Mark Jin 笔记

前言 | Introduction

中文： 本文是Andrew Ng在Startup School上的演讲内容，分享了他在AI Fund构建初创公司过程中学到的经验教训。AI Fund是一个风险工作室，平均每月构建一个初创公司。通过与企业家共同创立公司、编写代码、与客户交谈、设计功能、确定定价，他们积累了大量实际构建初创公司的经验。

English: This is Andrew Ng's speech at Startup School, sharing lessons learned from building startups at AI Fund. AI Fund is a venture studio that builds an average of one startup per month. Through co-founding startups, writing code, talking to customers, designing features, and determining pricing with entrepreneurs, they have gained extensive hands-on experience in building startups.

第一部分：AI技术栈与机遇 | Part 1: AI Stack and Opportunities

AI技术栈的层次结构 | The Hierarchy of AI Stack

中文： 我认为AI技术栈分为几个层次：最底层是半导体公司，然后是建立在其上的云服务和超大规模计算提供商，再上面是AI基础模型公司。尽管很多公关宣传和炒作都集中在这些技术层面，但事实上，几乎从定义上来说，最大的机会必须在应用层，因为我们实际上需要应用来产生更多收入，这样它们才能负担得起基础模型、云服务和半导体技术层的费用。

English: I think of the AI stack as having several layers: at the lowest level are semiconductor companies, then clouds and hyperscalers built on top of that, and AI foundation model companies built on top of that. Even though a lot of PR excitement and hype has been on these technology layers, it turns out that almost by definition, the biggest opportunities have to be at the application layer because we actually need applications to generate even more revenue so that they can afford to pay the foundation, cloud, and semiconductor technology layers.

智能体AI的兴起 | The Rise of Agentic AI

中文： 如果你问我AI领域最重要的技术趋势是什么，我会说是智能体AI的兴起。大约一年半前，当我开始四处演讲试图说服人们AI智能体可能会成为现实时，我没有意识到去年夏天左右，一群营销人员会抓住这个术语，把它当作贴纸贴在所有能看到的东西上，这几乎让它失去了一些意义。

English: If you ask me what's the most important tech trend in AI, I would say it's the rise of agentic AI. About a year and a half ago, when I started to go around and give talks to try to convince people that AI agents might be a thing, I did not realize that around last summer, a bunch of marketers would get a hold of this term and use it as a sticker and slap it on everything in sight, which made it almost lose some of its meaning.

中文： 我们使用大语言模型的方式通常是提示它输出内容。我们让大语言模型输出内容的方式就像你去找一个人，或者在这种情况下是AI，要求它为你写一篇文章，从第一个词到最后一个词一气呵成，永远不使用退格键。人类在被迫以这种线性顺序写作时，并不能发挥最佳写作水平。

English: The way a lot of us use LLMs is to prompt it to output something. The way we have an LLM output something is as if you're going to a human, or in this case an AI, and asking it to please type out an essay for you by writing from the first word to the last word all in one go without ever using backspace. Humans don't do our best writing being forced to type in this linear order.

中文： 通过智能体工作流，我们可以让AI系统首先写一个文章大纲，然后如果需要的话进行一些网络研究，获取一些网页放入大语言模型的上下文中，然后写初稿，然后阅读初稿并批评它，修改它，等等。这样我们最终得到一个迭代工作流，你的模型进行一些思考和研究，进行一些修改，回去做更多思考，通过多次循环，虽然速度较慢，但能提供更好的工作成果。

English: With agentic workflows, we can go to an AI system and ask it to please first write an essay outline, then do some web research if it needs to, and fetch some web pages to put in the LLM context, then write the first draft, then read the first draft and critique it, and revise it, and so on. We end up with this iterative workflow where your model does some thinking and research, does some revision, goes back to do more thinking, and by going around this loop many times, it is slower, but it delivers a much better work product.

第二部分：构建初创公司的最佳实践 | Part 2: Best Practices for Building Startups

专注于具体想法 | Focus on Concrete Ideas

中文： 在AI Fund，我们只专注于具体想法。对我来说，一个具体的想法，一个具体的产品想法，是指定了足够细节的想法，工程师可以去构建它。例如，如果你说"让我们使用AI来优化医疗资产"，这实际上不是一个具体想法，太模糊了。如果你告诉我写软件来使用AI优化医疗资产，不同的工程师会做完全不同的事情。

English: At AI Fund, we only focus on working on concrete ideas. To me, a concrete idea, a concrete product idea, is one that specifies enough detail that an engineer can go and build it. For example, if you say, "let's use AI to optimize healthcare assets," that's actually not a concrete idea. It's too vague. If you tell me to write software to use AI to optimize healthcare assets, different engineers would do totally different things.

中文： 相比之下，如果你有一个具体想法，比如"让我们写软件让医院患者在线预订MRI机器时段以优化使用率"。我不知道这是一个好想法还是坏想法，实际上已经有企业在做这个了。但它是具体的，这意味着工程师可以快速构建它。如果这是个好想法，你会发现；如果不是好想法，你也会发现。但拥有具体想法能让你获得速度。

English: In contrast, if you had a concrete idea, like "let's write software to let hospitals, let patients book MRI machine slots online to optimize usage." I don't know if this is a good or bad concrete idea. It's actually a business already doing this. But it is concrete and that means engineers can build it quickly. If it's a good idea, you find out; if it's not a good idea, you will find out. But having concrete ideas buys you speed.

快速构建-反馈循环 | Rapid Build-Feedback Loop

中文： 为了更快发展，我经常思考的另一件事是构建-反馈循环，当涉及到我们如何使用AI编码辅助构建时，这正在快速变化。当你构建很多应用程序时，最大的风险之一是客户接受度。很多初创公司的困难不是因为我们无法构建想要构建的东西，而是因为我们构建了某些东西，结果发现没人关心。

English: In order to go faster, the other thing I often think about is the build-feedback loop, which is rapidly changing when it comes to how we build with AI coding assistance. When you're building a lot of applications, one of the biggest risks is customer acceptance. A lot of startups struggle not because we can't build whatever we want to build, but because we build something and it turns out nobody cares.

中文： 事实证明，通过AI编码辅助，快速工程变得可能，这在以前是不可能的。工程速度正在快速提升，工程成本也在快速下降。这改变了我们推动初创公司围绕这个循环的机制。

English: It turns out that with AI coding assistance, rapid engineering is becoming possible in a way that just was not possible. The speed of engineering is going up rapidly and the cost of engineering is also going down rapidly. This changes the mechanisms by which we drive startups around this loop.

原型开发与生产代码的区别 | Prototyping vs Production Code

中文： 当我思考我所做的软件时，我可能把它分为两大类。有时我构建快速粗糙的原型来测试想法，比如构建新的客户服务聊天机器人，让AI处理法律文档等。构建快速粗糙的原型来看看我们认为它是否有效。另一种类型的软件是编写、维护生产软件，维护遗留软件，这些大规模的生产就绪代码库。

English: When I think about the software that I do, I maybe put it into two major buckets. Sometimes I've built quick and dirty prototypes to test an idea. Say, build a new customer service chatbot, let's build AI to process legal documents, whatever. Build a quick and dirty prototype to see if we think it works. The other type of software is write, maintain production software, maintain legacy software, with these massive production-ready code bases.

中文： 根据你信任的分析师报告，很难找到关于这方面的严格数据。在编写生产质量代码时，也许我们使用AI系统能快30%到50%。很难找到严格的数字。但在构建快速粗糙原型方面，我们不是快50%，我认为我们轻松快10倍，可能远超过10倍。

English: Depending on which analyst report you trust, it's been hard to find very rigorous data on this. When writing production quality code, maybe we're 30 to 50% faster with AI systems. Hard to find a rigorous number. But in terms of building quick and dirty prototypes, we're not 50% faster. I think we're easily 10 times faster, maybe much more than 10 times faster.

第三部分：AI编码辅助的演进 | Part 3: Evolution of AI Coding Assistance

编码辅助工具的发展 | Development of Coding Assistance Tools

中文： 在AI辅助编码领域，我认为三四年前是代码自动补全，由GitHub Copilot推广。然后是Cursor和Windsurf这一代AI增强的IDE。我们大量使用Windsurf和Cursor。然后从大约六七个月前开始，出现了新一代高度智能化的编码助手，包括大量使用O3进行编码。Claude Code非常棒。

English: In terms of the AI assistance coding landscape, I think three, four years ago was Code Autocomplete, popularized by GitHub Copilot. And then there was the Cursor Windsurf generation of AI-enabled IDEs. We use Windsurf and Cursor quite a lot. And then starting, I don't know, six, seven months ago, there started to be this new generation of highly agentic coding assistants, including using O3 a lot for coding. Claude Code is fantastic.

中文： 有趣的是，如果你甚至落后半代或一代，与使用最新工具相比，实际上会产生很大差异。我发现我的团队现在采用与三个月甚至六个月前完全不同的软件工程方法。一个令人惊讶的事情是，我们习惯于将代码视为非常有价值的工件，因为它很难创建。但由于软件工程成本正在下降，代码不再像以前那样是有价值的工件。

English: The interesting thing is if you're even half a generation or one generation behind, it actually makes a big difference compared to if you're on top of the latest tools. I find my team is taking really different approaches to software engineering now compared to even three or six months ago. One surprising thing is we're used to thinking of code as this really valuable artifact because it's so hard to create. But because the cost of software engineering is going down, code is much less of a valuable artifact as it used to be.

双向门与单向门决策 | Two-way Door vs One-way Door Decisions

中文： 你们中的一些人可能听说过Jeff Bezos关于双向门与单向门的术语。双向门是你可以做出的决定，如果你改变主意，可以相对便宜地回来、逆转它。而单向门是你做出决定后，如果你改变主意，逆转它的成本非常高或非常困难。

English: Some of you may have heard of Jeff Bezos' terminology of a two-way door versus a one-way door. A two-way door is a decision that you can make if you change your mind, come back out, reverse it relatively cheaply. Whereas a one-way door is you make a decision and if you change your mind, it's very costly or very difficult to reverse.

中文： 选择软件架构和技术栈曾经是单向门。一旦你建立在某个技术栈上，设置了数据库架构，就很难改变它。我不想说它完全是双向门，但我发现我的团队更经常在某个技术栈上构建，一周后改变主意，让我们丢弃代码库，在新技术栈上从头重做。

English: Choosing the software architecture of your tech stack used to be a one-way door. Once you're built on top of a certain tech stack, you set a database schema, really hard to change it. I don't want to say it's totally a two-way door, but I find that my team will more often build on a certain tech stack a week later, change your mind, let's throw the code base away and redo it from scratch on the new tech stack.

第四部分：赋能每个人构建AI | Part 4: Empowering Everyone to Build AI

学习编程的重要性 | The Importance of Learning to Code

中文： 我觉得现在是赋能每个人构建AI的好时机。在过去一年中，很多人建议其他人不要学习编程，理由是AI会自动化它。我认为我们回头看时会发现这是有史以来最糟糕的职业建议，因为当更好的工具让软件工程变得更容易时，应该有更多人去做，而不是更少。

English: I feel like it's actually a good time to empower everyone to build AI. Over the last year, a bunch of people have advised others not to learn to code on the grounds that AI will automate it. I think we'll look back on this as some of the worst career advice ever given because as better tools make software engineering easier, more people should do it, not fewer.

中文： 我有一个有争议的观点，我认为现在是每个工作角色的人都学习编程的时候了。事实上，在我的团队中，我的CFO、人才主管、招聘人员、前台人员，他们都知道如何编程。我实际上看到他们所有人在所有工作职能上表现得更好，因为他们会编程。

English: I have a controversial opinion which is I think it's time for everyone of every job role to learn to code. And in fact, on my team, my CFO, my head of talent, my recruiters, my front desk person, all of them know how to code. And I actually see all of them performing better at all of their job functions because they can code.

产品管理成为瓶颈 | Product Management Becomes the Bottleneck

中文： 随着软件工程变得更快，我看到的另一个有趣动态是产品管理工作——获取用户反馈、决定构建什么功能——越来越成为瓶颈。在过去一年中，我的更多团队开始抱怨他们在产品工程和设计方面遇到瓶颈，因为工程师变得如此之快。

English: With software engineering becoming much faster, the other interesting dynamic I'm seeing is that the product management work, getting user feedback, deciding what features to build, that is increasingly the bottleneck. Over the last year, a lot more of my teams have started to complain that they're bottlenecked on product engineering and design because the engineers have gotten so much faster.

中文： 也许四五年前，硅谷有这些稍微可疑的经验法则，但仍然是经验法则。我们会有1个产品经理对4个工程师或1个产品经理对7个工程师的比例。随着工程师变得更快，我没有看到产品管理工作以工程相同的速度变快，看到这个比例在转变。

English: Maybe four or five years ago, Silicon Valley used to have these slightly suspicious rules of thumb, but nonetheless rules of thumb. We'll have 1 PM to 4 engineers or 1 PM to 7 engineers ratio. With engineers becoming much faster, I don't see product management work becoming faster at the same speed that engineering is, seeing this ratio shift.

第五部分：获取快速反馈的策略 | Part 5: Strategies for Getting Rapid Feedback

反馈策略组合 | Portfolio of Feedback Tactics

中文： 我发现对初创公司领导者来说重要的是，因为工程进展如此之快，如果你有获取快速反馈的好策略来更快地塑造你对构建什么的看法，这也能帮助你更快发展。我将介绍一系列获取产品反馈的策略，从更快但可能不太准确的，到更慢但更准确的策略。

English: I find as startup leaders, because engineering is going so fast, if you have good tactics for getting rapid feedback to shape your perspective on what to build faster, that helps you get faster as well. I'm going to go through a portfolio of tactics for getting product feedback, from the faster, maybe less accurate, to the slower, more accurate tactics.

中文： 获取反馈最快的策略是自己看产品，凭直觉判断。如果你是主题专家，如果你知道自己在做什么，这实际上出奇地好。稍微慢一点的是去问三个朋友或队友获取反馈。再慢一点的是问三到十个陌生人获取反馈。

English: The fastest tactic for getting feedback is look at the product yourself and just go by your gut. If you're a subject matter expert, if you know what you're doing, this is actually surprisingly good. A little bit slower is go ask three friends or teammates to get feedback. A little bit slower is ask three to ten strangers for feedback.

中文： 事实证明，当我构建产品时，我认为我学到的最重要技能之一是如何坐在咖啡店里，如何坐在那里。当我旅行时，我经常坐在酒店大堂。事实证明，我学会了发现人流量大的地方，非常尊重地抓住陌生人，询问他们对我正在构建的任何东西的反馈。

English: It turns out when I build products, one of the most important skills I think I learned was how to sit in the coffee shop, how to sit in there. When I travel, I often sit in the hotel lobby. It turns out I learn to spot places of high foot traffic and very respectfully grab strangers and ask them for feedback on whatever I'm building.

第六部分：理解AI的竞争优势 | Part 6: Understanding AI as Competitive Advantage

AI知识的稀缺性 | Scarcity of AI Knowledge

中文： 我发现理解AI实际上让你走得更快。事实证明，当涉及到成熟技术，如移动技术时，很多人长期拥有智能手机。我们大致知道移动应用能做什么。如果你看成熟的工作角色，如销售、营销、人力资源、法律，它们都非常重要，都非常困难，但有足够多的营销人员做营销足够长时间，营销策略在去年没有太大变化。

English: I've seen that understanding AI actually makes you go faster. It turns out that when it comes to mature technology, like mobile, many people have had smartphones for a long time. We kind of know what a mobile app can do. If you look at mature job roles, like sales, marketing, HR, legal, they're all really important, all really difficult, but there are enough marketers that have done marketing for long enough and the marketing tactics haven't changed that much in the last year.

中文： 但AI是新兴技术。因此，如何很好地使用AI的知识并不广泛。所以真正理解AI的团队确实比不理解的团队有优势。而如果你有人力资源问题，你可能可以找到知道如何做好的人。但如果是AI问题，知道如何实际做到这一点，可能让你领先于其他公司。

English: But AI is emerging technology. And so the knowledge of how to do AI really well is not widespread. And so teams that actually get it, that understand AI, do have an advantage over teams that don't. Whereas if you have an HR problem, you can find someone that knows how to do it well, probably. But if an AI problem, knowing how to actually do that, could put you ahead of other companies.

技术决策的重要性 | Importance of Technical Decisions

中文： 像客户服务聊天机器人能达到什么准确度？你应该提示还是微调使用H&D工作流？如何让语音应用达到低延迟？有很多这样的决策，如果你做出正确的技术决策，你可以在几天内解决问题。如果做出错误的技术决策，你可能会在盲巷里追逐三个月。

English: Things like, what accuracy can you get for a customer service chatbot? Should you prompt or fine-tune using H&D workflow? How to get a voice app to low latency? There are a lot of these decisions that if you make the right technical decision, you can solve the problem in a couple of days. If you make the wrong technical decision, you could chase a blind alley for three months.

中文： 我对一件事感到惊讶，事实证明，如果你有两个可能的架构决策，这是一位信息。感觉如果你不知道正确答案，最多你会慢两倍。但我在实践中看到的是，如果你翻转错误的位，你不是慢两倍，你会花费大约10倍的时间追逐盲巷，这就是为什么我认为拥有正确的技术判断真的让初创公司走得更快。

English: One thing I was surprised by, it turns out if you have two possible architecture decisions, it's one bit of information. It feels like if you don't know the right answer, at most you're twice as slow. But what I see in practice, if you flip the wrong bit, you're not twice as slow. You spend like 10 times longer chasing a blind alley, which is why I think having the right technical judgment really makes startups go so much faster.

第七部分：AI构建模块的组合效应 | Part 7: Combinatorial Effects of AI Building Blocks

丰富的AI工具生态 | Rich Ecosystem of AI Tools

中文： 在过去两年中，我们有了大量精彩的生成式AI工具或生成式AI构建模块。部分列表包括：提示、智能体工作流、评估、护栏、RAG、语音应用、异步编程、大量ETL、嵌入、微调、图数据库、如何集成计算机使用、MCP、VZ模型。有一个长长的精彩构建模块列表，你可以快速组合来构建甚至一年前地球上没有人能构建的软件。

English: Over the last two years, we have just had a ton of wonderful Gen AI tools or Gen AI building blocks. Partial list includes: prompting, agentic workflows, evals, guardrails, RAG, voice app, async programming, lots of ETL, embeddings, fine tuning, graph DB, how to integrate computer use, MCP, VZ models. There's a long and wonderful list of building blocks that you can quickly combine to build software that no one on the planet could have built even a year ago.

中文： 当我了解这些构建模块时，这实际上是我脑海中的画面。如果你拥有一个构建模块，比如你有一个基本的白色构建模块，你可以构建一些很酷的东西。也许你知道如何提示，所以你有一个构建模块，你可以构建一些令人惊叹的东西。但如果你得到第二个构建模块，比如你也知道如何构建聊天，得到一些红色构建模块，也许一个小黄色的，更有趣。

English: When I learned about these building blocks, this is actually a picture that I have in mind. If you own one building block, like you have a basic white building block, you can build some cool stuff. Maybe you know how to prompt. So you have one building block, you can build some amazing stuff. But if you get a second building block, like you also know how to build chat, get a few red building blocks, maybe a little yellow one, more interesting.

中文： 获得更多构建模块，获得更多构建模块，很快，你可以将它们组合成的东西数量呈组合式或指数式增长。因此，了解所有这些精彩的构建模块让你能够将它们组合成更丰富的组合。

English: Get more building blocks, get more building blocks, and very rapidly, the number of things you can combine them into grows combinatorially or grows exponentially. And so knowing all these wonderful building blocks lets you combine them in a much richer combination.

第八部分：问答环节精选 | Part 8: Selected Q&A

关于AGI炒作的看法 | Views on AGI Hype

问题： 随着AI的发展，您认为人类开发工具更重要，还是学会更好地使用工具更重要？在智能正在民主化的世界中，我们如何定位自己保持重要性？

Question: As AI advances, do you think it's more important for humans to develop the tools or learn how to use the tools better? How can we position ourselves to remain essential in a world where intelligence is becoming democratized?

中文回答： 我觉得AGI被过度炒作了。因此，在很长一段时间内，人类能做而AI不能做的事情还有很多。我认为在未来，最强大的人是那些能让计算机完全按照你想要的方式工作的人。因此，我认为跟上工具发展，我们中的一些人有时会构建工具，但还有很多其他人构建的工具我们可以直接使用。知道如何使用AI让计算机做你想要它做的事情的人会比不知道的人强大得多。

English Answer: I feel like AGI has been overhyped. And so for a long time, there'll be a lot of things that humans can do that AI cannot. And I think in the future, the people that are most powerful are the people that can make computers do exactly what you want it to do. And so I think staying on top of the tools, some of us will build tools sometimes, but there are a lot of other tools that others will build that we can just use. People that know how to use AI to get computers to do what you want it to do will be much more powerful than people that don't.

关于AI安全的思考 | Thoughts on AI Safety

问题： AI有很大的积极潜力，但也有很多潜在的负面后果，比如加剧经济不平等等。我们作为AI构建者应该如何平衡产品构建与AI产品潜在的社会负面影响？

Question: AI has a lot of great potential for good, but there's also a lot of potential for bad consequences as well, such as exacerbating economic inequality. How do you think us as AI builders should balance our product building with also the potential societal downsides of some AI products?

中文回答： 看看你的内心。如果从根本上你正在构建的东西，如果你不认为它会让人们整体上过得更好，就不要做。我知道这听起来简单，但在当下做到这一点真的很难。在AI Fund，我们已经终止了多个项目，不是基于财务理由，而是基于道德理由，有多个项目我们看到经济案例非常可靠，但我们说，你知道吗，我们不希望这个存在于世界上，我们就基于这个理由终止了它。

English Answer: Look in your heart. And if fundamentally what you're building, if you don't think it'll make people with large better off, don't do it. I know it sounds simple, but that's really hard to do in the moment. At AI Fund, we've killed multiple projects, not on financial grounds, but on ethical grounds, where there are multiple projects and we looked at the economic case is very solid, but we said, you know what, we don't want this to exist in the world and we just killed it on that basis.

关于商业护城河的思考 | Thoughts on Business Moats

问题： 作为有抱负的创始人，在一个任何东西都可能在一天内被颠覆的世界中，我们应该如何思考商业？无论你有什么伟大的护城河、产品或功能，都可以通过Vibe编码竞争对手在几小时内复制。

Question: As aspiring founders, how should we be thinking about business in the world where anything can be disrupted in a day? Whatever great moat, product, or feature you have can be replicated with Vibe Coding competitors in basically hours.

中文回答： 事实证明，当你开始一项业务时，有很多事情需要担心。我担心的首要问题是，你是否在构建用户喜爱的产品？事实证明，当你构建业务时，有很多事情需要考虑：进入市场渠道、竞争对手、技术护城河，所有这些都很重要。但如果我要专注于一件事，那就是，你是否在构建用户真正想要的产品？在你解决这个问题之前，很难构建有价值的业务。

English Answer: It turns out when you start a business, there are a lot of things to worry about. The number one thing I worry about is, are you building a product that users love? It turns out that when you build a business, there are lots of things to think about: go-to-market channel, competitors, technology moat, all that is important. But if I were to have a singular focus on one thing, it is, are you building a product that users really want? Until you solve that, it's very difficult to build a valuable business.

中文回答（续）： 在你解决了这个问题之后，其他问题确实会发挥作用。你有渠道接触客户吗？长期定价是什么？你的护城河是什么？我发现护城河实际上往往被过度炒作。我发现更多企业往往从产品开始，然后最终演变成护城河。但消费产品品牌在某种程度上更具防御性。如果你有很大的动力，追赶你就变得更困难。

English Answer (continued): After you solve that, the other questions do come into play. Do you have a channel to get to customers? What is pricing long-term? What is your moat? I find that moats tend to be overhyped, actually. I find that more businesses tend to start up with a product and then evolve eventually into a moat. But consumer products brand is somewhat more defensible. And if you have a lot of momentum, it becomes harder to catch you.

关于教育AI的未来 | Future of AI in Education

问题： 在教育和AI的世界中，主要有两种范式。一种是AI可以让教师更有生产力，自动化评分和作业。另一种思路是每个学生都会有个人导师。您如何看待这两种范式的融合？未来五年教育会是什么样子？

Question: In the world of education and AI, there are two paradigms mostly. One is AI can make teachers more productive, automating grading and homework. Another school of thought is that there'll be personal tutors for every student. How do you see these two paradigms converge? And how would education look like in the next five years?

中文回答： 我认为每个人都感觉到教育技术即将发生变化。但我不认为颠覆已经到来。我认为很多人在尝试不同的东西。Coursera有Coursera Coach，效果很好。Deep Learning更专注于教授AI，也有一些内置聊天机器人。很多团队都在尝试自动评分。我认为教育将是超个性化的。但具体的工作流程是什么，是头像，还是文本聊天机器人，我觉得几年前的炒作说我们很快就会有AGI，一切都会变得如此简单，那是炒作。

English Answer: I think everyone feels like a change is coming in EdTech. But I don't think the disruption is here yet. I think a lot of people are experimenting at different things. Coursera has Coursera Coach, which actually works really well. Deep Learning is more focused on teaching AI, also has some built-in chatbots. A lot of teams have experimented in auto-grading. I do think education will be hyper-personalized. But what's the workflow, is it an avatar, is it a text chatbot, I think the hype from a couple of years ago that we'll have AGI soon and it'll be all so easy, that was hype.

中文回答（续）： 现实是，工作是复杂的。教师、学生、人们做的工作流程真的很复杂。在接下来的十年中，我们将研究需要完成的工作，并找出如何将其映射到智能体工作流程。教育是这种映射仍在进行中的部门之一，但还不够成熟，无法明确最终状态。所以我认为我们都应该继续努力。

English Answer (continued): The reality is, work is complex. Teachers, students, people do really complex workflows. And for the next decade, we'll be looking at the work that needs to be done and figuring out how to map it to agentic workflows. And education is one of the sectors where this mapping is still underway, but it's not yet mature enough to the point where the end state is clear. So I think we should all just keep working on it.

关于计算资源的未来 | Future of Computing Resources

问题： 随着我们向更强大的AI发展，您认为计算资源的发展方向如何？我们看到有人说要把GPU送到太空，有人谈论核电数据中心。您对此有什么看法？

Question: As we move towards more powerful AI, where do you think compute is heading? We see people saying let's ship GPUs to space, some people talking about nuclear power data centers. What do you think about it?

中文回答： 有一个框架可以用来决定什么是炒作，什么不是炒作。事实证明，在过去，因为AI如此新颖，少数公司可以说几乎任何话而没有人事实核查，因为技术不被理解。所以我的一个心理过滤器是，有某些炒作叙述让这些企业看起来更强大，这被放大了。

English Answer: There's one framework you can use for deciding what's hype and what's not hype. It turns out over the last period, because AI was so new, a handful of companies got away with saying almost anything without anyone fact-checking them because the technology was not understood. So one of my mental filters is there's certain hype narratives that make these businesses look more powerful that's been amplified.

中文回答（续）： 例如，这种想法认为AI如此强大，我们可能意外导致人类灭绝。这简直是荒谬的。但这是一个炒作叙述，让某些企业看起来更强大，它被放大了，实际上帮助了某些企业的筹资目标。AI需要如此多的电力，只有核电才足够好。风能、太阳能这些都不够。这不是真的。所以我认为很多这些太空中的GPU，我不知道，去做吧。我认为我们在地面GPU方面仍有很大发展空间。

English Answer (continued): For example, this idea that AI is so powerful, we might accidentally lead to human extinction. That's just ridiculous. But it is a hype narrative that made certain businesses look more powerful and it got ramped up and actually helped certain businesses' fundraising goals. AI needs so much electricity, only nuclear power is good enough for that. Wind, solar stuff, this is not true. So I think a lot of these GPUs in space, I don't know, go for it. I think we have a lot of room to run still for terrestrial GPUs.

结语 | Conclusion

中文： Andrew Ng在这次演讲中分享了许多关于在AI时代构建初创公司的宝贵见解。从专注于具体想法、利用AI编码辅助加速开发，到理解AI技术栈的机遇分布，再到负责任地构建AI产品，这些经验对于想要在AI领域创业的人来说都是非常实用的指导。

English: In this speech, Andrew Ng shared many valuable insights about building startups in the AI era. From focusing on concrete ideas and leveraging AI coding assistance to accelerate development, to understanding the opportunity distribution in the AI technology stack, and building AI products responsibly, these experiences provide very practical guidance for those who want to start businesses in the AI field.

关键要点 | Key Takeaways

中文：

专注具体想法：模糊的想法无法快速执行，具体的想法才能带来速度优势
拥抱AI工具：AI编码辅助正在革命性地改变软件开发速度
快速反馈循环：建立有效的用户反馈机制是成功的关键
技术判断力：在AI领域，正确的技术决策可以节省大量时间
负责任创新：在追求速度的同时，不能忽视道德和社会责任
全员编程：每个人都应该学会编程，以更好地利用AI工具
产品优先：专注于构建用户真正想要的产品，而不是过度担心竞争

English:

Focus on concrete ideas: Vague ideas cannot be executed quickly; concrete ideas bring speed advantages
Embrace AI tools: AI coding assistance is revolutionarily changing software development speed
Rapid feedback loops: Establishing effective user feedback mechanisms is key to success
Technical judgment: In the AI field, correct technical decisions can save significant time
Responsible innovation: While pursuing speed, we cannot ignore moral and social responsibilities
Universal coding: Everyone should learn to code to better utilize AI tools
Product-first: Focus on building products users really want, rather than over-worrying about competition

对创业者的启示 | Insights for Entrepreneurs

中文： 在AI快速发展的时代，创业者需要：

保持对新技术的敏感度和学习能力
建立快速迭代和验证的能力
培养正确的技术判断力
始终以用户需求为中心
在创新的同时承担社会责任

English: In the era of rapid AI development, entrepreneurs need to:

Maintain sensitivity and learning ability for new technologies
Build capabilities for rapid iteration and validation
Develop correct technical judgment
Always center on user needs
Take social responsibility while innovating

本文整理自Andrew Ng在Startup School的演讲，旨在为中文读者提供完整的中英对照版本。通过深入理解这些实践经验，希望能帮助更多创业者在AI时代取得成功。

This article is compiled from Andrew Ng's speech at Startup School, aiming to provide Chinese readers with a complete bilingual version. Through deep understanding of these practical experiences, we hope to help more entrepreneurs succeed in the AI era.