软件部分、电气工程部分。它包罗万象，我一直对复杂系统着迷。在特斯拉时，我对自动驾驶及其前景越来越感兴趣，也对我们可以用 AI 和计算机视觉彻底改变出行方式感兴趣。这让我对 AI 以及它能在世界上做什么、能带来什么改变越来越感兴趣。我并不一定想成为汽车人。我一直对不同事物充满好奇，我很好奇 AI 将如何影响我们与机器的交互方式，以及我们与信息的交互方式。当时，我对空间计算非常感兴趣，以及用一种完全不同于我们今天用键盘和鼠标（这太局限了）的方式与信息和复杂概念交互。我认为 AI 和计算机视觉将帮助我们真正改变这种与信息交互的界面。我设想虚拟现实或增强现实，你可以几乎触摸到分子，或者感受到混沌理论或引力波，这比在纸上阅读要直观得多。几乎就像抓住一个球就能感受到抛体运动一样直观，即使你不知道物理定律。我认为这可以真正改变我们学习和吸收世界的方式。

I think back then I probably saw it as a more tangible way to change things and it didn't feel abstract. It felt very tangible. You make a change and you see it and you see how it affects reality. I was always a thinker, I would explore different things. It was hard. Mechanical engineering is hard, but it's also very fulfilling and there was always a software component, so like in a hybrid car, you've got the entire system. It's not just the mechanical engineering part, there's always the software component, the electrical engineering component. It's a little bit of everything and I always was attracted to complex systems. When I was at Tesla, I got more interested in autopilot and the promise of it and also what we could do with AI and computer vision to completely change the way that we travel. That got me more and more interested in AI and what it could do in the world, what changes it could bring. I didn't necessarily want to become a car person. I always had this curiosity for different things and I was very curious about how AI would affect the way that we interact with machines and how we interact with information in general. At the time, I got really interested in spatial computing and just interacting with information and complex concepts in a completely different way than we interact even today, really, with a keyboard and the mouse, which is just so limited. I thought that AI and computer vision would help us really change this interface of interacting with information. I imagined virtual reality or augmented reality where you can almost touch molecules or you can get a sense for Chaos Theory or gravitational waves, and that is such an intuitive understanding of complex concepts versus when you read it on a page. It's almost like it's as intuitive as grabbing a ball and getting a sense of projectile motion even if you don't know the laws of physics. I thought, this can really change the way that we learn and the way we absorb the world.

这对我来说非常真实。我认为我们现在生活的现代世界中，我真正欣赏的一点是，有像 YouTube 这样的东西，如果你想理解某个东西，有那么多人在用各种不同的方式解释它，只要你足够坚定，就能找到一个人用完全适合你大脑的方式解释，让你快速理解。这一直是我的挣扎。我可以学得很快，但我不认为我和别人学习的方式完全一样。如果我能找到正确的概念切入点，我就能掌握它，甚至能理解那些在找到切入点之前觉得太复杂的东西。这实际上也是我对你们在 OpenAI 用这些智能体所做的事情感到兴奋的原因之一，因为智能体，如果你试图让它解释某事，它有无穷的耐心和适应性。如果你愿意对话并告诉它你需要什么，它会以你需要的方式解释给你听。这对我来说非常强大。

That feels so true to me. I think one of the things that I really appreciate about the modern world that we live in right now is that you have things like YouTube, where if you are trying to understand a thing, there are so many people trying to explain that thing in so many different ways that if you are determined enough, you can find someone explaining the thing in exactly the right way for your particular brain to understand it quickly. That was always my struggle. I could learn very quickly, but I don't think I learn exactly the same way that other people learn. If I can get the right conceptual hook on something, then I've got it and I can even understand the things that before I got the hook were too complicated. It's one of the things actually that really excites me about what it is that you-all are doing in OpenAI with these agents because the agents, if you are trying to get it to explain something to you, it's infinitely patient and it's adaptable. It will explain things to you in the way that you need it to explain things to you if you're willing to have a conversation and tell it what it is that you need. That feels very powerful to me.

我完全同意。

I completely agree.

在我们进入激动人心的 AI 话题之前——我相信这也是大家想听的——我想先聊聊特斯拉。你在那里工作感觉如何？你后来担任 Model X 的产品主管，责任重大，Model X 是有史以来最令人惊叹、最具创新性的汽车之一。你并不认为自己是个汽车迷，却帮助制造了可能是过去四五十年来世界上最具颠覆性的汽车之一。跟我们说说那段经历吧。

Let's go back for a minute before we get on to all of the exciting AI stuff, which I'm sure is what everyone wants to hear us talk about. I want to hear a little bit about Tesla. What was it like working there? You had a pretty big responsibility there at the end where you were the head product manager for the Model X, which is one of the most amazing, innovative vehicles that anyone's ever created. For you not thinking of yourself as a car person, you helped make one of the most disruptive cars that the world has seen maybe in the past 40, 50 years. Tell us a little bit about that.

特斯拉是一个不可思议的地方，在某些方面，我觉得它和现在的 OpenAI 很相似。当然，特斯拉规模大得多，做的事情也完全不同，但那里聚集了大量才华横溢、聪明绝顶的人，他们对自己的工作充满热情，几乎像是一种精神追求。每个人都坚信自己所做的事情是最重要的。当你处理真正困难的问题时，这种信念非常强大。对特斯拉来说，它是在变革整个行业，同时也在创造和改变许多新行业。这极其困难，但也令人振奋、充满乐趣，我在短时间内学到了很多。我认为在短短三四年内从零到一造出一辆车是不寻常的，时间非常短。这类事情通常有很长的生命周期或时间线，涉及设计、原型、生产等。我在特斯拉学到的一件事是，总有另一种方法，即使看起来不可能，也总有另一种方法。一般来说，产品构建有两种方式：一种是追求极致精致，另一种是快速迭代、从用户那里获取大量反馈、以客户为中心快速改进。特斯拉我认为介于两者之间，两者兼而有之。这太棒了，在一个如此成熟的行业中第一次这样运作。我学到了很多，比如创造力和原创思维的力量，真正改变一切，质疑你所知道的，质疑为什么事情要以某种方式做。正是在那里，我开始对 AI 的力量以及它将如何改变我们所做的一切产生了浓厚兴趣。从某种意义上说，在我的职业生涯中，特斯拉真正激发了我从事 AI 工作的兴趣。当然，在从事 VR 和 AR 工作之后，我认为智能才是世界变革的根本属性。然后我越来越关注 AI 的应用层面，但真正理解通用智能意味着什么，以及我们如何构建它，如果构建了它，如何让世界变得更好。

Tesla was an incredible place and in some ways actually, I find it quite similar to OpenAI now. Obviously it was much bigger and working on something very different, but this high density of very talented, smart people that are just so passionate about what they're doing. It's almost like a spiritual pursuit. Everyone believed so hard in what they were doing and that being the most important thing. That is just so powerful when you're working on really hard problems. In the case of Tesla, it's transforming an entire industry versus creating many new ones as well as transforming them. It was incredibly hard, but also just invigorating and so fun and I learned so much in a short amount of time. I don't think it's normal to build a car from zero to one in just three, four years. It's a very short time. These things usually have this very long lifecycle or timelines in terms of design and prototyping and production and so on. One of the things that I learned at Tesla was there's always some different way, even if it seems impossible, there is always a different way. In products in general, there's these two ways of building products where you have the really, really polished stuff. Then this way of hacking and iterating and getting a lot of feedback from your user base and customer centricity iterating quickly on that. Tesla, I would say, was in-between, doing both. That was incredible, just the first time of operating like that in an industry that is so established. I learned a lot as perceived from just the power of being creative and thinking originally. Just really changing everything and questioning what you know, and questioning why things are done a certain way. That was a place where I started getting really interested into the power of AI and how it would change everything that we do. In a sense, in my career, it was the place that really catalyzed my interest in working in AI. Then of course, after working in VR and AR, I just thought, intelligence is really the fundamental property of how the world is going to change. Then I got more and more interested on just the application side of it. But really understanding what general intelligence meant and how we could build it and how we make things go well for the world if we do build it.

在我们转向 AI 之前，你能分享一个在 Model X 上学到的有趣技术问题或技术点吗？一些棘手、有趣或不同寻常的东西。

Before we move on to AI, what's if you can share an interesting technical problem or technical thing that you learned on the Model X, something that was tricky or interesting or different?

我可以讲很多，比如鹰翼门。那可能有点麻烦。

So many things I could talk about the Falcon doors. That could be problematic.

也许我们可以从高层次聊聊。那是一个有趣的设计选择。显然是一个全新的东西，作为工程师，我不知道具体实现细节，但我能想象让这个功能在技术上工作有多难。你们有没有意识到，车里肯定有几十个这样的东西：设计师有个想法，工程师必须决定或想办法让它工作。一般来说，你们如何平衡这两者？

Maybe at a high level, we can talk about that. That is an interesting design choice to make. Obviously a brand new thing and as an engineer, I don't know the details of the implementation, but I can imagine how difficult it was to make that feature of the car work, technically. Did you all have a sense for, I'm sure there're just dozens of these things in a car where like some designer has this idea that I want to do this thing, then some engineer has to go decide or figure out how to make the thing work. Just in general, how do you balance those two things?

Model X 有很多东西都感觉是在突破极限，以前从未做过，尤其是在那种车上。车门就是这样的功能，还有 HVAC 系统、HEPA 过滤器。这总是需要把整个团队或相关部分聚集在一起：设计、工程、制造、软件团队，或者如果相关的话还有电气工程师，真正把所有的部分整合起来。你可以一起设计，而不是交接后反复来回，或者设计出无法制造的东西。与不同背景、不同领域专长的团队合作，共同设计前所未有的东西，采纳新想法，同时迅速放弃旧想法并转向下一个，这非常强大。关键是在正确的时间解决正确的问题。

There are a lot of things about the Model X that felt just really pushing the envelope and just they had never been done before, or especially in that kind of car. The doors were a feature like that, or the HVAC system, the HEPA filter. It always required bringing together the whole team or the parts that would be working together. Design, engineering, manufacturing, the software side of a team, or maybe if it was relevant the electrical engineers and really bringing together all the pieces. You could design it together versus hand it off and then go back and forth or design something that could not be manufactured. That was very powerful in working with teams that have different backgrounds, domain expertise, figuring out how to design something that has never been done before, adopting new ideas, but also very quickly killing old ideas and moving on to the next one. Just figuring out the right problem to work on at the right time.

我认为这非常重要。这种你做完工作然后扔给下一个人或团队的做法，虽然有一定的效率，但如果你想创造全新的东西，这种瀑布式流程很难奏效。有很多笑话，比如作为机械工程师，我想问你，你有没有在机加工车间待过？因为机械工程师和机加工师傅之间常有矛盾，比如‘你给我这张图纸，根本没法加工’。或者在软件工程中，产品经理和工程师之间也有矛盾，产品经理说‘我们要做这个’，工程师说‘你疯了吗？’通常当所有人都参与讨论时效果更好。听到你说你们就是这样工作的，非常有趣。

I think that is an incredibly important thing. This idea of you do your work and then throw it over the wall to the next person or a team and the change that has to go do the next thing is that there's a certain efficiency that you can get from doing things that way. But if you're trying to make something brand new, it's very difficult to have these waterfall processes like that. There's so many jokes about, like one of the things that I was going to ask you about as a mechanical engineer is, hey, did you spend any time in the machine shop? Because there's this tension between mechanical engineers and machinists, like, 'you gave me this print and there's no way to make it.' Or, there's the tension in software engineering between the product managers and the engineers, the product manager says 'we're going to go do this thing' and the engineers are like 'are you crazy?' It usually works better when everybody is in the conversation. It's super interesting, to hear you say that's how you all did your work.

完全同意。你提到这个很有趣，因为作为机械工程师，我经常自己加工零件，只是为了理解约束限制和加工挑战。这和特斯拉很像，设计工程师经常在车间里装配、测试零件，与制造工程师紧密合作。我认为，正如你所说，对于超过一定规模的公司来说，这是大规模创新的关键。如果你只是把工作扔过墙，官僚主义和流程就会介入，创新就会变得困难。随着公司成长，它们可能会失去愿景，不再追求新想法。但如果你能打破这些，最小化流程和必须跳过的障碍，那么事情就容易得多。回顾起来，这是我在特斯拉学到的一个非常关键的东西。

Totally. It's funny that you mentioned it because as a mechanical engineer, I was often machining my own parts just to understand the constraint limitations and also just the challenges of doing it. It was very similar to Tesla where the design engineers were often on the floor fitting, testing the parts, and just working very closely with manufacturing engineers. I think that like you said, it's key to innovating at scale past a certain size of company. It's difficult to innovate if you're just throwing things over the wall and bureaucracy can kick in, or processes. As they grow, companies can lose their vision and stop pursuing new ideas. But if you cut through that and minimize the layers of processes and things or hoops that you have to jump through to get something done or bring some new idea, then I think it's much easier. So that was something actually quite critical looking back that I learned working at Tesla.

很久以前我听埃隆的一个采访，他描述了一件事，不是 Model X，而是另一款车，他们在制造某个部件时遇到了巨大困难。一旦他开始问正确的问题，结果发现问题不在于如何让这个东西变得可制造，而是这个东西为什么存在？它完全多余，设计上就有问题，真正的解决办法不是去解决那个棘手的难题，因为那个东西本身有点随意，改变初始条件，问题就更容易解决了。我觉得这是我很钦佩埃隆的一点——第一性原理。他总是能退后一步，问正确的问题：我们为什么用这种方式做这件事？什么是必要的，什么不是？

I was listening a long while ago to an interview that Elon was doing where he was describing this thing that was happening, not with the Model X, but another one of the automobiles where they were having a really challenging time getting something manufactured. As soon as he started asking the right questions, it turned out that the problem wasn't solving the problem of how to make this particular thing actually manufacturable. It was like, why did this thing exist at all? Like it was just completely unnecessary in such a way that they got designed and the real fix wasn't like go solve the nasty hard problem. Because the thing itself was a little bit arbitrary and it's like change the initial conditions and then the problem gets easier to solve. I think that is one of the things I admire a lot about Elon is like this first principles. They always like being able to step back and ask the right questions about why are we doing a thing the way that we're doing it? What is necessary and what is not?

我认为这极其重要——退后一步，我的意思是，既要有能力在需要时沉浸于细节、深入挖掘，也要能退后一步问正确的问题，团队要有高度的适应性和对模糊性的容忍。因为特别是当人们非常有经验时，他们会有固定的做事方式，所以你需要适应性强，同时要能相信和怀疑事物。这些是很难同时具备的品质和特质。

I mean, I think this is incredibly important - stepping back, I mean, having the ability to be immersed in details and dig deep when you need to, but also stepping back and asking the right questions and having this high degree of adaptability in the team and tolerance for ambiguity. Because especially when people are extremely experienced, they have a certain way of doing things, and so you need to be adaptable and also believe and disbelieve things at the same time. Those are hard qualities and traits to sit together.

还有关于大组织的问题，组织只有在为利益相关者解决问题的性质要求你变大时才应该变大。因为庞大几乎是一种熵，它迫使一些事情发生，仅仅因为整体的复杂性，没有人能掌握所有细节，于是你会发现自己狂热地、拼命地优化某个细节。如果你能完全退后一步，你会发现你拼命做的东西完全没必要。OpenAI 目前规模的一个好处是，在制度上和复杂性上，你们还有较少那种大组织会出现的奇怪熵。我发现你必须与之对抗，这非常难。因为如果你不反抗，你就是在让人们完全优化狭隘的东西，这基本上会扩散成混乱，人们优化错误的东西。

Then there's just something about big organizations, like organizations should only be big if the nature of the problem that they're solving for their stakeholders requires you to be big. Because bigness, it is almost a flavor of entropy that forces some stuff to happen where just because of the complexity of the whole, like no one has all of the details in their head and so we'd like, you can find yourself trapped in you know just feverishly, working as hard as you can on the details of something. If you could pull all the way back, you would just find that the thing that you're working so hard on is completely unnecessary. Since one of the great things about the size at OpenAI is at right now is you still institutionally and the complexity of things you can - You have less of that weird entropy that happens to big organizations. The thing that I've found is you just have to fight against it. It's super hard. Because if you're not pushing back against this thing, you're just letting people entirely optimize for the narrow thing, it just metastasizes into confusion basically, and people optimizing for the wrong thing.

于是惯性就继续了。

So then momentum just carries on.

是的。我们来谈谈 AI。先说说，你是怎么从特斯拉转到 OpenAI 的？因为你加入得很早，从一开始。一开始并不明显——一点也不明显。

Yeah. Let's talk about AI. Let's start with, how did you make the transition from Tesla to OpenAI? Because you were in very early. From the beginning. It wasn't obvious at the start that like-- not obvious at all.

一点也不。

Not at all.

你走到了现在的位置。是什么让你做出了这个跳跃？

You get to where you're at now. What made the leap?

在从事 VR 和 AR 工作后，我原本致力于定义空间计算的新界面，但那时我觉得 VR 和 AR 还有点早。不过那时我们开始对 AI 如何帮助我们重新定义与世界互动、吸收信息、创造事物以及它如何影响创造力产生了浓厚兴趣。就是整个增强智能的概念及其意义。我非常有兴趣了解更多，看看这能走多远，这个将智能作为基本属性、能产生广泛普遍影响的想法。当时我不确定它有多大可能走向 AGI（通用人工智能），但我非常想探索我们能走多远，它看起来像是我们最后要研究的东西。它似乎是我能从事的最重要的事情。对我来说，在一个关心确保它对世界有益的地方工作很重要。我加入 OpenAI 时它还是个非营利组织，公司的使命当时和现在一样：确保构建 AGI 对世界上每个人都好，人们能从中受益。显然，后来出于实际原因，我们调整了公司结构，成为有利润上限的有限合伙制，但使命不变，非营利组织监督公司使命。我只是追随我的好奇心和我当时认为最重要的事情。

After I worked in VR and AR, and was really intent on defining the new interface for spatial computing, back then it was a bit too early, I think, too early for VR and AR. But at that time we actually got really interested in how AI can help us redefine the way that we interact with the world and we absorb information and the things that we produce and how it affects creativity. Just this entire concept of amplifying our intelligence and what that means. I was really interested in learning more and seeing where this can go, this idea of pushing intelligence as a fundamental property that can have this very broad universal impact. At the time, I was unsure whether -- what the chances of that are to go all the way to artificial general intelligence. But I was just very interested in figuring out how far we could pursue it, and it really seemed like maybe the last thing that we'd ever work on. It seemed like the most important thing that I could work on. It was important to me to work on it at a place that cared about making sure that it goes well for the world. I joined OpenAI when it was a non-profit and the mission of the company was then and still is, to make sure that building AGI goes well for everyone in the world and people can benefit from what it will bring. Obviously since then, for practical reasons, we've evolved the structure of the company to have it be limited partnership with a capped profit. It still maintains the same mission and the non-profit oversees the mission of the company. But I just pursued my curiosity and what felt like the most important thing to me at the time.

老实说，我觉得这对任何人都是非常好的职业建议。能够选择你所做的事，相信你正在做的事情是你能够做出贡献的最重要的事情。我认为人们对此思考得不够刻意。

Which I like honestly I think is super good career advice for anyone. Being able to make choices about what you do, where you believe the thing that you're working on is the most important thing you can make a contribution to. I think people don't think deliberately enough about.

我认为这非常重要，因为当你从事非常困难的事情时，正是那种热情、那种天生的好奇心能够支撑你走下去。

I think it's so important because, when you're working on really hard things, it's that passion, that innate curiosity is the thing that can pull you through.

是的，百分之百。很高兴你这么说，因为我一直跟别人说这个。如果你和一群非常聪明、高度积极的人一起解决一个非常困难的问题，那很难。大多数时候你都在失败。你投入进去……

Yeah. A hundred percent I mean just really glad you said that because I say this to people all the time. If you're working on a really hard problem with a bunch of really smart, highly motivated people, it's hard. Like most days you're failing. You go in and.

没错。

Exactly.

你尝试一些东西，但行不通，你对自己感到沮丧，对周围的人感到沮丧，只有很少的东西能帮助你日复一日地坚持下去，直到真正解决问题，得到有意义的结果。如果你在解决问题之前放弃，那么你没有解决问题，除了累积的挫败感，你什么也没得到。我认为能让你坚持下去的少数几件事之一就是，你必须相信这是你能做的最重要的事情。你必须相信它很重要，光有钱不够，你妈妈希望你做也不够，它让你的简历好看也不够。你必须深深相信这是你能做的最重要的事情。

You're trying something and it doesn't work, and you're frustrated with yourself and you're frustrated with the people around you, and there are only a very small number of things that you can have that will help you do that day after day after day until you actually solve the problem, and you get something that matters. If you quit before you solve the problem, then you haven't solved the problem, you've got nothing but this accumulated frustration that you've had. I think one of the very few things that you can have that will get you through is, you have to believe that it's the most important thing that you could be doing. You have to believe that it matters, like money's not enough. Your mom wanting you to do it isn't enough. It looking good on your resume isn't enough. You have to just deeply believe that it's the most important thing you could be doing.

是的，没错。很难找到那种信念和信仰。你几乎需要一生去尝试，有时才能真正找到是什么带给你这种满足感。

Yeah, exactly. It's hard to find that faith and belief. You almost have to experiment a bit through, I mean your entire life and sometimes to just really find what that is, that really brings you this satisfaction.

是的。而且某个时候你还得找出自己应对挫折和失败的机制，因为这很难。我相信这对你所做的一切都是如此，因为你似乎反复选择了做非常困难的事情。

Yeah. Yeah, and at some point you also have to figure out what your mechanism is for dealing with that frustration of friction and failure because it's tough. I'm sure this is for everything that you've done because you seem to have repeatedly chosen to do very hard things.

我知道对我来说，我反复选择去做最重要的事，而它几乎总是你能选择的最难的事。所以能够长期坚持这一点，因为到了某个时候，你在特斯拉的职业生涯已经足够成功，从成功的角度来说，你本可以选择不去做最难的事——比如，我可以去做比在一个非营利组织里打造 AGI 稍微容易一点的事情，对吧？（笑）这听起来难如登天。

I know for me I repeatedly choose to do, the most important thing is almost always the hardest thing you could choose to do, and so just being able to sustain that over time because at some point too you probably had enough success from your career at Tesla where you could have chosen, just from a success perspective to not do the hardest thing, well, I can go do something slightly easier than try to make an AGI in a non-profit, right? (laughter) It sounds impossibly hard.

你这么说的话，确实如此。

When you put it like that, yes in fact.

我认为帮助 OpenAI 取得巨大成功的因素之一是你们拥有非常优秀的人才，每个人都在自己的领域里顶尖——无论是研究如何从 GPU 中榨取数值性能，还是懂得如何进行安全和对齐工作，或是懂得如何设计深度神经网络、理解分布式系统。你们在每个领域都有最顶尖的人，而且你们还有这个使命：如何解决这个极其复杂的问题——不仅是 OpenAI，人类已经思考了数千年——如何让它成为现实，并以一种为人类创造巨大利益的方式实现。但你们还有第三件有趣的事，那就是一种让人们专注于前进和进步的方法。你可以有使命，也可以有所有这些聪明人，但他们可能朝一千个不同的方向奔跑，他们的工作可能不会累积成进步。我认为这就是你们做到的非凡的第三件事，我不知道你是否认同这个观点，我只是好奇你的看法，或者那个缺失的元素是什么，因为很多实验室都有非常聪明的人，花了很多钱，也有有趣的智力使命，但他们仍然没能取得你们所取得的这种进步。

One of the things that I think has helped OpenAI be very successful is you have really excellent people, folks who are in their particular domain, whether it's figuring out how to wring numeric performance out of a GPU or if it's someone who understands how to do safety and alignment work or whether it's someone who understands how to architect a deep neural network, someone who understands distributed systems. You have just people who are at the very top of their game in each one of those areas, and you also have this mission, how do you go solve this incredibly complicated problem that not just OpenAI, but humanity's been sort of thinking about for thousands of years and how do you make that a reality and how do you do it in a way where it creates massive benefits for humanity. But you've got this third thing that's interesting, which is a way to keep people focused on moving forward and progress. You can have the mission and you can have all of these smart people, but they could be running in 1,000 different directions, and their work could not be accruing to a thing that's making progress. And I think that's sort of the extraordinary third thing that you all have been able to do, and I don't know whether you share that same perspective, I'm just sort of curious on your take or what that missing element is, because lots of labs are out there with really smart people spending a lot of money and they've got an interesting intellectual mission but they still haven't been able to make this sort of progress that you all have made.

这非常难。就像你说的，你可以拥有大量极具天赋的人，他们天生好奇，永远追求发现新事物，但这需要累积，你需要让所有聪明人一起在相似或相同的赌注上工作，你想激励人们，你不是雇佣聪明人然后告诉他们做什么——你想让他们有动力并且足够一致地做同一件事。在 OpenAI，我认为我们做得最好的一件事就是对我们最相信的事情下了一个或几个赌注，并且在很早的时候就达成一致，甚至在招聘阶段就把人带进来，确保他们真正认同这些事情。说“不”很难，尤其是当有这么多机会时，可以研究所有这些不同的想法。说“不”极其困难，你会怀疑自己。这些赌注可能需要一段时间才能见效，比如缩放定律和专注于一个大模型、大量数据，现在这很明显，但当时并非如此。在这方面达成一致非常困难，但我认为这回到了这样一个想法：弄清楚如何在正确的时间研究正确的问题，并对此抱有信念。

It's incredibly hard. Like you say, you can have these incredibly talented people in high density and they are innately curious and they're forever in pursuit of discovering something new, but that needs to compound, you need to have all the smart people working together on kind of similar or same bets, and you want to motivate people, you don't hire smart people, tell them what to do and -- you want them to be motivated and aligned enough to work on the same thing. At OpenAI I think one of the most important things that we managed to do well was take a bet or take a couple of bets on the things that we believed the most and get alignment on those very early on and even at the stage of recruiting people actually and bringing them in, that's most important and making sure they're really aligned on those things. It's hard to say no, especially when there is so much opportunity, it could be working on all these different ideas. It's incredibly hard to say no, and you doubt yourself. It might take a while for these bets to pan out, the scaling laws and focusing on one large model, a ton of data, which now it's obvious, but back then, not so much. Getting alignment on that is incredibly hard, but I think it goes back to this idea of figuring out how you work on the right problem at the right time, and having faith in that.

是的，我想深入探讨一下“说‘不’很难”这个概念。说“不”极其困难，因为作为 OpenAI 的 CTO，你面临的情况——我在过去二十年里也经常遇到——是世界上最聪明的人会带着非常好的想法来找你，你觉得很有趣，你是个好奇的人，你会说“太棒了，我喜欢这个”。然后，你知道那个想法不在你追求的道路上，如果你正在选择下一个最重要的事情，它可能不是下一个要研究的最重要的事情，而你要说“不”。同时你是个好人，和你一起工作的人也是好人，你不想让他们失望，不想让他们难过。所以我认为这真的是一门艺术，它有两部分：一是你自己要有信心和勇气说“不”，即使你自己也有不确定性，比如“我错了吗？我做对了吗？”；二是能够以一种方式传达“不”，它不是简单的拒绝，而是一种“不，但这里有另一件事，我认为如果你做这件事会更有趣、产生更大影响”的拒绝。这很难。

Yeah, I want to double-click on this notion of it's hard to say no. It's incredibly hard to say no, because the thing that you're faced with as CTO of OpenAI, and I have had a lot of this over the past two decades, is you will have the smartest people in the world coming to you with very good ideas that you think are interesting and you're a curious person and you're like, that's amazing, I love this. And then, you know that that idea is not on the path that you're pursuing and it might not be the next most important thing to go work on if you're choosing the next most important thing and just saying no, and you're also a good person and the people who you work with, are good people and you don't want to disappoint them and you don't want them to be sad, and so it's a real art form I think, and it's two parts, it's like having the confidence and the courage to say no, yourself, when you also have your own uncertainties like, 'am I wrong, am I making the right call?' and then being able to deliver the no where, it's not a no, it's sort of a no, but it's no, here's this other thing that I think if you do that it will be even more interesting and create more impact, it's hard

完全正确，这极其困难。与此同时，还要培养组织快速学习新事物、快速了解什么行不通、快速采纳有效方法并快速淘汰旧想法的能力。淘汰那些可能已经在运作但不如新事物有效的东西，这很难。

Exactly, it is extremely hard. Together with that goes building the muscle as an organization to learn new things quickly or learn what's not going to work very quickly and adopt what's going to work very quickly and kill the old ideas quickly. It is hard to kill things that are already maybe working but they're not working as well as something new that you could be doing.

是的，嗯，看，我认为这是你们做得非常好的另一件事，而且非常重要，那就是选择何时停止做某事。比如，几年前你们有一个非常棒的演示：一只机械手单手解魔方，这个演示试图让强化学习系统学习机器人运动学模型。这在技术上有趣，是一个超酷的演示，但你们决定这不属于路径，所以停止研究。对我来说这是一个艰难的决定，因为对某人来说那是大量工作，而且可能是他们最喜欢的事情。最终人们可能会辞职，因为你停止了这件事，而那是他们想做的，所以他们去别的地方做，但这很重要。真的非常重要。

Yeah, well, look, I think that's another thing that you all do really well, and it's very important is choosing when to stop doing things. Like, for instance, you-all had like an incredibly great demo a handful of years ago of a robotic hand that could single hand solve a Rubik's cube and it was a demo that was trying to get a reinforcement learning system to learn a robotic kinematic model. It's technically interesting work. It's a super cool demo, but like you-all decided, this isn't on the path, so we're going to stop working on this and that's a hard decision for me because that was a lot of work for someone and it was like their favorite thing in the world. It's like at the end people may quit because, you stop doing this thing and that's the thing they wanted to work on so they're going to go find some other place to go work on it but it's important. Really important.

是的，完全正确。当时这对公司来说是一个非常大的赌注。我们有那个项目，还有 DOTA。我们到了一个转折点：好吧，我们想学什么？这如何融入我们通往 AGI 的道路？有没有更好的方法？选择停止研究它，是因为认为有更好的方法。

Yeah. Exactly. At the time it was a very big bet for the company was making. And we had that and DOTA. We had this inflection point that okay what are we trying to learn? How does this fit in on our path to AGI and is there a better way? Choosing to stop working on it, thinking there's a better way.

你说了很多非常重要、深刻的话，比如你刚才说的，我认为也很重要的一点是：我们想学什么？我的意思是，如果更多人刻意问这个问题，世界会变得更好，人们会更成功。但我的意思是，这本质上，我认为是你们一直非常关注的事情之一。你们不是为了活动而活动，或者为了证明自己聪明而活动。而是我们通过正在做的事情有一个特定的学习目标，这不一定是 AI，可以是产品设计，或者育儿，或者其他任何事情。

You've said a lot of very important profound things like you just said something that I think is also very important is what are we trying to learn? I mean if more people asked that question deliberately, we would have a much better world and people would have more success. But I mean, that is in essence, I think, one of the things that you-all have always had pretty good focus on. It's you're not doing activity for the sake of activity or like doing activity for the sake of proving that you're smart. It's we have a specific thing we're trying to learn through these things that we're doing and it doesn't have to be AI. It could be product design or it could be like parenting or whatever.

你通过正在做的事情想学到什么？我们来聊聊。我是说，你们整个历程令人难以置信，但特别是过去一年甚至过去六个月，我想对很多人来说是震惊的。我关注你们的工作有一段时间了，所以过去六个月发生的事情对我来说是惊讶的，但不像那些什么都没看到然后突然 ChatGPT 出现、成为全世界最有趣东西的人那么震惊。聊聊那段历程吧，因为我认为 ChatGPT 只是你们长期努力中的一个点，甚至不是最后一个，所以另一件人们可能没有意识到的是，它是一个曲线上的点，更多的东西正在到来。那么你们是如何思考公众反应的呢？

What are you trying to learn through this thing that you're doing? Let's talk a little bit about, I mean, you-all have had an unbelievable total run, but in particular the past year or even the past six months have been, I think, shocking to a bunch of folks. I've been following what you-all have been doing for a while and so what happened the past six months, I mean, it was surprising to me. But not quite as shocking to folks who saw nothing, nothing and then all of a sudden ChatGPT emerges and it becomes the most interesting thing in the world. Talk a little bit about that journey because I think ChatGPT is just one point on a long set of efforts that you-all have been working on and it's not even the last thing, so that's the other thing people probably aren't internalizing that it is a point on a curve and more things are coming. So how have you-all thought about that in the context of how the public's reacting?

我们第一次考虑部署这个还处于研究阶段的模型时，那是个疯狂的想法。那时候在现实世界中部署大型语言模型并不正常。商业案例是什么？它到底能为人们做什么？能解决什么问题？我们并没有这些答案。但我们想，如果我们让它变得易于使用且便宜，高度优化，你不需要了解机器学习的各种细节，只需触手可及，那么也许人们的创造力会带来新的产品和解决方案，我们就能看到这项技术如何在现实世界中帮助我们。当然，我们有一个假设，但实际上只是把 GPT-3 放到 API 里。第一次看到人们与这个大型模型和我们正在构建的技术互动时，那是多年来一直在实验室里构建，没有现实世界背景和外界反馈的第一次。这是一个信念的飞跃，它将会教会我们一些东西，我们会从中学习，并希望能反馈到技术中。我们可以带回这些知识和反馈，并找出如何利用它使技术变得更好、更可靠、更对齐、更安全、更稳健，最终在现实世界中部署。我一直相信，你不能只在实验室里构建这种强大的技术，不与现实接触，却希望一切顺利，安全且对所有人有益。你确实需要想办法让社会参与进来，既要收集反馈和见解，也要让社会适应这种变化。最好的方式就是让人们实际与技术互动，亲眼看到，而不是告诉他们或只是分享科学论文。这非常重要。我们花了几年的时间才达到不仅通过 API 发布模型改进，而且第一个面向消费者的界面是 DALL-E，DALL-E 实验室。人们只需用自然语言输入提示，就能看到这些美丽、原创、惊人的图像。然后出于研究原因，我们尝试了对话界面，在 ChatGPT 中与模型来回交流。对话是一个非常强大的工具。苏格拉底式对话的理念以及人们如何学习。你可以互相纠正或提问，深入探讨更深层的真理。所以我们想，即使使用现有模型，把它放出去，我们也会学到很多。我们会得到大量反馈，并利用这些反馈让即将推出的模型（当时是 GPT-4）更安全、更对齐。这就是动机。当然，正如我们所见，仅仅几天内它就变得超级流行，人们喜欢与这个 AI 系统互动。

The first time that we thought about deploying this model that was just in research territory was this insane idea. It wasn't normal back then to go deploy a large language model in the real world. What is the business case? What is it actually going to do for people? What problems is it going to solve? We didn't really have those answers. But we thought if we make it accessible in such a way that it's easy to use and cheap to use, highly optimized so you don't need to know all the bells and whistles of machine learning, just accessible, then maybe people's creativity would bring to life new products and solutions, and we'll see how this technology could help us in the real world. Of course, we had a hypothesis, but really it was just putting GPT-3 in the API. The first time that we saw people interact with this large model and the technology that we were building, that for so many years had just been building in the lab without this real-world context and feedback from people out there, that was the first time. It was this leap of faith that it was going to teach us something, we were going to learn something from it, and hopefully we could feed it back into the technology. We could bring back that knowledge, that feedback, and figure out how to use it to make the technology better, more reliable, more aligned, safer, more robust when it eventually gets deployed in the real world. And I always believed that you can't just build this powerful technology in the lab with no contact with reality and hope that somehow it's going to go well and it's going to be safe and beneficial for all. Somehow you do need to figure out how to bring society along, both in gathering that feedback and insight, but also in adjusting society to this change. And the best way to do that is for people to actually interact with the technology and see for themselves instead of telling them or just sharing scientific papers. That was very important. And it took us a couple of years to get to the point where we were not just releasing improvements to the model through the API, but in fact, the first interface that was more consumer-facing that we played around with was DALL-E, DALL-E labs. Where people could just input a prompt in natural language and then you'd see these beautiful, original, amazing images come up. And then really for research reasons, we were experimenting with this interface of dialogue, where you go back and forth with the model in ChatGPT. And dialogue is such a powerful tool. The idea of Socratic dialogue and how people learn. You can correct one another and/or ask questions, get really into deep, deeper truth. So we thought if we put this out there, even with the existing models, we will learn a lot. We will get a lot of feedback and we can use this feedback to actually make our upcoming model, that at the time was GPT-4, safer and more aligned. So there was the motivation. And of course, as we saw in just a few days it became super popular and people just loved interacting with this AI system.

我个人对你们的工作感到兴奋的原因之一是，你们希望让很多非专家的人能够尝试这项技术，并想象如何用它来做他们认为重要的事情，这对我来说非常重要。也许你也有同感。但我成长的地方不是像沿海创新中心那样创造 AI 系统的地方。我的父母不是计算机科学家或工程师。而弗吉尼亚中部农村的人们面临的问题，我猜阿尔巴尼亚的人们面临的问题，有些是普遍的，但有些非常不同。如果你的整个世界观是：我上了斯坦福，我在世界上最大的科技公司之一工作，我在构建这项技术，我必须想象它所有可能的用途，你根本无法想象来自阿尔巴尼亚或弗吉尼亚农村的人的生活是什么样的。所以我认为，让这些东西成为平台，而不是仅仅在实验室里构建，所有重大决策都在不与现实世界接触的情况下做出，这极其重要。这是我在时间用完前想聊的最后一件事。但这带来了一个非常困难的问题：如何负责任地做 AI。因为你获得了许多人参与的巨大好处，但同时你也面临一大堆需要解决的问题，以确保它不会造成大量伤害。那么你们是如何思考这个问题的呢？

One of the reasons why just me personally, I've been excited about the work that you-all are doing is this notion that you want to really allow a lot of non-expert people to be able to play around with the technology and to imagine how they can use it for things that they think are important is super important to me. And maybe a little bit of same is true for you. But like I grew up, not in like one of the coastal innovation centers where things like these AI systems get created. You did not have computer scientist or engineer parents. And the problems that people have in rural Central Virginia, and I'm guessing the problems that people have in Albania, some of them are common across the board, but some of them are like very different. And some of them you can't even imagine if your entire worldview is like, I went to Stanford, I got a job at one of the biggest technology companies in the world and I'm building this technology and I have to imagine all of its possible uses. You just can't even imagine what life is like for someone from Albania or rural Virginia. And so I think it's really unbelievably important to have these things be platforms that aren't just getting built in a lab where all the consequential decisions get made without any contact with the real world. I mean, this is the last thing I want to chat about before we run out of time. But it creates this very hard problem of how you do responsible AI. Because you get this big benefit of lots of people participating, but then you get this big bucket of things that you have to go solve at the same time to make sure that it's not creating a whole bunch of harm. So talk a little bit about how you-all think about that.

是的，说得好。这些权衡和最小化。你不可能零风险，但真正要最小化那些伤害，并且能够快速响应和迭代，可能对模型本身进行更改，或引入工具或政策来遏制这些伤害。这非常困难，因为我们经常在公众视野中做这一切。我们没有关起门来做的特权。显然，这伴随着一定的责任。但我认为实际上没有其他办法。我认为这是唯一正确的做法。它确实需要在公众视野中，并且需要处于这种持续的迭代循环中，因为现在技术进步的速度是疯狂的。如果你把系统留在实验室里，如果我们从未发布 GPT-3 或 3.5，而是直接推出 ChatGPT 上的 GPT-4，那将会震惊世界，它已经震惊了。我们有了这个持续的发展周期。我认为这非常重要。但其中一点是，从每次部署中，从每次我们发布模型时，我们都学到了一些东西。

Yeah, that's well put. These trade-offs and minimizing. And you can't have zero risk, but really minimizing those harms and actually being able to respond quickly and iterate quickly on being able to maybe make changes to the models themselves or introduce tools or policies basically to contain those harms. That's really difficult because often we're doing all of this in the public eye. We don't have the privilege of doing it behind closed doors. And so obviously with that comes a certain responsibility. But I think actually there is no other way to do it. I think it's the only way to get it right. It does need to be in the public eye and it needs to be in this continuous iterative cycle because the rate of technological advancement right now is insane. If you hold the systems back in the lab, the difference between if we had never released GPT-3 or 3.5 and we had just gone out with GPT-4 on ChatGPT, that would have shocked the world, it already did. We had this continuous development cycle. I think that's really important. But one of the things is from each deployment, from every time that we put out a model, we learned something.

我们在早期开发周期或产品周期中了解到系统的安全性，安全深深嵌入并整合到开发和部署这些模型的每个阶段，我们不断调整做法，因为我们一直在学习新东西。可以说每周我们都在学习新事物。无论是关于如何选择、过滤和分析早期数据，还是关于基于人类反馈的强化学习（RLHF）过程——它让模型更对齐，或是我们在生产中使用的分类器，或是我们为开发者提供的工具，让他们能掌控并引导这些模型。所有这些环节贯穿了从研究到生产的整个生命周期。

We learn something about maybe the safety of our systems in the early development cycle or impose training or in the product cycle, safety is really deeply embedded and integrated at each stage of developing and deploying these models and we're constantly changing what we're doing because we're just constantly learning new things. Every week, I would say we're learning something new. Whether it's how you think about the data that you're selecting and filtering and analyzing the data early on or about the RL, Reinforcement Learning with human feedback process that makes these models more aligned or classifiers that we use in production or the tools that we're making available for developers to have control and be able to be in the driver's seat and steer of these models. All these pieces along the life cycle of taking research to production.

管理这些权衡确实很复杂。但我同意你的看法，我不知道是否有其他合理的替代方案。关键在于获取大量输入，这样你就能听到什么有效、什么无效、受到的审视、哪些问题看似重大实则不然，以及人们以你从未想象或预期的方式使用产品时出现的奇怪情况。

It's a complicated set of things to manage these tradeoffs. But I agree with you. I don't know if there is any other reasonable alternative and I think the trick is having lots and lots of inputs that are coming into you like where you can hear what's working, what's not working, what is the scrutiny, which of the problems that seemed substantial or not and which of the things that people are seeing in some weird permutation of how they're trying to use the product that you never imagined or intended.

正是如此。

Exactly.

这既令人兴奋，又是一种巨大的责任。

It creates - It is on the one hand very exciting. But it's also like a huge responsibility I think.

是的。我们正在做的事情将改变一切——改变我们的工作方式、互动方式、思考方式，以及生活的方方面面。

It is. We're working on something that will change everything, it will change the way that we work, the way that we interact with each other, and the way that we think, and everything really, all aspects of life.

最后一个问题，我问过每位播客嘉宾。我知道过去一年你非常忙碌，但我想知道你在工作之外有什么爱好。

Yeah, I have one last question for you that I ask everybody who's on the podcast. I know you probably have no free time given the intensity of the past really year. But I ask everyone what they do outside of work for fun.

我喜欢阅读和徒步旅行。徒步是我最喜欢的活动之一，亲近大自然。

I love reading and I love going for hikes. Hiking is one of my favorite things to do, being in nature.

我们住的地方很适合徒步，太好了。非常感谢 Mira 在百忙之中抽出时间进行这次对话。我学到了很多，也很享受这次交流，并且很高兴能经常与你合作。

We live in a good place for hiking, which is good. Awesome. Well, thank you so much Mira for taking time out of an incredibly busy schedule to have this conversation. I've learned a ton and just enjoyed this conversation and enjoy being able to work with you on a regular basis.

太好了，我也是。非常感谢。

Awesome. I do too. Thanks so much.

哇，与 Mira Murati 的对话真是太精彩了。作为密切的合作伙伴，我经常与 Mira 和她的团队合作，帮助他们开发大型 AI 系统，并思考如何将这些极其复杂的 AI 系统安全地部署到我们的产品中。但今天我对 Mira 有了更多了解。我知道她来自阿尔巴尼亚，但对她最初如何对科学和技术产生兴趣知之甚少。听到她讲述她的老师——总是走在前面，培养她的好奇心，让她在厌倦手头内容时去翻阅姐姐的课本——真是太棒了。她说姐姐比她大一岁半，当她厌倦姐姐的东西时，就去找其他东西学。你在我们的对话中多次听到类似的话：接下来学什么？为什么那件事值得学？这种总有更多东西可学的信念，我认为是推动 Mira 取得如此成功、并带领团队走向成功的关键之一。这对我们所有人来说都是一个很好的职业建议：有意识地思考当前的活动，将其视为学习的机会，从而在工作中不断进步，并更有目的地将精力投入到未来。听到她在特斯拉的经历也很棒，这塑造了她作为领导者的工作方式，以及她如何处理那些需要多学科交叉团队协作的复杂问题——必须将许多不同观点的人聚集在一起，完成这些超级复杂的任务。听她谈论对智能的热情，以及这对我们如何与复杂技术交互的意义，还有他们长期以来如何思考将所做的工作打包成大众可用的形式，真正激发许多人的想象力和好奇心。你赋予他们使用这项技术的能力，从他们的视角去做有趣的事情。总之，这是一次引人入胜的对话，其中还有很多精彩片段。在对话中，我注意到她随口说出的几句话，我认为都是非常有价值的智慧结晶。希望大家都能反思这次对话的意义。今天的时间就到这里。非常感谢 Mira Murati 的参与。如果您有任何想分享的内容，请随时发送邮件至 BehindTheTech@Microsoft.com。您可以在 YouTube 和任何常规播客平台关注我们。下次再见。

Wow, that was a fascinating conversation with Mira Murati. As close partners, I get to work with Mira and her team all the time, helping to develop some of the big AI systems that they're building and then figuring out how to safely deploy those unbelievably sophisticated AI systems into the products that we're building. But I learned a ton about Mira today that I didn't know before. I knew she was from Albania, but I had known relatively little about how she first got interested in science and technology in the first place. It was so great to hear about her teachers, always being ahead, and having those teachers who were nurturing the curiosity that she had her going through her sister's textbooks when she got bored with the stuff that she was working on. I think she said a year and a half older than she was and then when she got bored with her sister's stuff, figuring out what else there was to learn. I think you heard at a bunch of places in our conversation like that, what am I going to go learn next? Why's that thing important to learn? And this belief that there's always something more to go learn is one of the things I think that has driven Mira to such success and that the teams that she's responsible for leading to success. I think it's a good piece of career advice for all of us to be just very intentional about how we're thinking about the activity that we're doing right now as an opportunity to learn something that will help us get better and better at our jobs and to be more purposeful about how we invest more of our energy in something into the future. It was awesome to hear about her experience at Tesla, which I think has really shaped how she does her job as a leader and how she tackles these complicated things where there are multi-disciplinary, intersectional teams, where you have to pull a lot of people together with a lot of different points of view to do some of these super-complicated things. Just hearing her talk about her passion for intelligence and what that means for how we are going to interface with complicated bits of technology and how they really have been thinking for a long while about how they take what they do and package it in a way where lots of people can use it and where you really can unlock the imagination and the curiosity of a lot of other people. You're empowering them to use this technology to do the interesting things from their points of view. Anyway, that was just a fascinating conversation. There were more tidbits in there. I found myself during the conversation remarking on several points where she said something almost in passing that I thought were real super valuable nuggets of wisdom. I hope everybody gets a chance to reflect on what this conversation really means. And that's all the time we have for today. Big thanks to Mira Murati for joining us. If you have anything you'd like to share with us, please email us anytime at BehindTheTech@Microsoft.com. You can follow us on YouTube and on any of the usual places that you go get your podcasts goodness and until then, we'll see you next time.