AI 先驱吴恩达谈智能体工作流与编程的未来

未来最重要的技能之一，是能够精确告诉计算机你想要什么，让它们为你执行。在可预见的未来，懂得计算机语言、会编程的人，会比不会的人高效得多。我知道今年早些时候，有些领袖建议别人不要学编程，理由是 AI 会将其自动化。我认为，未来我们会把这话看作有史以来最糟糕的职业建议之一。

One of the most important skills for the future is the ability to tell a computer exactly what you want so they can do it for you. And for the foreseeable future, people that know the language of computers—being able to understand coding—will be able to do that much more effectively than people that don't. So, I know that even earlier this year there were some leaders that were advising others not to learn to code on the grounds AI will automate it. I think we'll look back on that as some of the worst career advice ever given.

听众朋友们，大家好。今天我们要分享今年 Masters of Scale 峰会上的另一场对话。Andrew Ng 是真正的 AI 先驱。他是 Coursera 和 DeepLearning.AI 的联合创始人，也是 AI Fund 的管理合伙人，AI Fund 是一个孵化新 AI 公司的创业工作室。与他同台的是 DJ Patil，奥巴马政府前首席数据科学家，Greypoint Ventures 的普通合伙人。这是一场关于 AI 现状、未来一代应如何对待这项技术、以及美国如何在全球 AI 竞赛中保持竞争力的精彩讨论。我迫不及待想与你们分享。让我们开始吧。

Hi there, listeners. Today, we are sharing another conversation from this year's Masters of Scale Summit. Andrew Ng is a true AI pioneer. He's co-founder of Coursera and DeepLearning.AI and managing partner at AI Fund, a studio that incubates new AI companies. He was joined on stage by DJ Patil, former chief data scientist under the Obama administration and general partner at Greypoint Ventures. It's a dynamic discussion about the current state of AI, how future generations should approach the technology, and how America can stay competitive in the global AI race. I can't wait to share it with you. Let's jump in.

我认识 Andrew 很久了。早在 AI 流行之前，他就一直在谈论和研究 AI。我记得我们曾一起讨论数据科学等想法，那时他就在谈 AI，我当时想，‘嘿，AI 还在寒冬呢。’但他一直走在这段不可思议的旅程上。给你一些 Andrew 的亮点：他是最早倡导使用 GPU 进行深度学习的人之一。

I've known Andrew for a long time. He's been talking and working on AI long before it was cool. I remember actually sitting down with Andrew when we were coming up with the ideas around data science and these things. And then he was talking about AI and I was like, 'Hey, we're still in the middle of winter for AI.' But he's been on this incredible journey. To give you some of the highlights of Andrew: he was one of the first people to advocate using GPUs for deep learning.

是啊，当初该买英伟达股票的。

Yeah, should have bought Nvidia stock.

我正想说，黄仁勋欠你多少架飞机？

I was going to say, Jensen, how many jets does Jensen owe you?

是啊，得看多大尺寸，对吧？

I know. Like, what size, right?

为他高兴。他给过你什么吗？

Happy for the guy. Did he ever give you anything?

他给了我几块 GPU。那挺好的。

He gave me a few GPUs. That was nice.

GPU。你从中得到了几块 GPU。好吧。你编写了第一门关于机器学习和 AI 的大型在线课程，后来催生了 Coursera。当时教授在线课程是激进的想法，它已经帮助了超过 1000 万学生。对吗？1000 万？

GPUs. You got some GPUs out of it. All right. You wrote the first major online course on machine learning and AI, which led then to Coursera. At the time it was radical thinking to actually teach an online course, and it's helped over 10 million students. Is that correct? 10 million?

是的。谢谢。没错。

Yeah. Yeah. Thank you. Yeah.

了不起的成就。你和 Jeff Dean、Greg Corrado、Rajat Monga 一起启动了 Google Brain 项目。还有很多，但你目前在做的事情：你有一个基金和一个工作室，投资和构建 AI。你还有当今 AI 领域一些最具开创性的高引论文。

Incredible accomplishment. Along with Jeff Dean, Greg Corrado, and Rajat Monga, you started the Google Brain project. Much more, but some of the things you're doing right now: you have a fund and a studio investing and building with AI. And you also have some of the most seminal cited papers in AI today.

我得承认，我可能拥有被引次数最高的论文。我不像以前那样频繁追踪引用数了。

I have to admit, I probably have the highest cited papers. I don't track my citation count as often as I used to.

没错。你可以让别人或 AI 系统去做。但还有件事你可能不知道。现在有多少人听说过智能体式系统？也许我该反过来问，好让大家注意到。猜猜‘agentic’这个词从哪来的？就是这位。很少有人知道，Andrew 其实是真正提出‘agentic’的人。我们就从这开始吧。‘agentic’背后的故事是什么？

Exactly. You can do it for everybody else or the AI systems. But we're also something that you don't know about. How many people have heard about agentic systems these days? Yeah, maybe I should ask it the other way around just to highlight people. Guess where agentic came from? This is the guy. It's a little known fact that Andrew was actually the guy who really came up with agentic. Actually, let's start with that. What's the story behind agentic?

大约两年前，我看到了 AI 中一个正在兴起的趋势，很多人对此兴奋，但科技界内部争论不休：有人写软件说它是智能体，另一些人说‘不，那不是智能体。是智能体吗？不是。’我想，‘这是在浪费时间。我们不如不要非黑即白地争论是不是智能体，干脆都叫智能体式，然后停止争论，继续工作。’于是，我实际上发起了一场没有公开宣传的运动，试图让更多人采用‘agentic’这个词。我没料到的是，几个月后，一群营销人员抓住了这个词，像贴纸一样贴在所有东西上，这反而推动了这场运动的爆发。但尽管炒作如此，我认为真正的价值也在快速增长。所以，这很令人兴奋。

So, almost two years ago, I saw this rising trend in AI that a lot of people were excited about, but within the tech community there was all this debate with some people that write software and say it's an agent, others say, 'No, that's not an agent. Is it agent? Not an agent.' I thought, 'This is a waste of time. Why don't we instead of having a binary agent/not agent, let's just call it all agentic and then stop arguing and get on with the work.' And so, I actually kind of ran a campaign that I didn't publicize, but I did it anyway to try to get more people to just adopt the word agentic. What I didn't realize was a few months later a bunch of marketers would get hold of this word and slap it as a sticker on everything in sight, and that helped the movement take off. But even though the hype has gone like that, I think the real value is really growing rapidly, too. So, that's been exciting.

好，我们继续这个话题。我想分四个阶段来谈。首先谈谈当下。作为元老之一，你如何看待 AI 的现状，特别是从智能体式的角度？因为我觉得我们都在被各种营销和炒作困扰，分不清真假。未来的大量工作是将这些惊人的智能体式 AI 能力映射到真实的业务流程中。我想我们正在做这件事。

Well, let's stay on that. I want to do this in four stages. Let's first talk about today. And talking about today and the state of AI through the lens of one of the OGs. What is your honest take of what AI can and can't do, especially through this lens of agentic, because I think we're all struggling with all the marketing and buzz out there of what's real and what's not. A lot of the work that lies ahead is to take these amazing agentic AI capabilities and map it to real business workflows. I think we'll be doing that.

什么是智能体式？我们先明确一下。很多人使用 AI 大语言模型时，是直接提示它并让它写出输出。这有点像对一个人（或这里的 AI）说：‘请写一篇文章，从第一个字到最后一个字一次性打完，中间不停顿思考，也不使用退格键。’人类不会以这种方式写出最好的文章，AI 也不会。智能体式工作流的思想是，我们可以让 AI 采取更迭代的方法，比如：‘先写大纲，然后做更多研究，再写初稿，然后批判它。’迭代工作流耗时更长，但对于很多任务——从医疗建议、法律建议、关税合规到编写代码等各种事情——这些智能体式工作流效果要好得多。但我们面前还有很多工作要做。

What is an agentic? Let's ground us there. So, a lot of us use AI large language models by prompting it and asking it to write an output. That's a bit like going to a human or in this case an AI and say, 'Please write an essay by just typing it out from the first word to the last word all in one go without stopping to think or without ever using backspace.' So, humans don't do our best writing like that and neither does AI. With agentic workflow, the idea is we can ask an AI to take a more iterative approach and say, 'First write an outline, then do some more research, then write the first draft, then critique it.' And so, the iterative workflow takes much longer, but for a lot of tasks from medical advice, legal advice, tariff compliance, writing code for a lot of different things, these agentic workflows work much better. But there's still a lot of work ahead of us.

我知道有些人说：‘哦，别担心。等 AGI 吧，它会解决所有问题。’我不喜欢这种‘等 AGI’的说法，我觉得那很炒作。你今天做的很多非常有价值的工作，是拿技术以及未来 6 到 12 个月可能实现的东西，直接去做有价值的事情。有时我想到一个类比：我在使用这些系统，因为我试图部署 AI 来真正帮助老年人进行医疗保健等。有时我会想：‘我是在厚冰上，还是在薄冰上？系统能做什么、不能做什么？’我猜很多人都在疑惑：‘它对这个管用吗？还是说很脆弱？’有时我们只得到 80%的解决方案，有时它大获成功，有时我们极度失望。作为 Andrew，你在创办公司、给那么多人建议，你是怎么看待这个问题的？

I know that some people say, 'Oh, don't worry about it. Wait for AGI. That'll solve all the problems.' I'm not a fan of this 'let's wait for AGI,' you know, that feels hypey to me. A lot of work that's very valuable that you're doing today is to take the technology and what may be possible in the next 6 to 12 months and just go do valuable stuff with it. Sometimes the analogy I think of is as I'm using these systems because I'm trying to deploy AI to really help senior citizens in their health care journey, different things. And sometimes I think, 'Am I on thick ice? Am I on thin ice with what the systems can and can't do?' I suspect a lot of people out there are wondering, 'Does it work for this problem or is it fragile?' Sometimes we just get to an 80% solution, other times it knocks out of the park, sometimes we're incredibly disappointed. How do you as Andrew, who's building companies, advising so many people, think about this?

是的，这很难。我认为，一个任务越接近纯文本处理，并且如果你有管道获取完成任务所需的所有信息（最好是文本形式），AI 就越容易完成它。

Yeah, it is tough. I think the closer a task is to only text processing, and if you have the plumbing to get all the information, hopefully in text, that people need to do a task, the easier it is for AI to do it.

如果你需要输入图像、语音对话，这并非不可能，但会变得更难。我经常问的一个问题是：人类知道很多东西，我们拥有大量上下文。那么，我们是否有数据管道来为 AI 系统提供类似人类执行该任务所需的上下文？对于许多多步骤流程，如果你能编写标准操作程序（SOP），那也可能是一个信号，表明值得尝试将 SOP 转化为多步骤智能体式工作流。很难确定什么能做、什么不能做，但我认为这些可能是评估成功可能性的一些建议。

If you need to feed in images, voice conversations, it's not impossible, but it gets harder. And then one question I often ask is humans know a lot of stuff. We just have a lot of context. And so, do we have the data plumbing to get the AI system similar context that a human would need it to do that task? And then I think for a lot of multi-step processes, if you can write a standard operating procedure, like an SOP, that also may be a sign that it's worth seeing if you can qualify the SOP in a multi-step agentic workflow. It's hard to determine what can and cannot be done, but I think these are maybe some suggestions for rating what's more or less likely to succeed.

让我们换个话题，谈谈教育。你通过 Coursera 改变并转型了学术界。你目前的在线课程有数不清的观看量和学习者。我相信每位家长都在问你，他们的孩子应该做什么来为 AI 做好准备。孩子们还应该学编程吗？计算机科学还有用吗？数据科学呢，还是说这是个坏主意？

Well, let's switch gears and go to education. You've changed and transformed academia through Coursera. Your current online courses have so many views and people taking them. I'm sure every parent is asking you what their kid should do to be prepared for AI. Should kids still learn to code? Is CS a thing? Is data science a thing or is that a bad idea?

是的，未来最重要的技能之一就是能够精确地告诉计算机你想要什么，让它们为你执行。在可预见的未来，懂计算机语言、会编程的人将比不会的人高效得多。我知道，甚至今年早些时候，一些领导者建议别人不要学编程，理由是 AI 会将其自动化。我认为我们回头会认为这是有史以来最糟糕的职业建议之一。我已经在我的团队中看到，很多硅谷团队里，不仅仅是软件工程师，还有市场营销、人力资源、分析师、财务人员——那些会编程的人开始远远甩开不会的人。所以，如果你的孩子想成为软件工程师，让他们学习用 AI 编程。即使他们不想，未来我们也需要更多软件创造者，而不仅仅是用户。与其让你的孩子长大后问‘有没有一个应用能解决这个问题？’，我希望他们说‘我为此建了一个应用’。在 AI 的帮助下，编程比以前容易多了。所以，不要手工编码，让 AI 帮你做，这样做的人会比不这样做的人更强大、更有效。这也是孩子们必须面对的一个挑战。这是一项新技能。就像今天，我无法想象上大学却不学习如何进行网络搜索，对吧？那会很奇怪，也会限制你的就业前景。未来，我认为如果你上了大学却不会创建软件，我们会说‘哦，那有点奇怪’。这会限制一个人的发展前景。

Yeah, so one of the most important skills for the future is the ability to tell a computer exactly what you want so they can do it for you. For the foreseeable future, people that know the language of computers and understand coding will be able to do that much more effectively than people that don't. I know that even earlier this year there were some leaders advising others not to learn to code on the grounds that AI will automate it. I think we'll look back on that as some of the worst career advice ever given. I'm already seeing on my teams, lots of Silicon Valley teams, not just the software engineers, but the marketers, HR professionals, analysts, finance professionals—the ones that know how to code are starting to run circles around the ones that don't. So, if your kid intends to be a software engineer, have them learn to code with AI. And even if they don't, it is becoming clearer that in the future we need a lot more creators of software, not just users. Rather than your kid growing up and asking, 'Is there an app for that?' I want them to say, 'I built an app for that.' With AI assistance, coding is much easier than it used to be. So, don't code by hand, get AI to do it for you, and people that do that will be more powerful and effective than people that don't. That's one challenge the kids system has to go through as well. There's this new skill. Just like today, I can't imagine going through college without learning how to do web search, right? That would be weird and limit your job prospects. In the future, I think if you go through college and come out not knowing how to create software, we'll go, 'Oh, that's kind of weird.' It will limit the prospects of what someone can do.

当我们思考孩子应该在什么时候接触这项技术时，这应该是什么样子的？特别是从我们在社交媒体上开始遇到的教训来看。前卫生局局长维韦克·穆尔蒂确实强调了这方面的挑战。你认为什么时候太早？什么时候是合适的时机，让人真正开始接触 AI，以确保他们成为真正的 AI 原住民，并获得这项技术的最大益处？

When we think about when should a kid get access to this technology? What does that look like? And specifically through the lens of some of the lessons we've really started to struggle with around social media. Surgeon General previously Vivek Murthy really highlighted the challenges happening around that. When is too early in your idea? When is the appropriate time for someone to really start having access to AI to make sure they're truly AI native and get the maximum benefits of this technology?

我知道，这很难。我觉得就像我提到孩子接触书籍一样，我们认为他们很小就可以，但也有一些书不适合两岁的孩子。我认为技术的一个挑战是，有些应用对很小的孩子来说没问题，但也有许多东西我们不会让小孩子使用。我的孩子分别是 4 岁和 6 岁，他们偶尔会用平板电脑，但我会陪着他们。我不会把它当作保姆，而是让他们做教育性的事情或一些奇怪的事情，然后我们一起讨论。所以，我认为媒介已经从其他东西变成了技术，但挑战在于，公司有什么商业动机去为孩子创造或不创造某些体验。作为父母，我们如何设置护栏来管理所有这些东西？就像我不会让我的孩子读某些非常不适合他们年龄的书一样，我也不会让他们做某些非常不适合他们年龄的事情。但考虑到某些类型公司的动机，他们可能会做我们作为父母不希望他们做的事情，这是一个挑战。

I know. I think it's difficult. I feel like when I was mentioning kids get access to books, we think really, really young, but there are also some books that are inappropriate for a 2-year-old. I think one of the challenges of technology is there are apps that are just fine for a very young child to use, but there are also a lot of stuff that we would not let a young child use. My kids are 4 and 6, they do use tablets occasionally, but I'm there with them when they're using it. I'm not using it as a babysitter, but having them do educational things or doing weird things and we talk about it. So, I think the medium has changed from other things to tech, but the challenge is what are the business incentives for companies to create or not create certain experiences for kids. And as parents, how can we have guardrails to curate the whole four things? Just like I don't let my kids read certain highly inappropriate books for their age, I don't let them do certain highly inappropriate things for their age. But that is a challenge given the incentives of certain types of companies to do things that we as parents may not want them to.

你有没有和那些公司或团体谈过，说‘嘿，别这样，这没帮助’？或者那样的对话是怎样的？因为有很多上过你课的人实际上正在做一些我们作为父母认为很有问题的行为。

Do you ever talk to some of those companies or groups and say, 'Hey, knock it off. That's not going to be helpful.' Or what's that conversation like because these are a lot of people who've taken your classes that are actually doing some of the behaviors that as parents we find really problematic.

你知道，硅谷 99%的工程师和商界人士都想做正确的事。做这些事的人，坦率地说，是我们的朋友。也许这个房间里现在就有人。我认为每个人都想做正确的事。我希望我们能找到一种方法，在涉及数十亿美元利益时，仍然始终做正确的事。这是一个真正的问题，我们确实看到少数人有时会在财务激励或其他激励足够大时做出不太正确的事。我希望我知道如何解决人类激励的问题，我想。

You know, 99% of engineers and business people in Silicon Valley want to do the right thing. The people doing these, frankly, they're our friends. Maybe some people in this room right now. I think everyone kind of wants to do the right thing. I wish we could find a way when there are billions of dollars at stake to still always do the right thing. It is a real problem and we do see a small number of people that will sometimes do not quite the right thing when the financial incentives or some other incentives are big enough. I wish I knew how to solve the problem of human incentives, I guess.

太好了。让我们换个简单点的话题。美国政策。你早期就倡导 GPU 政策，也是我最早看到与国际科技公司合作的人之一。鉴于你的位置，你看到了联邦政策在 GPU 上的摇摆不定，关于他们所谓的‘觉醒系统’的行政命令，以及试图加速 AI 采用的行政命令和政策。如果你有 5 分钟和总统交谈，你会给出什么建议，以负责任地释放 AI 的力量造福所有美国人，并确保国家竞争力？

Great. Well, let's switch to something easier. US policy. You were early in advocating for policies on GPUs, and you were one of the first people I saw working with international technology companies. Given where you sit, you've seen the whipsawing on GPUs from federal policy, executive orders about what they describe as woke systems, and also executive orders and policies trying to accelerate AI adoption. If you had 5 minutes with the president, what advice would you give to responsibly unleash the power of AI to benefit all Americans and ensure national competitiveness?

是的，我非常担心美国在 AI 方面的国家竞争力。

Yeah, I'm really worried about US national competitiveness in AI.

我认为现任政府在一些方面做得不错。前任政府的一些 AI 安全思路实际上是安全剧场，由游说者推动，散布恐惧以试图实现监管俘获、反开源法规。如果你不想与开源竞争，就编造一堆关于 AI 危险的说法，试图通过压制性的许可。所以我认为现任政府对此似乎没什么耐心。这很好。

I think some things that the current administration has done well, I think the previous administration had some AI safety types of thinking that was really safety theater driven by lobbyists, fearmongering to try to create regulatory capture, anti-open source regulations. If you don't want to compete with open source, make up a bunch of stuff about the dangers of AI to try to get stifling licensing passed. So I think the current administration seems to have very little patience for that. That's good.

让我担心的事情：我们都是移民。很多学生是移民。如果我们在高技能移民上设置障碍，我真的很担心美国的竞争力。我 17 岁作为本科生来美国时，其实很懵懂。让学生来美国成长并成为高技能人才很重要。我还担心科学经费削减，对科学和 AI 的投资减少。大学有问题可以解决，但削弱国家执行科学的能力令人担忧。

Things that worry me: both of us are immigrants. Many students are immigrants. I really worry about American competitiveness if we make it harder for high-skilled immigration. When I came to the US as an undergrad at 17, I was pretty clueless. Letting students come and grow up to be higher skilled is important. I also worry about defunding of science, decreased investments in science and AI. Universities have issues we can fix, but diminishing the ability to execute science is concerning.

在国家政策方面，我担心我们对台积电的依赖。台积电亚利桑那是芯片制造商。有趣的是，中国最近禁止某些英伟达芯片进口，这表明中国正在摆脱对台湾台积电的依赖，而美国却变得严重依赖台湾制造。如果台湾发生任何事，无论是自然灾害还是人为事件，半导体生态的中断对美国的伤害可能大于中国。AI 半导体是一个瓶颈。另一个大瓶颈是能源。建设数据中心是将电力转化为智能。很多朋友在电厂许可上卡壳。能源容量是我担心的另一个瓶颈。

In terms of national policy, I worry about our reliance on TSMC. TSMC Arizona is a chip maker. It's interesting to see China ban certain Nvidia chip imports, signaling independence from TSMC in Taiwan while the US becomes heavily reliant on Taiwan manufacturing. If anything happens in Taiwan, disruption could hurt the US more than China. AI semiconductor is a bottleneck. The other big bottleneck is energy. Building a data center turns electricity into intelligence. Many friends are stuck in permitting for power plants. Energy capacity is another bottleneck I worry about.

当你想到中国和欧洲，两种截然不同的 AI 监管方式。对世界来说，正确的策略是什么？是一个中央主干带分支，还是许多不同的树组成模型联邦？

When you think about China and Europe, two radically different approaches to AI regulation. What's the right strategy for the world? Is it a central trunk of AI with branches or many different trees in a federation of models?

我认为我们需要多个分支。否则，就像移动生态有两个守门人，安卓和 iOS，除非他们允许你做某些事，否则你无法实验。我希望 AI 不会最终被少数守门人限制创新。过去一两年，中国在发布开放权重模型方面领先，任何人都可以免费下载。中国开放权重模型的累计采用量可能很快超过美国。美国的闭源模型仍然更好，但开放权重模型是 AI 供应链的关键部分。我认为我们国家在这方面投资不足。至于欧洲，我希望他们能觉醒并加快步伐。我听到欧洲监管者说他们想成为 AI 监管的领导者。这不是获得竞争优势的方式。更多刹车，更少油门，赢得比赛？

I think we need multiple branches. Otherwise, like the mobile ecosystem with two gatekeepers, Android and iOS, unless they let you do certain things, you can't experiment. I hope AI doesn't end up with a few gatekeepers limiting innovation. Over the past year or two, China has pulled ahead in releasing open weight models that anyone can download for free. Cumulative adoption of Chinese open weight models may soon surpass US ones. US closed models are still better, but open weight models are a key part of the AI supply chain. I think we aren't investing enough as a nation. As for Europe, I wish they would wake up and get going faster. I've heard European regulators say they want to be leaders in regulating AI. That's not how you gain competitive advantage. More brakes, less gas, win the race?

让我们转向未来。下一代学生倾向于解决什么问题？这对未来 24 个月意味着什么？

Let's turn to the future. What problems do next-generation students gravitate to? What does that tell you about the next 24 months?

在硅谷，我们大多数人都热爱 AI。但很多人可能低估了全国民众对 AI 的不信任。我感到紧迫，需要团结起来，讲述一个令人信服的故事，解释为什么 AI 对世界有益。我们对生产力提升感到兴奋，但一个担心失业的客服中心员工，或听到政客说 AI 会让工作消失的快餐店员工，会产生恐惧。要赢得人心，我们需要确保技术真正惠及每个人。AI 可以让个人更高效、更有生产力。我们需要提供工具，教人们使用，并改进它们。有了 AI，我们可以有 10 倍的营销人员、10 倍的分析师、10 倍的金融专业人士。但前面还有很多工作。我担心我们还没有赢得很多人的信任。

In Silicon Valley, most of us love AI. But many may underestimate the distrust people across the nation have for AI. I feel urgency to get our act together to tell a compelling narrative explaining why AI is good for the world. We're excited about productivity, but a contact center worker scared of losing their job or a fast food worker hearing a politician say AI will make their job go away creates fear. To win people over, we need to ensure our technology genuinely benefits everyone. AI can make individuals more effective and productive. We need to make tools available, teach people to use them, and improve them. With AI, we can have 10x marketers, 10x analysts, 10x finance professionals. But there's a lot of work ahead. I worry we haven't won the trust of many people.

你最喜欢用 AI 做什么？

What's your favorite way you use AI today?

天哪。也许我分享一个不太为人知的用法。大家都说，‘说吧。’

Gosh. Maybe I'll share one that is not widely known. Everyone's like, 'Go on.'

我把 AI 当作头脑风暴伙伴来用，甚至比我的朋友知道的还要多。诀窍在于，你得使用多个模型。

I use AI as a brainstorming companion much more than even my friends know. And the trick is, it turns out you have to use multiple models.

多个模型。帮朋友问问。明白了。

Multiple models. Asking for a friend. I see.

是的。我用 AI 来编程——我喜欢 Claude Code，也越来越常用 OpenAI Codex。但头脑风暴时，我用多个模型。事实证明 AI 很聪明，但输入上下文很困难。所以头脑风暴时，我发现很多不只是‘说些东西然后给我想法’，而是要确保有延展的对话，比如对话是‘给我三个想法’，然后我给出反馈。

Yeah. I use AI for coding—I love Claude Code and increasingly OpenAI Codex as well. But for brainstorming, I use multiple models. It turns out AI is very smart, but getting context in is difficult. So when brainstorming, I find a lot of it is not just 'say some stuff and give me ideas.' It's making sure you have an extended conversation where the conversation is 'give me three ideas,' then I'll give feedback.

语音还是文字？

Voice or text?

都可以。开车时用语音。坐着的时候，我发现开车时我跟 AI 聊很多。然后我会说，‘帮我总结一下。’然后发给团队，在开车时就把工作做了。

Either one. When I'm driving, voice. And when I'm sitting, I find when I'm driving, I talk to AI quite a lot. And then I'll say, 'Summarize it for me.' And I'll send it to my team and just get work done when I'm driving.

最后 10 秒，你希望人们用 AI 更多关注什么问题？

In the final 10 seconds, what's a problem that you would wish people would focus on more using AI?

我应该去动手构建。我认为你们每个人——现在都是构建的好时机。所以如果你从我信仰中带走一件事，那就是去动手构建。现在你可以构建很多以前不可能实现的酷东西。所以，构建、构建、构建。

I should go and build stuff. I think every one of you—this is a wonderful time to build. So if there's one thing you take away from what I believe in, just go and build stuff. There's so much cool stuff you can now build that just was not possible before. So build, build, build.

我认为这是完美的结束方式。构建、构建、构建。Andrew Ng，女士们先生们。感谢你的工作，Andrew。感谢你的研究。感谢你来到这里。

I think that is a perfect way to end on. Build, build, build. Andrew Ng, ladies and gentlemen. Thank you for your work, Andrew. Thank you for your research. Thank you for being here.

谢谢。

Thank you.

Andrew 和 DJ 的对话展示了 AI 是一个生态系统。美国要赢得 AI 竞赛，需要多管齐下。芯片和 AI 基础设施固然重要，但人才同样关键。我们必须继续吸引全球顶尖人才，为世界各地的学者和创新者提供机会，让他们在这里建立成功的生活和事业。

Andrew and DJ's conversation shows how AI is an ecosystem. For the US to win the AI race, it will take a multifaceted approach. It's chips and AI infrastructure, yes, but it's also talent. We must continue to attract top global talent and offer opportunities to scholars and innovators from all around the world to build a successful life and career here.

你可以在 Masters of Scale 的 YouTube 频道上找到这段完整视频以及更多峰会内容。

You can find the full video of this and more from the Summit Stage at the Masters of Scale YouTube channel.