模块 #1：理解生成式 AI 的力量及其驾驭之道

我的理解

本课以现场问答和投票形式开场，揭示了一个核心矛盾：生成式 AI 入门极易，但真正用好却极难，这是因为 AI 本质上是一场范式转变，而非普通工具升级。讲师援引比尔·盖茨的判断，将生成式 AI 与 GUI 并列为他一生中见过的仅有两项革命性技术，由此说明范式转变的量级。这场转变带来三重核心挑战：需要「去学」（unlearn）旧有隐性假设（如程序员遇到问题就想微调模型）、目标模糊（没有公认的「用好 AI」标准），以及对被淘汰的深层恐惧。课程的关键洞察是：当构建成本趋近于零，人的核心价值将从执行重复劳动转向品味与判断力。理解范式转变的本质，是真正掌握生成式 AI 的根本前提。

原文

Lesson 1 of 68 视频（英文） - 模块 #1：理解生成式 AI 的力量及其驾驭之道 / Video - Module #1: Understand the power of GenAI and how to yield it 麦克，但是你能吗？Mic and but could you. Hide transcript Transcript 00:00 麦克，但是你能吗？Mic and but could you. 00:03 我把我们的问题翻译成英文 I’ll translate our question into English 00:05 因为并非所有人都说中文。because not everyone is Chinese speaking. 00:08 那么，So the question is 00:10 当构建成本趋于零时，when the cost of building converges to zero, 00:14 匠人该如何维持自身价值呢？how can a craftsman keep their value? 00:20 没错。Right. 00:22 手工艺如何保持其价值？How can craftsmanship keep its value? 00:28 我觉得那就是品味的价值所在。I think that is the value of taste. 00:31 Yan 刚刚写了一篇关于这个的文章。Yan just wrote an article about this. 00:34 我们如同工匠或精通某种技艺之人，We have like as a craftsman or someone who master a craft, 00:39 无论是在数据、科学还是设计领域，whether it’s data, science or design, 00:44 我们的真正的价值在于我们的品味与判断力。our value is actually our taste and judgment. 00:51 但我们靠做那些枯燥 But we are paid by doing the tedious work, 00:54 重复的工作来赚钱。the repetitive work. 00:58 所以，构建过程其实去除了重复性工作 So building actually takes away the repetitive part 01:03 我们在品味上的价值将变得愈发重要。and our value in taste is just going to be more important. 01:13 好，我可以用一个非常简单的比喻来说明。All right, I can give a very simple analogy. 01:16 这并不是第一次发生。This is not the first time it’s happening. 01:18 想想汽车吧。Think about cars. 01:19 当造车越来越容易，When building cars is easier and easier, 01:21 高档品牌该如何维持其价值？how can the high end brands still keep their value? 01:29 好的。All right. 01:30 嗯，才两分钟。Two minutes into the yeah. 01:33 等会儿我就开始发送请求了。Time I’ll start sent a request. 01:38 嗯，已经批准了。Yeah, already approved. 01:40 好的。Okay. 01:48 好的。Okay. 01:49 所以，大家可以马上开始问卷调查，So we can start with the survey right away 01:52 在等待其他人的同时，while we are waiting for others, 01:54 大家可以去 Slido 输入这个号码，you can go to slido and type this number 01:58 或者扫描二维码。or you can scan QR code. 02:00 如果你用的是笔记本或台式电脑，If you are using a laptop or desktop, 02:05 可以直接访问 Slido，you can just go to Slido 02:09 输入号码 997-0750，and type this number 997-0750 02:13 我们来做几个选择题，and we’ll go to a few multiple choice questions 02:17 了解一下现场情况。just so we can get a sense of the room. 02:23 好的，我们开始投票吧。Okay, I’ll start the poll. 02:27 第一个问题。First question. 02:28 你试过我们预先准备的 Have you tried our pre work 02:30 那个紧急 WBR 请求吗？the urgent WBR request? 02:35 是的。Yes. 02:36 很少。Little. 02:36 有很多。There’s a lot. 02:46 如果你不知道那是什么。If you don’t know what it is. 02:50 这就是 the。This is the. 02:52 这就是我所讲的。This is what I’m talking about. 02:56 我在公告里强烈建议大家 In my announcement I strongly encouraged everyone 03:00 提前试用一下这个。to try this ahead of time. 03:04 不过没问题。But it’s okay. 03:05 我们将一步步带大家走完整个过程。We’ll walk through the entire thing. 03:09 10 票。10 votes. 03:10 太好了。Great. 03:10 太好了，太好了。Great, great. 03:14 我其实挺惊喜的 I’m actually pleasantly surprised 03:15 因为我记得上一期只有大约 20%的人尝试过这个 because I remember last cohort only maybe 20% try this 03:21 而 70%到 80%的人根本没试。and about like 70, 80% not at all. 03:25 做得很好。So good job. 03:29 下一个。Next one. 03:31 你用过以下这些 AI 编程工具吗？Have you used the following AI coding tools 03:34 或者我们叫它们代理式编程工具。or let’s call it agentic coding tools. 03:37 无所谓。Whatever. 03:37 Cursor。Cursor. 03:38 Claude Code、Codex、CLI、Gemini、Trae、Copilot。Claude Code, Codex, CLI, Gemini, Trae, Copilot. 03:44 以上皆非。None of the above. 03:48 很好。Nice. 03:49 很好。Nice. 03:50 很多人试了很多方法。A lot of people tried a lot of things. 03:52 哇哦！Wow. 03:52 好的。Okay. 03:54 作为背景，Just as a context, 03:56 这应该是第三期或第四期学员。I think this is the third cohort or fourth cohort. 04:01 我们已经做到了这一点。We did this. 04:02 我们大概是从 Cohort 8 或 Cohort 9 开始做这个 We started doing this probably in Cohort 8 or Cohort 9 04:08 大约半年前的事。and half a year ago. 04:11 我觉得有一半的学生 I think half of the students 04:13 从来没有试过这些东西。never tried any of this. 04:17 现在大概只有 10%或 17% And now only about 10% or 17% 04:21 ，但还是有一些。but still some. 04:24 没错。Right. 04:26 好的。Okay. 04:28 Claude Code 也是最受欢迎的。And Claude Code is the most popular too. 04:34 嗯。Yeah. 04:34 你感到惊讶吗？Are you surprised? 04:37 既是，也不是。Both yes and no. 04:39 我知道 Claude Code 是 I am aware that Claude Code is 04:41 最受欢迎的工具之一。among the most popular tool. 04:42 我感到惊讶是因为 I’m surprised in the sense that 04:43 我不认为它是一个很好的工具。I don’t think it’s a very good tool. 04:46 什么？What? 04:48 这是一个非常不得人心的看法。That’s a very unpopular opinion. 04:51 你想再多说一点吗？Do you want to elaborate? 04:55 我来试着解释一下这两句话。I’ll try to explain two sentences. 04:56 第一个是从功能等价性的角度来看。The first is from a feature parity perspective. 04:59 Claude Code 提供的大部分功能 Nearly all of the features Claude Code provides 05:02 Cursor 或 Codex 也都提供了。is also provided in either Cursor or Codex. 05:06 从开放性的角度来看，Claude Code。From an openness perspective, Claude Code. 05:10 它运行在定制的二进制文件上 It runs on a customized binary 05:14 带有 a。with a. 05:19 我说不是公开的代码。I would say not public code. 05:21 但 Codex 完全是开源的。But Codex is totally open source. 05:24 这就是两个原因。So those are the two reasons. 05:26 我写了一篇分析这个的文章。I wrote an article analyzing this. 05:30 我把链接贴在这里。I’ll post paste the link here. 05:34 有时候人们会谈到 harness，对不对？Sometimes people talk about the harness, right? 05:36 他们说 And they say like the harness, 05:37 Claude Code 提供了更好的 harness，like Claude Code provides a better harness, 05:42 所以用起来更方便。so it’s easier to use. 05:44 你觉得这是真的吗？Do you find that to be true? 05:45 有人在进行基准测试。There are people doing benchmarks. 05:48 完全正确。Exactly that. 05:48 而且我认为结果恰恰相反。And I think the result is quite opposite. 05:51 这取决于不同的场景。It depends on different scenarios. 05:53 不过我来贴上链接吧。But let me paste the link. 05:56 是的，我认为 Claude Code 之所以流行 Yeah, I do think the popularity of Claude Code 06:02 并不是因为它是个更好的工具，is it’s not because it’s a better tool, 06:05 而是更多由于人们的感知而非实际实质。but because, I don’t know, like due to perception more than substance. 06:11 好，下个问题。Okay, next question. 06:13 你用过以下这些 AI 代理工具吗？Have you used the following AI agent tools? 06:17 我把这个叫做 AI Agent Tools。I call this AI Agent Tools. 06:19 我还没想出更好的名字，I don’t have a better name, 06:22 但它基本上提供了一个网页入口，but basically it gives you a web portal 06:25 让你来操作。to do something. 06:27 你不必使用命令行 You don’t have to use a command line 06:29 ，或者至少用 Perplexity，对吧？or wow, perplexity at least, right? 06:58 好吧，你不必用它们任何一个 All right, you don’t have to use any of them 07:01 只是大致感受一下就好。but just get a sense. 07:05 嗯。Yeah. 07:05 下一个问题，也是最后一个问题：The next question, last question is 07:07 你用过以下这些 AI 生产力工具吗？have you used the following AI productivity tools? 07:11 这些算是小众工具 These are kind of the niche tools 07:14 吧，你看，对不对？like you can see, right? 07:17 生成式 AI 编程工具。The AI coding tool. 07:18 它们非常通用 They are quite general 07:19 你可以做任何事情。you can do anything. 07:21 而且 AI 代理工具变得更加专业化了。And the AI agent tools became more specialized. 07:26 然后，AI 生产力工具 And then the AI productivity tools, 07:28 更加专业化。they are more specialized. 07:29 例如，Gamma 用于制作演示文稿。Like for example Gamma is for presentation. 07:32 Granola 专门用于会议记笔记。Granola is for taking taking notes in meetings. 07:35 Notebook ll 用于消耗知识并 Notebook ll consuming knowledge and generating 07:41 像笔记本和播客那样生成内容。like notebooks and podcasts. 07:45 Genspark 几乎类似于 Manus Genspark is almost like Manus 07:48 ，但更侧重于演示呈现。but more geared towards presentation. 07:51 Hadrian 正在生成一个数字虚拟形象。Hadrian is generating a digital avatar. 07:57 酷。Cool. 07:58 课后我会把结果分享给大家。I’ll share the results with the class afterwards. 08:01 不过谢谢你的回答。But thank you for answering this. 08:04 停下来，我们继续讲座吧。Stop this and go back to our lecture. 08:11 我们将用大约 15 分钟时间讨论如何学习 AI We’ll spend about 15 minutes talking about how to learn AI 08:17 ，然后休息 10 分钟 and then about 10 minutes taking a break 08:20 ，让大家有机会在课堂上自我介绍，and give everyone a chance to introduce yourself in front of the class, 08:24 这很有必要，因为你们都来上课并支付了学费。which is quite useful because you all came to the class and paid the tuition. 08:32 所以你们很可能也面临着类似挑战 So you probably face similar challenges 08:37 和应用场景。and similar applications. 08:40 所以只要主动连接并展现自己，So just by connecting and putting yourself out there, 08:43 就能很好地与志同道合的人建立联系。it’s a very good chance to connect with like minded individual. 08:48 好的，接下来 Yan 将 Okay, and then Yan is going to work through 08:50 深入讲解构建过程，deep dive into a building process 08:53 并向我们传授几项重要原则，and teach us a few important Principles, 08:57 尤其是在上下文工程和上下文架构方面。especially in context engineering context architecture. 09:02 那么接下来我们开放问答环节。Then we’ll open up for Q and A. 09:05 欢迎随时打断我提问。Feel free to interrupt me anytime and ask a question. 09:09 那我们就开始第一个模块吧。So let’s start with module one. 09:12 第一模块是学习如何学习 AI。Module one is learning to learn AI. 09:16 因此，第一模块的议程非常简单。So our agenda for the first module is simple. 09:19 首先，我们来探讨 First we’ll develop understanding 09:21 一下为什么 AI 这么难掌握？why is AI so hard to master? 09:25 这个所谓的范式转变究竟是什么？What is this so called paradigm shift. 09:28 接下来，我们 Then we’ll see it in action 09:29 通过一个动手演示 with a hands on demo 09:31 来实际操作，让这些概念变得生动起来。to make these concepts real. 09:37 好的。All right. 09:39 你也有这种感觉吗？Do you share this feeling? 09:40 AI 容易上手但难用好？AI is easy to start but hard to use? 09:44 那么，什么才算在使用 AI 呢？Well, what is using AI? 09:47 你只需打开 ChatGPT、Cursor 或 Claude Code，You just open ChatGPT or Cursor or Claude Code, 09:51 开始输入文字，这就算在使用 AI 了，start typing and that is using AI 09:54 你能从中学习到 fine tuning、LangChain、harness 等许多概念。you can learn a lot of concepts such as fine tuning, LangChain, harness. 10:00 但是，这能让你 But does that make you use like, 10:02 成为一个优秀的 AI 构建者吗？does that qualify you as a good like AI builder 10:06 或者，你能 or can you can, can you claim that 10:08 声称自己正在使用生成式 AI 吗？you are using it? 10:12 嗯，这个很难说。Well, it’s hard to say. 10:14 这就是我们共同感受到的核心问题。So this is the core problem we all feel. 10:18 生成式 AI 入门非常简单 AI is incredibly easy to start 10:22 但要用好却异常艰难 but frustratingly hard to use well 10:25 我们遇到了许多问题症状。and we have a lot of symptoms. 10:30 所以我想知道，这些症状你是否觉得很熟悉。So I wonder if these symptoms sound familiar to you. 10:33 比如，我在使用 chat 时，For example, I use chat, 10:36 感觉卡住了。but I feel stuck. 10:37 我无法突破它。I cannot move beyond it. 10:39 我被各种炒作淹没了。I’m drowning in hype. 10:41 我分不清什么是真正的机会 I cannot tell what’s a real opportunity 10:43 什么只是噪音 and was just noise 10:44 也不知道在实际工作流程中尝试使用它时会怎样。or when I tried to use it in my real workflow. 10:49 它既笨重又粗糙 It’s chunky and it’s clunky 10:51 ，常常失灵。and often fails. 10:53 或者它适合做演示 Or it’s great for demo 10:55 ，但不可能让它可靠地用于生产环境。but it’s impossible to make it reliable for production. 11:00 原因是，AI 是一种范式变革。And the reason is because AI is a paradigm shift. 11:05 比尔·盖茨说他一生中，Bill Gates said in his entire life, 11:08 或者一生中。or entire life. 11:10 这篇文章是比尔·盖茨在 2023 年 So this article was published by Bill Gates in 2023, 11:14 GPT-4 推出后不久发表的，对吗？right after the launch of GPT4. 11:19 你看，文章的第一句话 And you can see in the first sentence of the article 11:22 说，在比尔·盖茨的整个一生中 is in Bill Gates lifetime 11:26 ，他见过两次 he has seen two demonstrations of technology 11:27 让他震惊为革命性的技术演示。that strikes him as revolutionary. 11:33 当然，AI 时代已经来临。Of course the age of AI has begun. 11:36 所以，他认为具有革命性的第二项技术 So the second technology that he feels is revolutionary 11:39 就是 AI。is AI. 11:41 那么，第一项技术 So what was the first technology 11:44 是他认为最具革命性的？that he struck him as revolutionary? 11:49 欢迎直接在聊天框中输入你的 Feel free to just type your answer 11:52 答案。in chat. 11:55 第一个堪比 AI 的革命性技术是什么？What was the first revolutionary technology that is comparable to AI? 12:01 互联网，Web。The Internet, Web. 12:05 不，那不对。No, that’s not. 12:07 互联网不是正确答案，Internet is not the correct answer, 12:09 至少不是他的答案。or at least not his answer. 12:11 PC。PC. 12:13 不，不，我指的是 No, no, I mean 12:15 工业时代是在比尔·盖茨之前。industrial was before Bill Gates. 12:20 我觉得手机是发明了的 I think mobile phone was invented 12:22 比尔·盖茨出生之前就 before Bill Gates. 12:23 但是移动互联网也不是解决方案。But mobile Internet was not the answer either. 12:26 电力是被发明出来的。Electricity was invented. 12:28 是的。Yes. 12:28 图形用户界面。GUI. 12:29 嗯。Yeah. 12:30 所以，比尔·盖茨已经年过七十了。So Bill Gates was like more than 70 years old. 12:33 在他看来，第一个 The first revolutionary technology to him 12:35 与他认为与 AI 同等重要的 that is on the same level as AI 12:38 革命性技术是 gui，即图形用户界面。is gui, Graphical user interface. 12:44 那么第二项就是 AI。Then the second is AI. 12:45 这个比较是一切的钥匙。This comparison is the key to everything. 12:50 那个我稍后会详细讲讲。I’ll talk about that in more detail later. 12:53 但这个幻灯片仅供理解。But this slide is just to understand. 12:59 这是一场范式变革。It’s a paradigm shift. 13:01 我们必须理解这一范式转变。And we need to understand this paradigm shift. 13:04 这是为什么呢？Why is that? 13:05 因为新的范式会让人感到困惑迷茫。Because a new paradigm is disorienting. 13:08 它带来了诸多挑战，It brings a lot of challenges, 13:10 不仅限于技术层面，还包括认知和情感方面。not just technical, but cognitive and emotional. 13:14 这些挑战归结为三个核心挑战。This challenges boil down to three core challenges. 13:18 第一个是 unlearn。First is unlearn. 13:20 第二个是未知目标。Second is unknown target. 13:22 第三是恐惧并学习。The third is fear and learn. 13:25 你过去的专业知识可能会成为陷阱。Your old expertise can be a trap. 13:29 经典程序员的第一反应是 A classic programmer’s first instinct is to 13:31 例如，对一个模型进行微调。for example, fine tune a model. 13:34 如果你过去学过机器学习，Like if you’ve done machine learning in the past, 13:36 当某件事效果不佳时，if something doesn’t work well, 13:38 本能想去微调模型，you want to fine tune the model, 13:40 但这通常是错误的做法。which is often the wrong approach. 13:43 现在第二个挑战：未知目标。Now the second challenge, unknown target. 13:46 传统的基准测试已经过时了。The old benchmarks are gone. 13:49 我们不知道什么是好的样子。We don’t know what good looks like. 13:51 而且很难将信号与噪声恐惧区分开来。And it’s hard to separate signal from noise fear. 13:56 从专家转变 It’s tough to go from being an expert 13:58 为初学者是非常困难的。to a beginner. 14:01 同样，这种对变得无关紧要的恐惧是真实存在的。Again, this fear of irrelevance is real. 14:04 一个高中生在使用 AI 方面，可能 A high school student may just be better at using AI 14:08 比资深程序员还要出色。than an experienced programmer. 14:12 然后，我们会面临这种潜意识的挑战 And then we face this kind of subconscious challenge 14:15 或者说对变得无关紧要的恐惧。or of fear of irrelevance. 14:23 好，所有这些挫败感的根源 Okay, the root cause of all these frustrations 14:26 在于，我们根本不理解 AI 究竟是什么。is that we don’t fundamentally understand what AI is. 14:31 它不只是一个新的二号。It’s not just a new two. 14:33 那么观众中谁能 So who like among the audience, who do you think 14:36 解释一下，AI 具体是什么，生成式 AI 是什么，can explain what AI specifically what generative AI is 14:43 以及为什么它如此重要？and why it is so important? 14:49 喜欢的话，尽管输入你的答案就好。Feel free to type your answer if you like. 14:51 如果你觉得自己能解释 Just type like yes or yeah, 14:54 什么是生成式 AI 以及 like if you feel like you can explain what generative AI is 14:56 它为什么这么重要，就在聊天框里打‘yes’或‘yeah’吧。and why it is so important, then just type yes. 15:04 在聊天里。In the chat. 15:05 我不用点名，I don’t need to call you, 15:07 但我想大致感受一下，but I just want to get a sense of 15:09 有多少人像我们每天聊 AI 那样，对吧？how many people like we talk about AI every day, right? 15:12 我们经常说 AI 正在改变我们的生活。We talk about AI is changing our life. 15:15 AI 将带来如此根本的变化。AI is going to bring such fundamental change. 15:18 但我们真的理解 AI 是什么 But do we think we actually understand what AI is 15:21 、为什么它如此重要吗？and why it is so important? 15:26 我们能说得具体些吗？Can we be specific? 15:30 我还没看到答案 I haven’t seen an answer yet 15:32 所以我猜这是一个挺有挑战性的问题。so I just assume it’s a challenging question. 15:37 但今天，我们来真正 But today let’s really understand 15:39 搞清楚在这个语境下，AI 究竟是什么？what is the AI in this context? 15:42 仅仅是生成式 AI。Just generative AI. 15:43 假设我们说的是大型语言模型 Let’s say it’s large language models 15:46 以及基于这个模型的其他工具 and other harness on top of this model 15:48 Transformer，具体来说就是这样。transformers to be very specific. 15:55 为什么这点如此重要？Why is that so important? 16:01 好吧，假设今天的问题很有挑战性。Okay, so suppose it’s a challenging question today. 16:04 我们来理解这些看似高大上的 Let’s understand some big words 16:08 术语，其实需要至少分解一层 that are actually at least one level 16:12 才能 break down one level below 16:13 真正搞清楚本质。to get to the bottom of this. 16:18 所以让我们脚踏实地吧。So let’s ground. 16:19 我们第一个要拆解的大词 The first big word we want to break down 16:21 就是 paradigm。is paradigm. 16:24 范式转变。Paradigm shift. 16:25 什么是范式转变？What is the paradigm shift? 16:26 所以，我们用一个具体的故事来阐释范式 So let’s ground paradigm 16:29 吧。With a concrete story. 16:29 为了理解这个，我们来看看一个范式转变吧。To understand this, let’s look at a paradigm shift. 16:33 我们都生活在 iPhone 时代。We all live through the iPhone. 16:35 我们都认同 iPhone 是一个典范，对不对？We all agree that iPhone is a paradigm, right? 16:39 相对于 iPhone 之前的手机 A paradigm shift compared to the functional phones 16:42 这是一个范式转变。before iPhone. 16:45 所以 iPhone 代表了一种新的范式。So iPhone is a new paradigm. 16:47 发生了什么转变？What was the shift? 16:50 请随意输入你的答案。Feel free to type your answer. 16:52 iPhone 带来了怎样的范式转变？What was the paradigm shift of iPhone? 16:55 为什么 iPhone 从根本上与其他手机不同，Why was iPhone fundamentally different from other phones, 16:59 与 iPhone 之前的手机有何本质区别？from the phones before iPhone? 17:07 答案是没有键盘，Answer is no keyboard, 17:08 但这其实只是功能上的差异。but that is actually just a feature difference. 17:11 有很多类似于诺基亚的手机 There are a lot of phones that are like Nokia 17:14 诺基亚生产了一款无键盘手机。manufactured a phone with no keyboard. 17:17 摩托罗拉生产了没有键盘的手机。Motorola manufactured phone without keyboard. 17:22 随时随地都能接入互联网。Having access to the Internet everywhere all the time. 17:26 其他手机也能上网。Other phones also have access to the Internet. 17:29 其实我打算发一篇文章。Actually I’m going to post an article. 17:32 iPhone 没有 3G IPhone didn’t have 3G 17:35 并非第一款具备 3G 互联网功能的手机。was not the first phone that has 3G Internet. 17:40 其实是诺基亚第一个获得 3G 接入的。Actually it was Nokia that first had access to 3G. 17:45 所以 iPhone 几乎就是电脑应用程序。So iPhone almost is computer apps. 17:48 在 iPhone 生态系统中，用户可以选择 In iPhone ecosystem users can select apps 17:51 将应用安装到自己手机的系统里。to be in their phone their own system. 17:53 iOS，我是说其他手机有它们各自的系统。IOS I mean other phones have their own system. 17:58 iOS 的确有所不同。IOS is indeed different. 17:59 但这有什么区别呢？But how is that different? 18:01 App Store 生态系统和更好的 UI App store ecosystem, better ui 18:04 降低了使用手机的成本。reduce the cost for using the phone. 18:06 每个人都可以轻松学会。Everyone could easily learn. 18:07 iPhone 价格非常高。IPhone was very expensive. 18:09 其他手机便宜得多。Other phones was way cheaper. 18:11 强大的生态系统和平台。Strong ecosystem, platform. 18:13 平台，是的，我原本想直接谈 Platform, yeah, I think about converging to what I want to talk about 18:17 生态系统，但让我精确一点来说明。the ecosystem, but let me be very accurate and specific. 18:24 好吧，如果你用旧标准来评判第一款 iPhone Okay, so if you judge the first iPhone 18:28 ，它其实并不是一部特别出色的手机。by the old rules, it was not a very impressive phone actually. 18:32 所以，2007 年的那篇评论 So this is the 2007 review 18:35 指出，它在很多方面其实比 Nokia N95 shows it was actually worse than the Nokia N95 18:39 还要差。in many categories. 18:40 我还记得这部手机。I still remember this phone. 18:41 它还拥有出色的屏幕 It also has a great screen 18:42 支持多点触控，并且没有实体按钮。multi touch and no buttons. 18:45 其实有一些按钮 Actually with some buttons that are like 18:47 很多人都觉得它们更有用。a lot of people consider to be more useful. 18:52 如果你仔细观察，And if you look closely, 18:53 带互联网访问的 iPhone 没有 3G 功能，the Internet access iPhone doesn’t have 3G 18:58 但 Nokia 95 却有 3G。but Nokia 95 had 3G. 19:01 因此，这款手机比 95 So a worse phone compared to 95 19:04 在许多方面差。in many categories. 19:07 但这个评论其实 But this review is actually 19:09 仍陷于旧范式之中。was trapped in the old paradigm. 19:11 它用旧的规则 It was using the old rules 19:13 来对比这个新的范式。to compare this new paradigm. 19:16 iPhone 的真正变革体现在 The real shift of iPhone was twofold 19:19 硬件和软件两个方面。hardware and software. 19:21 在硬件上，它是一个 On hardware, it was a device 19:23 身体的自然延伸设备。that was a natural extension of your body. 19:26 iPhone 内置了许多传感器，包括陀螺仪 IPhone built a lot of sensors gyroscope 19:30 让你可以自然地、多角度使用 iPhone，so you can use iPhone naturally multi attach, 19:33 而无需按按钮。no buttons. 19:35 一切都围绕着如何将它 Everything is about how to use this 19:37 当作你身体的自然延伸来使用。as a natural extension of your body. 19:41 iPhone 之前，它并非为任何特定功能而构建。It was not built for any specific function before iPhone. 19:47 如果你想在手机上拥有某个功能，If you want to have a function on your phone, 19:50 它必须由第一方或少数第三方开发者来开发。it has to be developed by either first party or a few third party developers. 19:55 而且它必须实现这一功能 And it has to make it happen 19:58 以特定方式在手机上。on the phone with a specific way. 20:01 就像苹果的 iPhone 一样 IPhone like Apple 20:04 他们并不关心 was not concerned with what function 20:06 手机上应该放什么功能。should we put on the phone. 20:08 我们刚刚打造了最强大、最直观的平台 We just built the most powerful and intuitive platform 20:12 使用起来非常便捷。to use. 20:14 所以苹果当时的目标 So that was the goal of Apple then 20:17 就是打造一个平台 on the software side, it was a platform 20:19 让每个人都能访问移动连接和计算资源。that gave everyone access to mobile connection and compute. 20:24 所以苹果并没有开发出租车 App。So Apple didn’t build a taxi app. 20:27 他们打造了那个平台 They built the platform 20:29 让 Uber 诞生的 that allowed Uber to exist. 20:31 所以在第一年，So in year one, you know, 20:33 iPhone 跟 Nokia 相比，iPhone compared with Nokia, 20:36 有好的也有坏的，some, some good, some bad, 20:37 有优点也有缺点。some pros, some cons. 20:39 但到了第 10 年，But in year 10, 20:41 你用整个生态系统 you are using the entire ecosystem 20:42 与诺基亚相比。to compare with Nokia. 20:44 那根本无法相比。And that is just no comparison. 20:49 所以，缺少按钮就是一个症状。So no button is one symptom. 20:54 你得看看这个范式。You have to look at the paradigm. 20:57 这种范式就是，我希望这款手机 The paradigm is I just want this phone 21:01 尽可能自然，to be as natural as possible, 21:03 施加尽可能少的限制或主观意见。as little constraints or opinions as possible. 21:09 我想让它使用起来很直观，I want it to be intuitive to use, 21:11 但又不想这样。but I don’t want to. 21:15 乔布斯为何要移除物理按钮？Why did Jobs remove the physical button? 21:18 因为物理按钮是 Because the physical button is 21:19 针对某些特定应用设计的。for some specific applications. 21:22 就像今天，如果你用 Instagram，Like today, if you use Instagram, 21:24 你就不想要那些按钮了。you don’t want the buttons. 21:26 对不对？Right? 21:26 但是没有。But without. 21:28 但是乔布斯发明 iPhone 的时候，But when Jobs invented the iPhone, 21:33 Instagram 还没出现，Instagram was not there, 21:37 但他有远见，but he had the vision of 21:39 要去掉实体按钮，let’s remove the buttons 21:41 让其他应用 so that other applications can exist 21:43 不再受物理按钮的限制。without the limitation of having these physical buttons. 21:46 这就是那个范式。So that is the paradigm. 21:47 这种范式就是 The paradigm is 21:49 去除约束。let’s remove the constraints. 21:50 我们尽量让它直观且通用。Let’s make it as intuitive, as general as possible. 21:55 好。Cool. 21:58 好的，我们不必深入 iPhone 的细节，Okay, so we don’t have to go to the details of iPhone, 22:03 我只是想用这个例子来说明一下。but I just want to use the example to talk about. 22:07 其实要理解一个范式，并不容易。It’s actually not easy to understand a paradigm. 22:10 你必须触及本质，And you have to go to the essence, 22:12 回归基础，you have to go to the fundamentals 22:13 才能理解这一范式转变。to understand this paradigm shift. 22:16 即使到了今天，我们都知道，Even like today, we all know, 22:18 大家都认同，iPhone 是一次范式变革。we all agree, that iPhone was a paradigm shift. 22:22 当时范式转变究竟是什么，还并不十分明显。It was still not quite obvious what the paradigm shift is. 22:28 因此，理解 AI 的范式转变 So understanding the paradigm shift of AI 22:31 可能更有挑战，但我会尽量讲清楚。may be more challenging, but I’ll try to make it clear. 22:37 那么，AI 的范式转变 So the paradigm shift of AI, 22:38 是什么呢？AI 的范式转变究竟是什么？what is the paradigm shift of AI? 22:41 我们来回顾一下 Let’s go back to the value 22:42 所有技术所带来的价值。of all technology. 22:44 如果你采用最抽象的形式，And if you do the most abstract form, 22:46 我们这项技术的价值其实非常简单。the value of our technology is actually simple. 22:50 它让人类能够访问计算资源。It gives human access to compute. 22:52 这就是所有技术所带来的价值。That is the value of all technology. 22:55 所有数字公司、所有软件公司、All the digital companies, all the software companies, 22:58 所有基于字节运营的公司，all the companies that operate on bytes 23:00 都为人类提供了对计算资源的访问。give human access to compute. 23:05 在 AI 出现之前，我们有两种方法来实现这一点。And before AI, we had two ways to do that. 23:09 首先是编程。The first is programming. 23:10 威力强大，但难度很高。Very powerful, but very hard. 23:12 第二个是 gui，The second is gui, 23:13 很简单，但局限性很大。very easy, but very limited. 23:18 AI 是第三个界面。AI is the third interface. 23:21 还记得 interface 吗？Remember interface? 23:23 我们首次 For the first time, 23:24 能够通过自然语言来访问 we can access powerful compute 23:27 强大的计算能力。using natural language. 23:29 这正是比尔·盖茨把 AI 比作 gui 的原因，This is why Bill Gates compares AI to gui, 23:34 因为所有价值都源于访问计算能力。because all the value comes from accessing compute. 23:38 现在，我们拥有了一个全新的界面。And now we have a different interface. 23:41 所以 GUI 和 AI So that’s why GUI and AI 23:45 是同等革命性的技术接口级别。are the same level of revolutionary technology interface. 23:51 这正是这个比较如此完美的原因。This is why the comparison was so perfect. 23:56 编程是通用的，但很难。Programming is general purpose, but hard. 24:00 GUI 虽然简单 The GUI is easy 24:01 ，但构建针对特定任务的 GUI 实际上非常昂贵。but for specific tasks GUI was actually very expensive to build. 24:07 得有人来做一个按钮 Someone needs to build a button 24:09 但做这个按钮需要雇一大堆工程师 and to build the button they need to hire a lot of engineers 24:12 而工程师费用很高。and engineers was expensive. 24:14 所以 GUI 虽然好用 So GUI was easy to use 24:17 但构建成本却非常高。but very expensive to build. 24:20 自然语言既通用又易学 Natural language is both general purpose and easy to learn 24:22 因此打破了传统的权衡取舍。so it breaks the old trade off. 24:29 大家明白了吗？Is this clear? 24:31 如果清楚了，我们来看一些例子 If it’s clear, let’s go to some examples 24:33 直接感受它的实际效果吧。and just see it in action. 24:37 Neph 理论。Neph theory. 24:39 目前为止有什么问题吗？Any questions so far? 24:48 好的，我们来看看它的实际效果吧。Okay, let’s see it in action. 24:53 真正的学习是在一个循环中发生的。Real learning happens in a loop. 24:55 吸收、思考、行动。Absorb, ponder and act. 24:58 但我们大多数人只是卡住不动 But most of us get stuck 25:00 只顾着吸收。and absorb. 25:02 我们面临的障碍并非信息不足，Our barrier isn’t a lack of information, 25:04 而是行动不足。it’s a lack of action. 25:06 因此，在我们的课程中 So through our course 25:09 我们专注于付诸行动。we are focused on doing the action. 25:13 无论是在 Yan 使用 WBR 例子进行的讲座 And whether it’s Yan’s lecture using the WBR example 25:18 还是其他项目和我们的顶石项目，or other projects and our capstone, 25:23 我们都注重行动。we focus on action. 25:26 课程接近尾声时 And towards the end of my course 25:28 我会讨论通过名词学习的一些陷阱 I’ll talk about some traps in learning through nouns 25:31 以及为什么我们应通过动词来学习。and why we should learn through verbs. 25:36 好吧，我们来做些什么吧。But okay, let’s do something. 25:39 这是一个简单的实际任务。Here’s a simple real world task. 25:43 利用 Google Sheets 向每个人发送个性化电子邮件。Sending a personalized email to everyone 25:47 只需将我的目标描述给 AI Just describe my goal to the AI 25:49 如你所见原封不动地即可。exactly as you see here. 25:51 我甚至都不知道 I didn’t even know 25:52 有 Google Apps Script 这个东西。Google Apps script. 25:53 我刚跟 AI 聊了 I just talked with AI 25:55 才知道有这个工具。to know that this tool existed. 25:59 我有一个 Google 表格 I have a Google spreadsheet 26:00 里面有两个名为 email 的列。with two columns named email. 26:02 我想向这个邮箱地址发送邮件。I want to send an email to email address. 26:04 基本上，我想把这个技术栈分享给大家 Basically I want to share this stack with all of you 26:08 但要让它个性化且视觉效果出色。but make it personalized and visually appealing. 26:12 所以我查看列表 So I go to the list 26:14 并通过询问 AI 来操作。and by asking AI. 26:18 所以，So this is, you know, 26:19 我刚把这个输入给 ChatGPT，I just typed this to ChatGPT 26:22 它就给我生成了，and ChatGPT gave me, you know, 26:24 这些命令和 JavaScript 代码。this commands and the JavaScript. 26:28 我按照刚刚复制过来的 JavaScript I follow the JavaScript 26:32 代码来运行。I just copied here and run. 26:33 但是第一次我需要设置几个权限。But the first time I need to do a few permissions. 26:40 好，希望它能成功。Okay, I hope it works. 26:48 好的，你应该能 Okay, you should receive the email 26:49 在我们的邮箱收到那封邮件。in our mailbox. 26:53 如果你还没做 Let me know if you haven’t 26:54 或者已经做了，请告诉我。or if you did. 26:55 我就假设你已经做了 I will just assume you did 26:57 继续往下说吧。and go on. 27:03 好了，那是我们刚才的快速演示。Okay, so that was our quick demo. 27:06 现在我们来搞清楚刚才发生了什么。Now let’s understand what happened. 27:11 有人收到那封邮件了吗？Did anyone got the email? 27:15 我忘了在这里写上我的名字。I forgot to put my name here. 27:18 好的，酷，太酷了。Okay, cool, cool, cool. 27:19 有人做了。Someone did. 27:20 好的，虽然可能要等一会儿 Okay, it may take a while 27:23 但很棒。but great. 27:26 我们用这个例子来搞清楚发生了什么 Let’s use the example to understand what happened 27:29 以及这三个计算接口的作用。and understand these three interfaces for computing. 27:32 另外，为什么 AI 如此重要？And also why is AI so important? 27:35 生成式 AI 的真正价值 What’s actually the value 27:37 和机会究竟是什么？and what is actually the opportunity? 27:40 AI 的机会是什么？What is the opportunity of AI? 27:42 它只是取代我们吗？Does it just replace us 27:43 还是带来了全新的东西？or does it bring something new? 27:46 好的，我们来拆解一下这两个旧的界面。Okay, so our two old interface, 27:53 假如没有 AI，我们回到 Suppose there is no AI and let’s go back 27:55 生成式 AI 发明之前的世界，to the world before generative AI was invented, 27:58 那时我们有编程，也有 GUI。we have programming and we have gui. 28:02 我们来分析一下刚才发生了什么。Let’s break down what just happened. 28:03 在旧世界里，In the old world, 28:05 我们其实有三个选择 we actually had three options 28:07 来做这项任务。for this task. 28:09 首先是体力劳动，对不对？First is manual labor, right? 28:10 我们随时可以复制粘贴。We can always copy paste. 28:12 假设我们要发送一千封 Suppose we want to send a thousand emails, 28:14 个性化的电子邮件。personalized emails. 28:16 我们只需复制粘贴 We just copy paste 28:17 ，然后改一千次名字。and change the name a thousand times. 28:18 那是手动劳动。That’s manual labor. 28:20 第二点是编程，Number two is programming, 28:21 这曾经是一个很高的门槛。which was a high barrier. 28:24 我得知道怎么做这个 I have to know how to do this 28:27 但这并不容易。and it’s not easy. 28:30 尤其是，你要把这里的所有内容都记牢。Especially you have to remember everything here. 28:33 所有的语法和所有命令。All the syntax and all the commands. 28:36 这可不容易。It’s not easy. 28:37 第三种方法是 gui。The third way is gui. 28:40 我可以购买现成的工具 I can buy a pre built tool 28:42 而且有一些特定的现成工具。and there are certain pre built too. 28:45 例如，Mailchimp For example mailchimp, 28:46 是一个由 AI 驱动的平台 AI powered platform 28:48 ，可以免费发送个性化电子邮件。for free sending customized emails. 28:53 所以，第三个选项——GUI So the third option, the GUI 28:54 是一个巨大的商业机会。is a massive business. 28:56 比如 Mailchimp，这家公司只是简单地在邮件脚本上 For example mailchimp, a company that just put button on email script 29:01 加了个按钮，实现自动邮件发送功能，就被卖了 120 亿美元 sending the automatic email automation was sold for 12 billion 29:06 ——而这正是我刚才演示的内容，对吧？for doing what I just did, right? 29:13 为什么？Why? 29:13 但为什么这家公司价值如此之高？But why is this company so valuable? 29:17 因为这项任务很有价值。Because the task is valuable. 29:20 这个任务的价值在于 The task, the value is from 29:22 自动化重复一千次的复制粘贴操作。automating this copy paste a thousand times. 29:26 因此节省了大量人力小时。So the man hour, human hour saved. 29:31 这就是这个按钮的价值，对不对？Here is the value of this button, right? 29:35 这是可以量化的。It can be measured. 29:36 三号与一号之间的价值增量（delta） The value, the delta between number three and number one 29:39 可以通过自动化节省的劳动力量来衡量。can be measured by how much labor was saved from automation. 29:43 这就是为什么第三项如此宝贵的原因。So that’s why number three was so valuable. 29:47 但为什么选项二编程方式 But why didn’t option two programming work 29:50 没有奏效呢？for most people? 29:53 因为从理论上讲，Because in theory, 29:55 自动化只是一次性投资。automation is a one time investment. 29:58 写好代码，Write the code, 29:58 你就解放了。then you are free. 30:00 现实情况却是这张图。In reality it’s this picture. 30:03 我们以为能节省时间，We think we’ll save time, 30:05 结果却花更多时间调试自己写坏的脚本。but we spend more time debugging our own broken script. 30:09 编程的入门障碍太高 The barrier to entry for coding is too high 30:13 结果又太不确定。and the outcome is too uncertain. 30:16 这就是自动化谬论。This is the automation fallacy. 30:19 我觉得这个漫画已经有 20 多年历史了. I think this comic is over 20 years old. 30:22 这是一部很老的漫画，It’s a very old comic, 30:24 但它讲的是事实。but it’s true. 30:26 在生成式 AI 发明之前，Until the invention of AI, 30:28 如果你想花时间进行自动化，if you want to spend time to automate, 30:31 那实在是太难了，也太不确定。it’s just too hard and too uncertain. 30:36 这让我们回到了 AI 范式转变的话题。This brings us back to the AI paradigm shift. 30:39 我们现在拥有这个新界面，We now have this new interface, 30:41 自然语言，natural language, 30:42 它能满足人类的所有需求。think of all human needs. 30:45 像 Mailchimp 这样的图形用户界面，GUI like mailchimp, 30:48 满足了集中化的生产力需求。solved centralized productivity needs. 30:51 所以假设我们将所有人类需求 So suppose we put all human needs, 30:53 放入这个 2x2 的度量框架中。all human demand into this two by two metric. 30:56 这是生产力需求吗？Is it a productivity demand? 30:58 这是娱乐方面的需求吗？Is it an entertainment demand? 31:00 这是不是一种集中需求 Is it a concentrated demand 31:02 ，即很多人有相同的需求？meaning a lot of people share the same demand? 31:04 或者这是长尾需求 Or is it a long tail demand 31:05 也就是说它是个性化的？meaning it’s individualized? 31:07 我有这个需求 I have this demand 31:09 但你和我没有同样的需求。but you don’t share the same demand as I do. 31:12 Excel、Windows、Office，Excel, Windows, Office, 31:16 这些属于集中需求。those are concentrated demand. 31:18 电子邮件自动化，即为集中需求。Email automation, that is concentrated demand. 31:21 很多人有相同生产力需求。A lot of people share the same productivity demand. 31:25 因此，GUI 解决了这个问题。So GUI solved this. 31:29 随后，传统的网页门户解决了集中式娱乐，Then our old web portals solved centralized entertainment, 31:33 而 TikTok 等平台则解决了长尾娱乐。and platforms like TikTok solved long tail entertainment. 31:38 这得益于推荐算法 And this was made possible by recommendation algorithm 31:40 以及传统的机器学习或深度学习技术。the old machine learning or deep learning. 31:44 此外，iPhone 的发明 And also as well as invention of iPhone, 31:47 也让这些视频的生产成本大大降低。making production of these videos much cheaper. 31:51 但是最后一个方框怎么样呢？But what about the last box? 31:53 生产力的长尾效应，The long tail of productivity, 31:55 就是那些数以百万计的小型定制任务，all those millions of small custom tasks 32:00 从来没有人为其开发过 GUI。no one ever built a GUI for. 32:03 这个问题之前没有被解决。It wasn’t solved before. 32:05 但是，软件 3.0 以人类自然语言作为接口，But software 3.0, the human natural language as interface, 32:10 能够解决这个问题。can solve this. 32:14 因此，这个用户生成软件象限，So this quadrant, the user generated software, 32:19 正是 AI 所解决的问题。is what AI solves. 32:23 这就是新的机遇。That is the new opportunity. 32:26 那就是增量部分。That is the incremental thing. 32:28 那就是新事物。That is the new thing. 32:30 如果你只关注如何用 AI 取代 If you focus on how to use AI to replace, 32:33 比如邮件自动化，for example email automation, 32:34 你就是在和一个非常强大的 GUI 竞争。you are competing with a very powerful gui. 32:39 GUI 已经做得非常出色了。The GUI already does the job very well. 32:41 这是一场激烈的竞争。It’s a hard competition. 32:43 但如果你将努力和注意力 But if you focus on your effort and attention 32:46 聚焦于长尾生产力需求，on this long tail productivity demand, 32:49 就能发现许多以往不可能的新机会 you can see like a lot of new opportunity 32:53 如今变为可能，that wasn’t possible before became possible 32:55 即用户生成软件。is user generated software. 33:00 这就是主要机会。This is the main opportunity. 33:01 我们正从必须编写左侧代码的世界 We are moving from a world where you had to write the code 33:05 ，转向只需描述右侧目标的 which is on the left, to one where you just have to describe the goal 33:09 世界。which is on the right. 33:10 这就是颠覆，This is the disruption 33:12 也是我们课程的全部焦点。and it’s the entire focus of our course. 33:17 这就是为什么构建是最佳途径 This is why building is the best way 33:21 真正学习在新范式下。to truly learn in this new paradigm. 33:24 我想强调，I want to repeat, 33:24 在这个生成式 AI 的新范式中，构建是真正学习的最佳途径。building is the best way to truly learn 33:31 在过去的每一次范式转变中，In every past paradigm shift, 33:34 就像过去的范式转变，like paradigm shift in the past, 33:36 获胜者都有一点共同之处。the winner had one thing in common. 33:39 他们不只是用户，更是创造者。They weren’t just users, they were builders. 33:43 看看 PC 或 GUI 互联网，If you look at PC or GUI Internet, 33:46 如果你想抓住机会，if you want to capture the opportunity, 33:47 就必须成为一名建设者（builder）。you have to be a builder. 33:52 这些亿万富翁有何共同之处？What do these billionaires have in common? 33:56 如果你追溯 If you trace the start 33:56 这些成功者的 of their success 33:57 起点，就会发现他们都来自建造者。to all builders. 34:02 但我要强调，But I want to be clear, 34:02 我们还处于非常早期的阶段。we are still very early. 34:05 这个生成式 AI 时代还处于非常早期的阶段。This AI era is still very early. 34:08 我们并不处于 AI 的笔记本电脑时代。We are not in the laptop era of AI. 34:11 我们现在处于主机的时代。We are in the mainframe era. 34:13 这些工具功能强大，The tools are powerful, 34:14 但还很原始、不成熟。but they’re still raw. 34:16 所以来看看这张图片。So look at this picture. 34:17 这就是 IBM 的主机。This is the IBM mainframe. 34:19 看看这台机器有多少 Just look at how many wires 34:21 电线伸出来。come from this machine. 34:22 光是试着操作它就让人望而却步。It’s super daunting to even try to operate. 34:27 我根本不知道怎么操作这样的机器。I wouldn’t know how to operate such a machine. 34:29 这是 PDP10，This is PDP10, 34:31 被视为第一台个人电脑。which was considered the first personal computer. 34:34 虽然只有衣柜那么大，但仍然非常庞大且令人畏惧，Size of a closet, still very huge and very daunting, 34:39 不过看起来还算可控。but seems manageable. 34:42 我想用比尔·盖茨 I want to use the example of Bill Gates 34:45 抓住 PC 机遇的例子 and how he captured the PC opportunity 34:47 ，来加深我们的忧虑，同时为我们的行动提供一些指导 to elevate our concern and also give us some guidance 34:54 ，并放大我们的焦虑感。in our action, elevate our anxiety. 34:57 很多人都在焦虑：嗯，A lot of people are anxious about, well, 35:00 我该怎么办？what should I do? 35:02 我会不会干脆就被 AI 取代了。Would I be just replaced by AI. 35:04 但看看比尔·盖茨的故事，But if you look at the story of Bill Gates, 35:08 我们会发现他其实花了很长时间 we can actually see it took him a long time 35:13 才抓住那些机会。to capture those opportunities. 35:17 来看看比尔·盖茨吧。Look at Bill Gates. 35:18 他并非仅仅从哈佛大学 He didn’t just drop out of Harvard 35:21 辍学。1975 年第一台个人电脑推出时 in 1975 when the first PC was released. 35:26 他从中学时代起，就 He had been programming since middle school 35:28 在早期的巨型机上进行编程了。on early mainframes. 35:31 不是这个，是 PDP 10。Not this one, but PDP 10. 35:33 他高中时学校有一台 His high school had a terminal 35:34 能连接 PDP 10 的 that has access to PDP 10, 35:36 终端机，那正是右边的那台电脑。which was the computer on the right. 35:39 所以大家都知道，他 So we all know that he dropped out 35:41 从哈佛辍学了。from Harvard. 35:42 就像大多數人一樣，當你聽到比爾·蓋茨時，As most people, when you hear about Bill Gates, 35:46 大家都知道他著名的故事是從哈佛輟學，his famous story was dropping off from Harvard, 35:49 但其實他在中學時就是一個熟練的程式設計師。but in middle school he was a proficient coder. 35:53 他在高中时其实创办了一家公司 In high school he actually founded a company 35:56 还被一家计算机中心公司聘用 and was hired by this computer center corporation 35:58 那家公司提供计算机时间的时分共享服务。which was a timeshare of computer time. 36:03 所以他当时就已经很会编程了。So he was already very good at coding then. 36:07 这就是 Altair 8800 那种个人电脑。This is the kind of the personal computer, 36:16 我的意思是，这是第一个 yen，I mean, this was the first one yen, 36:19 我相信它被送到了西雅图的计算机博物馆，I believe went to the Seattle Museum, Computer museum 36:22 并在那里出售了实际的机器。and sell the actual machine. 36:24 它体积仍然很大，It’s still pretty big, 36:27 但一个人就能买来用于个人任务。but one person can buy this and use it on their personal task. 36:31 因此，当 Altair Opportunity 出现时，So when Altair Opportunity appeared, 36:35 他和保罗·艾伦是最先准备好抓住这个机会的人 he and Paul Allen were the first one ready to seize it 36:38 因为他们早已开始构建了。because they had already been building. 36:40 他们懂得如何针对这个进行构建。They know how to build for this. 36:42 他们通过实际构建 And through building they actually understand 36:44 比别人更深刻地理解了这个机会。the opportunity better than other people. 36:47 是的，机会属于那些 Yeah, the opportunity belongs to those 36:50 抢先一步开始构建的人。who start building before everyone else. 36:53 我不重复讲这个故事了，但我记得 I won’t repeat the story, but I remember 36:57 这家公司的 CEO 说过 this company’s CEO said 37:00 公司发布电脑时，when the company released the computer, 37:02 有很多人——超过几十个人—— a lot of people, like more than dozens of people 37:06 找上门来，想为他们写一个解释器。came to them and want to write an interpreter for this company. 37:11 只有 Bill 和 Alan Bill and Alan were the only ones 37:13 真正通读了手册 actually read through the manual 37:14 并成功实现了它。and made it happen. 37:16 因此，他们已经准备充分了 So they were well prepared 37:19 在真正机会来临之前。before the actual opportunity came along. 37:24 嗯。Yep. 37:25 所以，构建是学习的最简单最佳途径 So building is the simple best way to learn 37:27 有两个原因。for two reasons. 37:29 它既是罗盘 It is a compass 37:30 又是推进器。and it is a propeller. 37:33 这是什么意思？What does it mean? 37:34 它是一枚罗盘。It is a compass. 37:36 在满是炒作的海洋里。In a sea of hype. 37:38 唯一重要的是 The only thing that matters is 37:40 这个新技术能否让我的项目成功运行？does this new technique make my project work? 37:46 它是否让我觉得我学到了 Does it make my like I learned 37:48 成千上万的生成式 AI 提示技巧或 prompt engineering 原则。there are thousands of prompting tricks or prompt engineering principles. 37:55 但它真的能给我带来更好的结果吗？But does it actually give me better result? 37:59 因此，通过构建能获得更好的结果 So the better result through building 38:01 它既是一个过滤器，也是一个指南针 is a filter and is a compass 38:04 帮助辨别什么有用，什么无用。of what is useful and what is not useful. 38:07 第二个，The second is, 38:08 更重要的是，I guess more importantly 38:10 它是我们前进的动力。is it is our propeller. 38:12 建造带来的满足感会驱使你 The satisfaction of building will propel you 38:15 持续迭代，这种 to keep iterating in a way 38:17 动力是任何作业都无法比拟的。no homework assignment ever could. 38:21 我们可以给你上这堂课 We can give you the lecture 38:22 ，也可以给你布置大量作业。we can give you a lot of homework. 38:25 但假如你动力十足 But suppose you are so motivated 38:27 ，会把所有作业都做完。that you’ll complete all the homework. 38:31 那如果你做完所有作业呢？What about after you finish all the homework? 38:33 但如果你坚持构建，But if you keep building, 38:35 构建的过程本身就会带给你持续工作的满足感，building itself would give you the satisfaction 38:39 让你不断工作、不断学习。to keep working and keep working and keep learning. 38:42 所以这就是推动力 So that is the propeller 38:43 ，它为你提供了学习的飞轮效应 and that gives you the flywheel 38:46 ，让你知道该学什么。to learn and know what to learn. 38:50 但构建并非像 But building is not as simple as 38:52 ‘咱们就直接建吧’那么简单。let’s just build. 38:55 有一点要记住 One thing to keep in mind 38:57 ：它要求我们保持愚蠢 is it requires us to stay foolish 38:59 ，因为我们正在开创一个新的范式。because we are building a new paradigm. 39:04 因此，保持愚蠢比以往任何时候都 So stay foolish becomes more important 39:07 更加重要。than ever. 39:10 我们必须保持耐心。And we must be patient. 39:13 这个例子说明了汽车 This example was how long it took the car 39:15 成为主流用了多久。to became mainstream. 39:18 我们都知道，汽车取代了马。We all know that car is a replacement of horse. 39:22 显然，发动机技术 Obviously engine is a better technology 39:24 显然比用马拉货要先进得多。than using horse and cargo. 39:27 但是，汽车用了将近 70 年 But it took car almost 70 years 39:30 的时间才取代马匹。to replace the horse. 39:32 那这是为什么呢？And why is that? 39:34 因为你需要的不仅仅是一辆车，Because you don’t just need a car, 39:37 还需要道路、加油站、装配线，you need roads, gas stations and assembly lines, 39:41 服务店、汽车商店和驾驶学校。and service shop, car shop and driver school. 39:45 因此，建立生态系统需要时间。So the ecosystem development takes time to build. 39:50 我们如今在 AI 领域看到，We are seeing that in AI today, 39:52 那些如此智能的模型。models that smart. 39:54 但是你需要构建并利用 But you need to build, harness 39:58 AI 的软件或生态系统。and you need to build software or ecosystem for AI. 40:02 你需要变革组织结构 And you need to change your organization 40:04 以更有效地发挥 AI 的强大威力。to better leverage the power of AI. 40:08 因此，生态系统需要时间来发展。So the ecosystem takes time to develop. 40:12 这是本课程的核心理念，This is the core concept of the course, 40:14 也就是我们这门课要聚焦的内容。of our course. 40:16 我们现在教你 We are now teaching you 40:17 成为一名程序员。to become a coder. 40:19 我们正在教你将 AI 编码 We are teaching you to use AI coding 40:21 作为接口来使用。as an interface. 40:23 编程并非目的，Coding is not the purpose, 40:26 它只是解决问题的通用接口。it is the general interface to solve problems. 40:30 就像回到我们现在的状态一样。Like going back to where we are today. 40:34 我们现在所处的生态系统 Where we are is a lot of ecosystem 40:37 大多不是为 AI 而建，was built not for AI, 40:40 而是为 AI 之前的软件时代构建的。but for the software before AI. 40:45 因此，编程就是接口 So coding is the interface 40:49 与现有生态系统整合。to integrate with this existing ecosystem. 40:53 所以再说一遍，编程并不是目的。So again, coding is not the purpose. 40:56 我们学习 AI 并不 We are not learning AI 40:57 是为了编程。in order to code. 40:59 我们学习 AI 是为了完成任务 We are learning AI to achieve our tasks 41:01 、利用计算资源。to use compute. 41:03 编程只是一个通用的手段 Coding is just a general purpose 41:06 用来解决问题。to solve problems. 41:09 所以要说得非常具体。So to be very specific. 41:11 具体的。Specific. 41:12 我们当下的现实就是这个样子。Our reality today is this. 41:15 我们用自然语言指示 AI We use natural language to tell an AI 41:17 编写代码。to write code. 41:18 这段代码将现有的 API 整合起来 That code glues together existing APIs 41:22 用于访问计算资源。to access compute. 41:25 那我们为什么要这么做呢？And why do we do that? 41:26 因为 AI 读过 Because the AI has read 41:28 我们没读过的所有 API 文档。all the API documents we haven’t. 41:32 它就像一个通用翻译器。It acts as the universal translator. 41:36 这正是当前的分割点。This is the split spot right now. 41:39 这就是我们希望你们 And this is the advantage we hope to give you 41:40 通过学习本课程所获得的独特优势。by learning this course. 41:45 很多人会止步于使用聊天功能。A lot of people will stop at using chat. 41:48 那很强大。That is powerful. 41:49 但这就好比把一个人放在汽车前头。But it’s like putting a person in front of a car. 41:56 而且这确实在历史上发生过。And that actually happened in history. 42:00 我相信英国有过一条法律 I believe in Britain they had a law 42:03 要求在汽车前方派人 that you have to put someone in front of a car 42:06 举旗引导 with a flag to get the car 42:08 以防汽车引发任何事故。so the car doesn’t cause any accident. 42:12 那并不是 That was not the way 42:13 发挥汽车威力正确的方法。to leverage the power of the car. 42:17 但如果你只是与 AI 闲聊，But if you are just chatting with AI, 42:19 你就相当于那个站在车前挥旗的人。you are that person in front of the car. 42:23 AI 不会自动运行 The AI is not running by itself 42:25 虽然 AI 很强大 AI is powerful 42:27 但每一步都需要人类的互动。but every turn it requires interaction. 42:30 但是如果你能学会把代码当作通用接口 But if you can learn how to use code as the general interface 42:36 通过代码访问计算资源，and use code to access compute, 42:40 同时利用 AI 来辅助编程，and you can use how to use AI to leverage coding, 42:44 那你就掌握了最佳结合点。then this gives you the sweet spot. 42:48 如今，你能做的已经 Today you can do so much more 42:50 远超单纯的绘图。than just charting. 42:54 好吧，这段旅程不会顺利无阻。Okay, the journey is not going to be smooth. 42:57 你会面临挑战，You will hit challenges, 42:59 但这些挑战正是机会。but those challenges are the opportunity. 43:03 如果事情那么容易，它也就没有价值了，对不对？If it were easy, it wouldn’t be valuable, right? 43:07 我们常常会为这些挑战感到沮丧。Often we are frustrated by these challenges. 43:11 但你想想看。But think about this. 43:12 如果谁都能一蹴而就 If anyone can do this in one shot 43:14 那还有什么意义呢？then what’s the point? 43:16 任何人都能做到这一点。Anyone can do this. 43:17 没有模式，There is no mode, 43:18 也没有差异化因素。There is no differentiator. 43:22 克服了这个挑战—— By overcoming the challenge, 43:23 你搞定了 10 个错误，the fact that you overcame 10 errors 43:25 而别人却放弃了—— and someone else gave up 43:27 这正是你全部的竞争优势所在。is your entire competitive advantage. 43:31 这就是为什么 Which is why 43:32 我们要回到该节标题的原因。to the section title. 43:35 这正是我们必须保持愚蠢的原因。Which is why we must stay foolish. 43:38 我们必须抱持初学者的心态。We must adopt a beginner’s mind. 43:41 你过去的专长能 Your old expertise can blend you 43:43 引领你探索新的可能性。to new possibilities. 43:48 这句话有两种解读方式。This quote can be read in two ways. 43:51 初心里蕴藏着无限可能。In the beginner’s mind, there are many possibilities. 43:53 但专家中寥寥无几。But in the experts there are few. 43:55 所以可以說 So you can say 43:56 ，專家懂得很多。experts know a lot. 43:58 因此，他们知道哪条路径可行 So they know which path works 44:03 就能避免犯错 and can just avoid the mistakes 44:05 也避免浪费时间和精力。or avoid wasting time and energy. 44:10 但请记住，But remember, 44:11 我们现在进入了一个全新的范式。we are now in a new paradigm. 44:13 没有人是专家。Nobody is the expert. 44:15 所以，如果你假装成专家，So if you pretend to be the expert, 44:18 就会错失许多机会。you are miss the many opportunities. 44:21 这就是初学者心态的原因。That’s why the Beginner’s mind. 44:23 这本书名叫《初心》。This book is called the Beginner’s Mind. 44:25 禅宗的心境。The Zen’s Mind. 44:26 禅者的心是初心。The Zen’s mind is the beginner’s mind. 44:29 所以保持愚昧 So stay foolish 44:30 抓住眼前的机遇吧。and see the opportunities in front of you. 44:32 别让过去的经验 Do not let your past experience 44:35 遮蔽了你的双眼，看不到真正的机遇。blind you from seeing the real opportunity. 44:42 好了，现在该轮到我们 Okay, so now goes to our causal action 44:44 对你们的因果行动了。to you. 44:46 现在是我们转变身份的时候了。It’s time to shift our identity. 44:49 这就是新的口号。This is the new mantra. 44:50 用户反馈这个 app 很糟糕。The users say this app is bad. 44:54 开发者说，我可以改进它。The builder say I can improve it. 44:56 用户说，我需要的那个 app 不见了。The user say the app I need is missing. 45:00 建造者说，我能把它建起来。The builder says I can build it. 45:03 所以就让它实现吧。So just make it happen. 45:05 努力让它实现 Try to make it happen 45:06 ，你会惊讶自己能走多远。and you will be surprised how far you can go. 45:10 在寻找问题时，When you look for problems, 45:13 不要妄图一步登天成为下一个马克·扎克伯格。do not try to be the next Mark Zuckerberg 45:21 要现实一点，管理好你的期望。Be realistic and manage your expectation. 45:25 请关注右下这个象限。Focus on this bottom right quadrant. 45:28 所以，如果我们把所有需求 So if we again put all the demand 45:32 和机会都放回这个 2x2 矩阵中，and all the opportunities into this two by two matrix, 45:36 它是有价值的，还是具有颠覆性的？is it valuable or is it disruptive? 45:42 有些事情以前虽然可行，There are things that were possible before, 45:44 但成本太高了。but way too expensive. 45:45 比如，Like for example, 45:46 为每个人创建一个个人网站，a personal website for everybody 45:48 就能把成本降到零。drops the cost to zero. 45:51 所以，如果某事物价值不高但具有颠覆性，So if it’s not so valuable but disruptive, 45:56 这就是你今天能做的事。this is what you can do today. 46:00 这些是我们课程中的原型。This is the prototypes from our course. 46:02 当我交给你们这些项目时，When I give you the projects, 46:05 它们会停留在这个象限中。they stay in this quadrant. 46:08 我们这么做的原因是 The reason we do this is 46:10 希望培养构建的心态和习惯。we want to build the mindset and habit of building. 46:14 我们希望培养起构建的心态与习惯，We want to acquire the mindset and habit 46:15 也就是围绕「构建」这一点。of building. 46:17 因此，當時間、生態系統和創造力開始發揮作用時，So if when time, ecosystem and creativity comes in play, 46:22 我們有些人就能抓住那些 10 倍機會，some of us can capture those 10x opportunities, 46:26 那些寶貴的機會。those valuable opportunities. 46:29 但你必须几乎按这个顺序来做 But you have to do this almost in order 46:33 才能做得正确，因为你，to do this right because you, 46:37 没人能瞬间就成为建筑大师。you do not become a master builder just instantly. 46:42 往往时机和生态系统都尚未成熟。Often the timing and the ecosystem is just not ready yet. 46:47 需要很多次迭代 It takes a lot of iterations 46:49 要真正培养出品味和创造力 to actually acquire the taste and creativity 46:52 去发现机会。to see the opportunity. 46:54 下一张幻灯片我将举一些例子。I’ll give some examples in the next slide. 46:58 所以我们从这里起步，So we start here, 47:00 养成习惯，develop our habits 47:01 从而发现那些 10 倍增长的机会。and this is how we find the 10x opportunities. 47:04 历史证明了这种模式。And history shows this pattern. 47:07 首先，我们将新技术应用到旧问题上，例如网页视频 First we apply new tech to old problems like video on the web 47:12 ，或者马克·库班通过将视频引入互联网 or Mark Cuban built his fortune by bringing video to the Internet 47:16 而积累财富，都是将新技术用于旧问题 applied the new tech to old problems 47:19 。随后，Google 涌现出原生新技术解决方案。then new tech native solutions emerged at Google. 47:25 互联网初期，At the beginning of Internet, 47:26 Google 并不特别有用，Google is not going to be very useful 47:29 因为可供搜索的网站很少。because there are not many websites to search from. 47:31 你想使用像 Yahoo 这样的目录 You want to use a category like Yahoo 47:35 来访问，了解有哪些网站可用。to access to know what websites are out there. 47:39 但是当信息呈指数级增长时，But when information grows exponentially, 47:41 你就无法再依赖 Yahoo then you can no longer use Yahoo 47:44 来获取所有信息了。to access all the information. 47:47 你需要全新的技术原生解决方案 You need a new tech native solution 47:48 像 Google 那样。like Google. 47:49 但请记住，在新技术初期，But remember at the beginning of the new tech, 47:51 Google 是没用的。Google is not useful. 47:53 只有 Google 才会变得有用 Google only becomes useful 47:55 当新技术成熟时。when the new tech matures. 47:57 这就是为什么预测未来如此困难的原因。So that’s why it’s very hard to predict the future. 48:00 这就是为什么很难预测 That’s why it’s very hard to predict 48:02 哪种 AI 原生技术 what is an AI native technology 48:04 会比以往壮大 10 倍。that is 10 times bigger than before. 48:07 几乎无法预测。It’s almost impossible to predict. 48:09 但要计算它。But to compute it. 48:10 从打造人们真正需要的东西开始。Starting with building simple things people really want. 48:14 另一个例子就是 Instagram。Another example is Instagram. 48:16 Instagram 一开始并不是我们如今所知的社交媒体。Instagram started not as the social media we have today. 48:20 它最初是一个未来的应用，It started as a future app, 48:21 只是一个简单的滤镜应用。just simple filter app. 48:23 但是 Instagram 的不同之处在于 But the difference of Instagram is 48:25 Instagram 非常出色。Instagram was very good. 48:27 于是大家都开始用 Instagram So everyone was using Instagram 48:31 上传了大量照片 and they start uploading a lot of pictures 48:32 并开始分享图片。and they start sharing pictures. 48:34 接着，Facebook 抓住了机会 And then Facebook saw the opportunity 48:35 收购了 Instagram。and bought Instagram. 48:37 Instagram 与市场上其他数十款滤镜应用的区别在于 The difference between Instagram and all the other dozens of filter app on the market 48:42 Instagram 非常出色。was Instagram was very good. 48:47 因此，为了让我们的教学内容清晰明确，So to be crystal clear in what we teach, 48:50 我们现在教你如何构建大型语言模型（LLM）。we are now teaching you how to build an LLM. 48:56 这个观点是：浏览器发明之时，And the proposition is when the browser was invented, 48:59 机会不在于造一个更好的浏览器，the opportunity wasn’t to build a better browser, 49:03 而在于创建网站。it was to build websites. 49:06 iPhone 推出时，When the iPhone came out, 49:07 机会不在于再造一个 iPhone，the opportunity wasn’t to build a new iPhone, 49:10 而在于为 iPhone 开发应用。it was to build apps on iPhone. 49:13 所以我们教大家如何用 AI 来构建解决方案 So we are teaching you how to build solutions with AI 49:18 我们相信这是一种能随着 AI 发展而成长的技能或专长。and we believe this is skill or expertise that can grow with AI. 49:25 当 AI 变得更强大，When AI becomes better, 49:27 你就能构建更多东西。you just can build more things. 49:30 所以学习 AI 虽然难，So learning AI is hard, 49:33 但过程其实很简单。but the process is simple. 49:37 首先，大多数人觉得自己落后了 The first thing is most people think they are behind 49:40 是因为他们不熟悉足够的 AI 术语。because they do not know enough AI terms. 49:43 请选择我们的框架。Choose our frameworks. 49:45 这通常不是真正瓶颈 This is usually not the real bottleneck 49:47 在实际项目中。in real projects. 49:49 差距不在于 The gap is not 49:51 谁无法解释 RAG。who is not who can explain rag. 49:55 AI 能比你更好地解释 rag。AI can explain rag much better than you do. 49:59 真正的差距在于谁能拆解复杂任务、The real gap is who can break down a messy task, 50:03 有效管理上下文、manage context well, 50:05 判断生成式 AI 与人类各自该做什么、decide what AI should do versus what humans should do, 50:09 诊断问题并不断迭代。diagnose failure and keep iterating. 50:12 还记得我刚才说过的话吗？Remember what I just said? 50:16 分解任务、管理上下文、Break down, manage context, 50:19 决定 AI 该做什么、人类该做什么、decide what AI should do, what humans should do, 50:23 诊断并迭代。diagnose iterating. 50:25 这些全都是动词。Those are all verbs. 50:27 Twos 是组成部分。Twos are components. 50:28 执行才是真正的优势所在。Execution is the real advantage. 50:31 这就是为什么人们即便掌握所有词汇 So that is why people can know all the vocabulary 50:34 ，却仍无法产出任何有用的成果。and still fail to ship anything useful. 50:37 动词是我们课程的核心操作能力。Verbs are the operating ability in our course. 50:41 要特别注意动词。Pay attention to verb. 50:42 不如关注 Yan 在做什么 Pay attention to what Yan is doing 50:45 与其只是学习这个概念 instead of just let’s learn this concept. 50:48 学习概念太简单 Learning concept is so easy 50:50 、太静态 and so static 50:51 、太琐碎了。and so trivial. 50:52 但是学习动词才是最难的部分。But learning verb is the hard part. 50:54 这就是我们所提供的。And that’s what we deliver. 50:57 但我们并不是反对工具。But we are not anti tool. 50:58 工具至关重要。Tools matter. 51:00 但工具只有 But tools only matter 51:00 当你掌握如何 once you know how to operate them 51:01 在实际工作中运用它们时，才真正重要。inside real work. 51:05 这就是为什么本课程聚焦于 This is why this course focus on 51:07 让 AI 在实际应用中可靠的框架。frameworks that makes AI reliable in practice. 51:11 我们不教你那些玩具般的项目，We don’t teach you toy projects, 51:13 而是教你如何在实际工作中让它变得可靠。we teach you how to make it reliable in real work. 51:17 比如，上下文架构、Like for example context architecture, 51:19 委托谱系之类的，还有代理循环。delegation spectrum, like agent loops. 51:23 这些是我们必须创造出来的术语 Those are terms we have to invent 51:26 为了描述实际传授给你们的专长。in order to describe the actual expertise we teach you. 51:30 但还是要再次强调，注意动词。But again, pay attention to verb. 51:33 是的，要学名词，So yes, learn the nouns, 51:35 但真正的杠杆作用来自于动词。but the real leverage come from the verbs. 51:37 这正是了解 AI That is the difference between knowing AI 51:39 与真正用它构建的区别所在。and actually building with it. 51:43 本课程旨在填补五个具体差距 The course is designed to close five concrete gaps 51:45 这些差距将普通 AI 用户与真正 AI 构建者区分开来。that separates ordinary AI users from real AI builders. 51:51 第一个是输出差距。The first is output gap. 51:52 大多数人认为更好的提示就是答案 Most people think better prompting is the answer 51:54 或者不是答案。or it’s not. 51:55 真正的差异在于，AI 是否能 The real difference is whether AI sees 51:58 看到正确的上下文约束、标准以及先前的决策。the right context constraints, standards and prior decisions. 52:02 第二个是质量差距。The second is quality gap. 52:04 当生成式 AI 的输出很差时，When AI output is bad, 52:06 大多数人只会重试。most people just retry. 52:07 优秀的构建者会诊断出真正的原因。Strong builders diagnose the actual cause. 52:10 缺少上下文、上下文饱和、指令模糊或模型限制。Missing context, saturated context, ambiguous instructions or model limits. 52:15 他们明白真正导致错误的原因是什么。They understand what what actually caused the error. 52:19 它们能调试并改进它。They are able to debug and make it better. 52:22 第三是委托差距。The third is delegation gap. 52:24 大多数用户仍把 AI Most users still operate AI 52:25 当作一个速度极快的实习生来使用。as a very fast intern. 52:28 他们需要持续进行监督。They have to supervise constantly. 52:30 我们希望你们逐步实现将完整任务交给 AI We want you to move towards handing off complete tasks 52:34 带有明确目标、边界和检查机制。with clear goals, boundaries and checks. 52:36 接下来，我们来谈谈如何信任‘不信任 AI’，Next, we’re going to talk about how to trust not trusting AI, 52:42 而是通过构建反馈循环、上下文和架构，but build the feedback loop and the context and the architecture 52:47 让 AI 自我迭代，直至达到预设的终点。so that AI can iterate on itself until it crosses a predefined finishing line. 52:54 第四个是累积差距。The fourth is the compounding gap. 52:55 没有系统，Without a system, 52:56 每个好的结果都是一锤子买卖，every good Result is one off 52:58 每个错误都会反复出现。and every mistake gets repeated. 53:00 我们将演示如何把经验 We’ll show you how to turn experience 53:02 转化为可复用的资产和团队记忆。into reusable assets and team memory. 53:05 第五个是杠杆差距。The fifth is the leverage gap. 53:07 前四个要素到位后，Once the first four are in place, 53:09 AI 就不再只是一个执行工具。AI stops being just an execution tool. 53:13 它开始真正成为你的思想伙伴 It starts becoming a real thought partner 53:16 能帮助你发现表面的盲点 and can help you reason surface blind spots 53:18 并扩展你的判断力。and extend your judgments. 53:21 因此，这个课程的承诺不只是 So the promise of this class is not just 53:23 让你了解更多 AI 术语。that you will know more AI terms. 53:26 就是要你填补这五个差距 It is that you are close these five gaps 53:28 ，以更高水平来运用 AI。and operate the AI at a much higher level. 53:32 希望这个前景能让你对本课程充满期待。Hope the promise gets you excited about the course. 53:36 它涉及两件事。It’s about two things. 53:37 心态与幸福。Mindset and happiness. 53:40 这就是秘诀。And this is the recipe. 53:41 这个配方非常简单。The recipe is super simple. 53:42 化妆品，正是人们想要的。The makeup, something people want. 53:44 循环。Loop. 53:46 如果你想构建产品，If you want to build, 53:47 只需发现问题、just discover a problem, 53:48 开发解决方案、build a solution, 53:50 克服挑战、overcome challenges, 53:50 找到市场，然后不断重复。find the market and repeat. 53:53 这是一个简单但却艰难的过程。This is the simple yet hard process. 53:58 这门课程的结束 And the end of this course 53:59 只是你真正学习之旅的起点。is just the beginning of your real learning. 54:02 我们的目标不是给你 Our goal isn’t to give you 54:03 一次性知识的提升。a one time knowledge bump. 54:06 它的目的是赋予你思维方式和习惯 It is to give you the mindset and habit 54:08 改变人生轨迹。to change your entire trajectory. 54:11 所以，希望在每一个节点，就像每一个分叉一样，So hopefully every nod, like every fork, 54:14 你选择不同的路径，you choose a different path, 54:15 以不同的方式做事，you do things differently, 54:16 用不同的视角看待事物 you’ll see things differently 54:18 并选择不同的道路。and you select a different path. 54:19 而且随着时间的积累，它将形成真正的竞争优势。And with time it will compound into real advantage. 54:26 你不会孤单一人。You won’t be alone. 54:28 我们建立了一个社区，支持你 We have built a community to support you 54:30 在这一旅程中前行，一起分享项目、提问 on this journey, share projects, ask questions 54:32 并学习。and learn together. 54:34 今天课后我会发邀请给你 After today’s course I will send an invitation to you 54:37 你就能获得社区的所有权限，so you will get all the access to the community, 54:41 包括自定进度的课程。including the self paced course. 54:45 我主要给你们展示中文版本 I’ll basically show you there is a Chinese version 54:49 因为我的学生大多说中文，because most of my students are Chinese speaking, 54:53 但我们构建了真实的翻译流程，将所有内容都转为英语，but we built authentic flow to translate everything into English, 55:00 包括比如评论。including for example the comments. 55:07 我们过去的学生，以及被社区吸引而加入的人，现在已经有了超过 400 个真实项目。There are over, I believe now over 400 real projects 55:17 你随时可以提问并进行讨论。And you can always ask questions and do discussions. 55:19 我们到了这里。Here we are. 55:22 回答所有问题。Answer everything. 55:23 你将获得社区的终身访问权限 You get lifetime access to the community 55:27 以及在社区中提问的终身权限 and lifetime access to asking questions in this community 55:31 他们会为你解答。and they will answer it for you. 55:34 好，我们开始吧。Okay, so it’s time to get started. 55:38 首先，关于作业，我们有五个小项目。First, homework, we have five mini projects. 55:40 它们看似复杂，但借助 AI 的帮助 They look complex, but with AIs help 55:43 你只需大约 5 到 10 分钟就能构建每一个。you can build each one in about five to 10 minutes. 55:46 所以试试分享它吧。So try and share it. 55:49 我已经看到 Sasha 发布了项目。I already see Sasha posting the project. 55:52 太好了。Great. 55:53 所以把你的答案发到这里来。So post your answer here. 55:56 每项任务都不应超过 10 分钟。Each of them should take no more than 10 minutes. 56:00 其次，下周的 Second, for the next week, 56:01 任务我们特意设计得看起来很复杂，we specifically designed these tasks to look complex 56:06 不试试看，你可能会觉得 and without trying, you may think 56:09 每个都要花好几个小时才能解决。it will take hours to solve. 56:14 但试试用 AI 吧 But just try with AI 56:15 你会发现真正的神奇之处 and see the real magic happens 56:17 在于开始尝试的那一刻。when you start trying. 56:20 其次，So second, 56:21 下周我想让你找出 for the next week I want you to find 56:23 日常生活中的五个小问题或 maybe five small problems or frictions 56:26 不便之处。in your daily life. 56:28 就把它们写下来吧。Just write them down. 56:29 训练我们的人去发现机会。Train our man to see opportunities. 56:32 所以这是在总体部分。So this is in the general section. 56:33 我已经发了一个。I already posted one. 56:35 我想把我的中文视频 I want to translate my Chinese video 56:36 翻译成英文文章，然后发到 X 上。into English articles then post them on X. 56:40 把日常生活中遇到的事情记下来 Just jot down what you encountered in your daily life 56:44 就写在这里吧。and just write it down here. 56:47 一旦养成发现问题的习惯，Once you build that habit of discovering problems, 56:50 你就会在日常生活中每天看到问题 you see problems every day in our daily life 56:54 这些就是机会。and those are the opportunities. 56:56 第三，送你一份礼物。Third, a gift for you. 56:58 我们来构建一个人。We build a person. 56:59 哦，其实像‘是的，Oh actually the introducing yourself like yeah, 57:01 这个社区就是这样’一样介绍自己。this community is so. 57:08 我不清楚，I don’t know, 57:09 我希望能打造一个紧密团结的社区。I hope to build a close knit community. 57:14 我们大家齐聚一堂，目的相同 We all came here for one purpose 57:17 通过相互连接，我们应当能够彼此支持。and by connecting we should be able to support each other. 57:21 当然，你已经加入了 And of course you’ve joined 57:23 Superlinear Academy 这个规模更大的社区。the much larger community of Superlinear Academy. 57:29 如果你看看这里的人数 If you look at how many people here there are 57:32 已经大约有 14,000 人了。about 14,000 people already. 57:37 不过这个社区规模较小 But this community is small 57:40 小社区的优势在于大家关系紧密 and the small community can have the advantage of being close 57:45 我希望我们都能花 10 分钟 and I hope we all give 10 minutes 57:48 时间互相认识一下。for us to introduce each other. 57:52 所以我希望我们能互相深入了解 So I hope we will know each other well 57:57 知道如何彼此帮助 and we will know how to help each other 57:59 和支持 and how to support each other 58:00 这正是 AI 无法取代的巨大优势。and that is a lot of benefit that AI cannot replace. 58:07 好，我们来总结一下。Okay, let’s summarize. 58:09 AI 很难学 AI is tough to learn 58:10 因为它是一种全新的范式。because it’s a new paradigm. 58:14 这一范式是将自然语言作为接口 This paradigm is natural language as interface 58:17 计算的。to compute. 58:19 最好的学习方法就是在早期与 AI 一起构建 The best way to learn is to build with AI while early 58:23 这意味着挑战其实是变相的机会。which means challenges are just opportunities in disguise. 58:26 专注于培养建设者心态和习惯。Focus on acquiring builder mindset and habit. 58:29 打造人们真正需要的东西 Build something people want 58:31 并保持愚昧之心。and stay foolish. 58:33 好的，我们来休息一下。Okay, let’s go to our break. 58:45 所以你可以打开 So you can just turn on 58:50 我们的视频。your turn on our video. 58:54 介绍部分将谈谈我是谁——Fanfax The introduction will be who am I Fanfax 58:58 ，以及我如何帮助我的同学们。and how can I help my classmates? 59:03 我先简单自我介绍一下吧。Probably I can do a self introduction first 59:05 我叫 Yan，or others are thinking My name is Yan, 59:06 大家大多已经知道，I am as many of you already know 59:08 我是 Superlinear Academy 的讲师兼联合创始人。and both instructor and a co founder of Superlinear Academy 59:14 下面我来讲两个关于我的故事。and I can tell two stories about who I am. 59:18 首先，我们有一个小型 E ink The first is we have this small E ink 59:22 哎呀，它现在开始闪烁了。Oops small E ink saying it begins flashing now 59:26 它显示了过去七天我使用了多少 tokens。which counts how many tokens I use for the past seven days 59:33 过去七天我用了 I used it’s externally refreshing so it’s a bit hard to tell 59:35 超过 50 亿 tokens，I think seven days I used more than 5 billion tokens 59:39 过去 30 天用了 140 亿 tokens，and for the past 30 days I used 14 billion tokens 59:45 这花费了 11,000 美元，但节省了我超过 300 小时。and that would cost $11,000 and saved me more than 300 hours. 59:53 让我瞧瞧能不能让它在这里对焦。Let me see whether we can get it focused here. 59:56 嗯，这是个很有趣的项目。Yep so it’s a fun project. 59:59 我凡事都用 AI I use AI for everything 01:00:01 ，第二个有趣的事实是，and the second fun fact is 01:00:04 过去一个月我一直睡不好觉 I had a hard time trying to sleep 01:00:07 ，于是就把所有数据 for the past month so I just exported everything 01:00:10 都导出来了。from my Apple Health. 01:00:12 只需让 AI 直接去做，输入 AI Just ask AI to Do it into the AI 01:00:14 写一个技能，让它进行分析。write a skill, ask it to do analysis. 01:00:16 它对许多变量进行了回归测试 It did a regression test on a lot of variables 01:00:20 ，发现最有影响力的变量是‘晚上工作’。and found out the most significant variable was I work at in the evening. 01:00:25 于是我停下了那个习惯 So I stopped doing that 01:00:28 过去三周平均每天多出 1.2 小时 and gained 1.2 hours daily in average 01:00:32 相当于每天多睡 1.2 小时。Like having 1.2 hours more sleep every day. 01:00:36 我对那个故事有另一种解读。And I have a different interpretation of that story. 01:00:40 原因是你的 AI 使用量太多了。The factor was you were using AI too much. 01:00:44 那就是让你睡不着觉的原因。That’s what caused the loss of sleep. 01:00:46 是的，没错。Yeah, that’s true. 01:00:47 没错。That’s true. 01:00:48 解决办法就是少用 AI And the solution to that is use less AI 01:00:51 ，这样我就能重新入睡了。and then I get into sleep again. 01:00:57 所以某种程度上，So to some extent, 01:00:58 AI 试图自我毁灭。the AI was trying to kill itself. 01:01:03 好啊，很乐意把话筒 All right, happy to hand the mic 01:01:04 交给下一个想第二个发言的人。to the next one who wants to go second. 01:01:08 我可以接下来发言。I can go next. 01:01:10 谢谢，Yan。Thanks, Yan. 01:01:12 我叫 Maggie And my name is Maggie 01:01:14 ，是西雅图的一名数据科学家 and I’m a data scientist in Seattle 01:01:16 ，目前在亚马逊工作。and working for Amazon. 01:01:19 有趣的是，And the fun fact is 01:01:20 我刚刚读完了刚刚分享的那本书。I just finished the reading of the. 01:01:25 就是刚刚分享的那本书。Of the book just shared. 01:01:31 是的。It is. 01:01:32 嗯，这是一位非常著名的日本僧人 Yeah, it is from a very famous Japanese monk, 01:01:36 ——大师的话。the Master. 01:01:38 他曾是史蒂夫·乔布斯的老师 And he used to be the teacher of Steven Jobs 01:01:43 教导他禅宗的脱俗心境 and taught him the out of the Zen mind 01:01:46 这对乔布斯的职业生涯和他的人生都大有裨益。and which is helpful in Job’s career and his life. 01:01:55 第三个问题是 And the third question was 01:01:58 你怎样帮助同学？how can you help classmates? 01:02:02 是的。Yeah. 01:02:04 说实话，我现在公司里的工作 To be honest and the current work now in company 01:02:09 还能用一些基础的 AI 工具 I can still handle with some very basic AI tools 01:02:14 ，或者传统方法来应付。and or just do it in a traditional way. 01:02:18 但说实话，But to be honest, 01:02:19 我知道这不会持续太久。I know that will not last very long. 01:02:23 所以我觉得应该多分享一些思考、多提一些问题 And so I think try to like share more thinkings, share more questions 01:02:30 ，包括课堂上、工作中 in the, in the class and from my work 01:02:33 、项目中，以及工作之外的个人兴趣 and also from the like the projects or the personal interest out of the work work 01:02:38 ，也许还能帮助同学们一起思考 and maybe help the classmates to think together 01:02:42 。因为说实话，刚开始的几个问题 because to be honest for a couple of questions at the beginning 01:02:47 和通常使用工具的情况是这样。and usual access the tools. 01:02:52 我还没用过的大多数工具，Most of the tools I haven’t used, 01:02:55 我都没尝试过。I haven’t tried. 01:02:58 所以是的，我来这里也是想打下基础 And so yeah and I also came here to want to build a foundation 01:03:03 ，让我们在市场上工具太多时不会迷失方向。of how to be not lost when we have like too many tools there in the market. 01:03:13 嗯。Yeah. 01:03:13 好的。Okay. 01:03:14 然后传给下一个。And pass it to next. 01:03:20 我会安排订单，确保周末顺利度过。I’ll assign orders to make a smooth weekend. 01:03:27 哎呀，被叫到了。Oh gosh, called out. 01:03:28 对不起。Sorry. 01:03:28 我总是对这些东西感到特别紧张。I always get so nervous of these things. 01:03:30 不过我的名字叫 Reagan。But my name is Reagan. 01:03:31 我住在加拿大不列颠哥伦比亚省的 Squamish。I live in Squamish, B.C. 01:03:33 加拿大。canada. 01:03:34 我在 Lattice 公司做软件工程师 I work for a company called Lattice as a software engineer 01:03:38 一直在想一个关于我的有趣的事儿 and been trying to think of an interesting thing about me 01:03:41 结果来晚了。and I’m coming up late. 01:03:42 但我住在 Squamish，But I live in Squamish, 01:03:44 这是一个大型户外活动胜地。which is a big outdoor place. 01:03:46 我一直都在徒步和骑摩托车。I hike and motorbike all the time. 01:03:48 而且就像帮助团队其他人 And just as far as like helping others on the team 01:03:51 同时自己也得到帮助一样。and also being helped. 01:03:53 我很喜欢这个关于社区的演讲，I love this talk of community and 01:03:56 就像鼓励大家发声、提问一样。Kind of like speaking out, asking questions 01:03:59 我特别兴奋，能和一群 and I’m like really excited to be in a room 01:04:01 同样感到有些迷茫却又充满期待的人共处一室。of people who are feeling also kind of lost but excited 01:04:05 到目前为止，我超级喜欢所听到的一切。and I’m just really liking what I’m hearing so far 01:04:08 我觉得保持投入、积极提问，and I think myself staying engaged and asking questions 01:04:11 而且对所有同学来说，我热爱那种协作精神。and also for all classmates, I love that kind of collaboration. 01:04:15 谢谢大家。So thanks. 01:04:16 很好。Nice. 01:04:17 谢谢 Taz。Thanks Taz. 01:04:22 好的，谢谢。Okay, thank you. 01:04:22 Yuzheng，你好，我是 Teh。Yz Hi, I’m Teh. 01:04:25 抱歉，有点紧张，因为 Sorry it feels a bit nervous because 01:04:27 我长期关注你的频道 I’ve been watching your channel for long 01:04:28 现在亲自跟你说话感觉有点怪怪的。so it feels a bit weird to speak to you in person. 01:04:31 嗯，谢谢。But yeah, thank you. 01:04:33 所以我是 Tae，So I’m Tae, 01:04:34 我是中国人，I’m from China 01:04:35 但现在在澳大利亚墨尔本工作。but I’m now working in Melbourne, Australia. 01:04:38 我不确定 I’m not sure 01:04:38 你们中有多少人身在澳大利亚。how many of you are based in Australia. 01:04:40 如果你是的话，很乐意联系。If you do, happy to connect. 01:04:43 我在咨询行业从事财务转型相关工作 And I work in consulting focusing on finance transformation 01:04:46 ，我之所以来这里 and I would say the reason why I’m here 01:04:48 ，是因为我不想再继续在这个领域干下去了。is because I don’t want to work in this field anymore. 01:04:53 所以我想多了解一些生成式 AI So I wanted to learn more about AI 01:04:57 看看职业生涯的下一个阶段 and see where I can move to for the next career, 01:05:01 能转向哪里。for the next stage of my career I suppose. 01:05:04 至于我如何帮助同学 In terms of how I can help my classmate 01:05:08 我想说我没有技术背景。I would say I come from a non technical background. 01:05:13 我的专业背景是金融领域 My background is in finance 01:05:14 所以 so I’m not sure 01:05:15 我 how I can help you guys 01:05:15 不确定自己能帮上什么忙，如果你们大多数人有技术背景。if most of you are from a technical background. 01:05:19 不过，我会说 But I would say 01:05:20 我视自己为一名问题解决者。I consider myself as a problem solver. 01:05:23 我可以瞧瞧。I would see. 01:05:24 我们来发现一个问题 Let’s discover a problem 01:05:26 ，看看我能如何帮助你，我猜 and see how I can help you I suppose 01:05:28 ，而且非常期待与大家建立联系。and really look forward to connecting with you all. 01:05:31 接下来传给下一位同学。I’ll pass on to the next classmate. 01:05:34 很好。Nice. 01:05:35 谢谢你，Brenda。Thank you Brenda. 01:05:39 谢谢 Suun。Thanks Suun. 01:05:41 我叫。My name is. 01:05:41 大家好，我叫 Brenda。Hi everyone, my name is Brenda. 01:05:44 我目前住在旧金山南湾地区。I am based in South Bay San Francisco. 01:05:47 我目前在 LinkedIn 担任数据科学家。I’m currently a data scientist at LinkedIn 01:05:54 有趣的是，我是从银行业转战科技行业的，and fun fact, I broke into a tech industry from a banking industry 01:06:01 在这个过程中，我自学了很多数据科学知识，and then during that process I studied a lot of like how to do data science 01:06:08 还看了 Yujian YouTube 频道上的面试准备视频等课程。and also some courses like interview prep videos from Yujian’s channel on YouTube 01:06:19 我非常感激，从那以后我就一直关注着你，so I really appreciate and since then I’ve been following you 01:06:25 以及你能提供的帮助。and what I can offer. 01:06:31 我可以贡献我的好奇心 I can offer my curiosity 01:06:35 提供互相帮助 and I can offer to help each other 01:06:37 或许一起头脑风暴 maybe brainstorm on things 01:06:41 我可以做个诚实的听众 and I can offer to be an honest audience 01:06:45 分享我的真实看法 to share my honest opinion 01:06:48 我非常非常期待 and I really really look forward 01:06:50 这个动手实践的课程 to this hands on practicing this class 01:06:54 用 AI 构建项目 to build with AI 01:06:57 并更多了解你们每一位。and to learn more about each one of you. 01:07:01 谢谢。Thank you. 01:07:02 很好，谢谢你。Nice, thank you. 01:07:03 你可以听到你对我第一个模块表现的 You get to hear your honest opinion 01:07:07 诚恳看法。about how I did in the first module. 01:07:09 好的，Jeremy。All right Jeremy. 01:07:13 大家好。Hi everyone. 01:07:14 所以，我是 Jeremy。So I’m Jeremy. 01:07:16 我本人是数学专业训练出身 I’m a mathematician by training 01:07:19 博士毕业后 and after my PhD 01:07:20 我在 AI 和医疗保健领域从事了一些年的研究工作。I spent some years doing research in the AI and healthcare. 01:07:25 之后，我在一家位于纽约的 And afterwards I was working at the New York based 01:07:29 量化数据初创公司工作，负责数学建模 quant data startup doing the mathematical modeling 01:07:33 以衡量某些 alpha 信号。to measure some alpha signals. 01:07:38 所以现在我正在创办一家 So now I’m doing a startup 01:07:40 利用 AI 技术的初创公司。using AI. 01:07:42 我试图利用 AI 来提升数学教育 I try to use AI to improve mathematics education 01:07:45 这是我的专业领域。which is my specialty. 01:07:52 这就是我来上这门课的原因，你知道，That’s why I came to this course, you know, 01:07:55 我想更好地学习如何构建东西，try to learn better in terms of building stuff 01:07:58 顺便分享一些我的趣闻。a little bit of fun facts of me. 01:08:03 所以我现在喝咖啡喝得太多 So I drink too much coffee right now 01:08:07 以至于咖啡反而能让我睡得更好。to the extent like coffee can make me sleep better. 01:08:13 不知道我的网络连不连得上。I don’t know if my Internet is okay. 01:08:17 是的。Yeah. 01:08:17 不过是的，这是关于我的趣闻。But yeah, that’s fun facts of me 01:08:21 在这个话题上，我觉得我可以提供 and in terms I think I can offer 01:08:26 一些从学术界关于 AI 的有趣视角，maybe some interesting perspective from academia about the AI, 01:08:30 提出一些关于可解释性的有趣问题，some interesting questions about interpretability 01:08:35 还分享我在 and also some interesting projects that I heard 01:08:36 医疗保健 AI 领域 in the AI in health care 01:08:39 听到的一些有趣项目，并且在课程中进行更多讨论。and also have more discussions along with the course. 01:08:45 嗯。Yeah. 01:08:47 谢谢。Thank you. 01:08:48 很好。Nice. 01:08:48 谢谢。Thank you. 01:08:50 伊莎贝尔。Isabel. 01:08:52 大家好。Hey everyone. 01:08:54 我叫 Isabel My name is Isabel 01:08:55 ，住在西雅图 Great Hill 地区 and I live in Great Hill Seattle area 01:08:58 ，就在 Kudai Bio 附近 just near Kudai Bio area 01:09:01 ，Kudai Bio 离 Issaquah 很近。which is near the Issaquah. 01:09:08 我在 T-Mobile 做 BI 分析师。I work as a BI analyst and T mobile. 01:09:11 我觉得最有趣的是 I think the fun fact is 01:09:13 上周六我参加了 Yuzheng 的活动 last Saturday I just joined Yujun’s event 01:09:17 在 Belleville place，我们终于见面了。in the Belleville place and we met in person. 01:09:23 那真是一个很棒的活动。That’s a really good event. 01:09:25 我决定加入这个社区 I decided to join the community 01:09:27 和这里的各位朋友一起学习 AI。and to learn the AI together with all the friends here. 01:09:32 有趣的是，另一个有趣的事实是，The fun fact is another fun fact is 01:09:36 我觉得两天前，I think two days ago 01:09:38 除了我一月份的工作，besides my work from January 01:09:40 我们还和一些朋友聚在一起。we have some friends together. 01:09:44 我们已经开发了一个 AI 产品 We already built an AI product 01:09:46 就在两天前，我把我的项目发到了我们的社区里。and two days ago I just post my project in our community. 01:09:51 大家可以搜索我发布的 You guys can search for my post 01:09:54 房地产 AI 分析器 which is for real estate AI analyzer 01:09:57 帖子，我欢迎大家试用它。and I appreciate everybody can try it. 01:10:02 我能帮助大家的地方 How I can help with everyone here 01:10:04 在于，我的背景主要是数据分析师。is that my background is more about the data analyst. 01:10:07 所以现在我也在学习：So right now I’m also learning: 01:10:10 我做了大量工作，研究如何利用 AI I did a lot of work on how to use AI 01:10:15 摆脱繁琐的数据处理任务使用 AI，to transition away from tedious data work using AI, 01:10:19 从而把更多时间投入到战略性工作中。so that I can spend more time on strategic work. 01:10:25 所以如果你们当中有数据分析师 So if anybody background in data analyst 01:10:27 或 BI 分析师背景的，我可以提供帮助。or BI analyst I can help. 01:10:30 想想我们该怎么进行这个转变吧。Just think how we can do the transition. 01:10:35 对，就是这样。Yeah, that’s it. 01:10:38 顺便一提，这是个很好的机会。By the way, this is a good opportunity to. 01:10:40 哦，谢谢你。Oh thank you. 01:10:41 介绍我们的社区，To introduce our community 01:10:43 有一个‘分享你的项目’专区，There is a share your project section 01:10:47 如果你在那里发布帖子，and if you post there 01:10:50 不仅会以应用内消息形式发送，it was sent not only in app message 01:10:53 还会通过电子邮件发送给超过 14,000 名社区成员。but also email to over 14,000 community members. 01:11:01 所以，这是一个绝佳方法 So that is a very good way 01:11:04 来获取应用初始用户牵引力和反馈。to get initial traction and feedback for your app. 01:11:09 我记得 Isabel 的 app And I remember Isabel’s app 01:11:12 在 48 小时内就收到了超过 100 条评论。received over I think 100 comments within 48 hours. 01:11:18 是的，这是最成功的发布之一。Yes, one of the most successful launches. 01:11:21 谢谢。Thank you. 01:11:21 感谢你建立这个社区。Thank you for you build the community. 01:11:24 嗯，你应该已经积累了不少早期用户了。Yeah, you probably got a lot of early users already. 01:11:28 所以我想，你能帮助同学们的一件事 So I guess probably one thing you can help our classmates 01:11:31 就是如何进行市场推广 is how to do go to market 01:11:33 以及如何宣传你的应用。and how to promote your app. 01:11:35 是的，那是个很好的观点。Sure, that was a very good point. 01:11:37 我很乐意这么做。I’m happy to do that. 01:11:38 嗯。Yep. 01:11:39 酷。Cool. 01:11:40 你也可以分享 You can also share the link 01:11:42 项目链接在聊天里。of your project in chat as well. 01:11:45 好，谢谢 Jung。Cool, thank you Jung. 01:11:50 嗨，大家好，Hi there everyone, 01:11:51 能听到我的声音吗？can you hear me? 01:11:52 嗯。Yep. 01:11:53 嗯，这是 Jung。Yeah, this is Jung. 01:11:55 我来自山景城。I’m from Mountain View. 01:11:58 我在 Google Photos 做过软件工程师 I worked at the software engineer at Google Photos 01:12:03 ，有趣的是，and the fun fact is 01:12:04 我加入 Google 纯粹是因为热爱摄影，I joined Google just because I love taking photos 01:12:08 总觉得自己在开发给自己用的应用。and I always feel like I’m building application for myself. 01:12:12 那么我能帮上什么忙？And so what can I help? 01:12:14 基本上，如果你有任何关于使用这些应用的问题 Basically if you have any questions about using the apps 01:12:17 ，或者发现了 bug，可以直接给我发消息。or you found any bugs, you could probably send a message to me. 01:12:20 我甚至可以优先为 iOS 进行修复。I could prioritize the fix probably even for iOS. 01:12:26 对我而言，我想加入这个课程的原因 And for me the reason why I want to join this class 01:12:29 很简单：我认为我已经持续使用 AI is just because I think I kind of keep using AI 01:12:33 将近一年了，并且一直试图按照公司的指示行事。for almost over a year and I try to follow the company the instruction. 01:12:37 内部文档也非常多。There are a lot of internal docs as well. 01:12:40 但你知道，在 Google 里你基本上只能用 Gemini But you know in Google you just like kind of limited to the Gemini 01:12:45 ，我可能访问权限有限 and I kind of have probably have limited access 01:12:47 ，或者已经对 Gemini 很熟悉了。or probably kind of already in a circle around with the Gemini. 01:12:50 我想知道别人是怎么用 AI 的 I want to know how the others people are using AI 01:12:54 但每次用 AI 时，我总是感到非常沮丧。and always, always I feel a lot of frustration while using the AI. 01:12:59 现在结果出人意料。The result is now unexpected. 01:13:01 我得进行大量的手动审核工作。I have to do a lot of manual review work. 01:13:03 我必须不停地提示 I have to just keep prompting 01:13:05 才能得到更好的结果 to make it better result 01:13:08 尤其是时间线预期 and especially the timeline expectation 01:13:09 总是比想象中慢。just like always slower than expected. 01:13:12 我必须拼命加班才能 I have to overwork to just like to meet 01:13:15 勉强赶上预设的时间表。my designed timeline. 01:13:19 所以我觉得这些就是 So I think those are the reasons 01:13:20 我来到这里的原因。of why I’m here. 01:13:21 所以我希望能学到很多东西 So I hope could learn a lot of things 01:13:24 和大家一起。with everybody together. 01:13:25 嗯，没错，那就是我的 nips。Yep, that’s my nips. 01:13:29 希望我们能从你这里获得更多配额 Can we get more quota from you 01:13:33 比如存储配额之类的。like storage quota from you hopefully. 01:13:39 所以时间有限 So we have limited time 01:13:40 ，我们有三个选择。but we have three choices. 01:13:43 我们可以选择再做两个介绍 We can do either do like another two introduction 01:13:46 或者让大家加快速度 or we can have everyone do it faster 01:13:49 或者把课延长五分钟左右 or we can extend our class a little bit like five minutes 01:13:52 我想我们结合二和三吧。I guess maybe we’ll do a combination of two and three. 01:13:55 所以大家动作快一点 So everyone be a little faster 01:13:57 我们就把课时稍稍延长一些。and we’ll extend our class a little bit. 01:14:01 阿尔弗雷德。Alfred. 01:14:06 你好。Hi. 01:14:06 你好。Hi. 01:14:06 那我加快速度做吧。So I’ll do it faster. 01:14:07 大家听得清楚吗？Can you hear me clear? 01:14:08 嗯。Yep. 01:14:09 所以我现在在加拿大不列颠哥伦比亚省的温哥华。So I am in Vancouver B.C. 01:14:11 所以我在加拿大。so in Canada. 01:14:12 所以它真的非常接近 reckon。So it’s really close to reckon. 01:14:15 所以我没有技术背景，So I am a non technical background, 01:14:17 我在 TD 商业银行担任账户经理。I work in a TD commercial banking as an account manager there. 01:14:21 所以我帮助企业获得商业贷款 So I help businesses to get their business loan 01:14:25 、抵押贷款、actually to get their mortgage or business mortgage business loan, 01:14:27 营运资金等融资支持。working capital, something like that. 01:14:31 所以这个有趣的事实是 So the fun Fact for this is 01:14:34 我的这位朋友曾在 XAI 担任高级软件工程师 because my friend actually work in as a senior software engineer in XAI 01:14:39 现在他正在 Y Combinator 旗下的一个营销公司工作。and then he’s now working in one of the YC dispatch actually Y Combinator 01:14:47 所以他偶尔也会做一些 AI 项目 So now and then he also building some AI things 01:14:50 然后他就跟我说，哦，对，就加入这个吧。so he just tell me like oh yeah, just join this. 01:14:54 就像他长期关注你们的社区 Like he follow your community for so long 01:14:57 然后觉得你们能帮我 and then he thinks that you guys can help me 01:14:59 填补空白，因为他给我举了个很简短的例子。to fill the gap because he gave me a really quick example. 01:15:03 就像是，哦，他 10 级，It’s like oh he’s level 10, 01:15:05 你们 20 级，you guys are level 20, 01:15:06 我才 1 级而已。I’m just level one. 01:15:07 所以他根本没时间 So he just don’t have time 01:15:08 让我去了解它。to let me learn about it. 01:15:10 所以这就是他把我介绍到这个课程的原因。So that’s why he introduced me to this course. 01:15:13 我知道它不便宜 I know it’s not that cheap 01:15:14 ，但我觉得真的很值得 but then I think it’s really worth it 01:15:15 ，因为我有一些想法。because I have some idea. 01:15:17 我也在学习加拿大移民顾问的相关内容。I’m also studying Canada immigration consultant too. 01:15:23 所以因为我已经在加拿大 So because I have been to Canada 01:15:25 待了 10 年从 2016 年起。for 10 years since 2016. 01:15:28 所以我也在尝试建立一个一人公司 So I’m also trying to build like a one man company 01:15:32 在未来毕业后，那叫什么来着？in the future once I graduated with what’s that called? 01:15:37 移民。Immigration. 01:15:38 所以，如果你有任何关于移民加拿大的问题，So if you have any immigration problem questions to Canada, 01:15:41 随时告诉我。let me know. 01:15:43 我正在研究那个。I’m studying that. 01:15:44 另外，我对股票交易 And also I’m really interested in stock trading 01:15:46 或波段交易特别感兴趣。or swing trading. 01:15:48 所以我想用 AI 来帮助我 So I want to use AI to help me 01:15:50 开发 AI 交易工具 to build AI trading tools 01:15:52 比如跟踪 Donald Trump 的情绪分析。like also sentiments on tracking Donald Trump. 01:15:56 没错，那里的一切。Right Everything in there. 01:15:57 所以我知道你们也聊金融 So I know you guys also talk about finance 01:15:59 话题在 YouTube 上。too in the YouTube. 01:16:01 嗯。Yeah. 01:16:02 所以我对这个 cohort 非常兴奋 So I’m really excited to this cohort 01:16:04 然后我们可以互相认识 and then get to know each other 01:16:06 也一起参与 and also participate in 01:16:07 ……抱歉，花了太长时间。Sorry it takes too long. 01:16:09 不过嗯，有事告诉我。But then yeah, let me know. 01:16:10 嗯。Yeah. 01:16:11 见到大家真高兴。So nice to see everyone. 01:16:12 嗯，好，谢谢。Yeah, nice, thank you. 01:16:14 听到大家的口碑传播 And it’s so great to hear the word of mouth 01:16:16 真是让我太高兴了！like makes me so happy. 01:16:18 谢谢。Thank you. 01:16:19 嗯。Yeah. 01:16:19 你没加班 You didn’t join the class 01:16:20 但它在社区里。but it’s in the community. 01:16:22 嗯嗯。Yeah, yeah. 01:16:26 Yanxian。Yanxian. 01:16:27 大家好，Hey everyone, 01:16:28 我是西雅图的 Yan Xian。this is Yan Xian here in Seattle. 01:16:30 我是一名消防工程师 I’m a fire protection engineer 01:16:32 ，因此可以说属于非技术背景 so we can say it’s a non tech background 01:16:34 ，不过我们在亚马逊工作。but we work at Amazon. 01:16:38 所以对我来说，最主要的是设计数据中心。So primary things is for me is a design data center 01:16:42 我迫切需要 AI 帮助的原因是，and whatever the reason why I need to desperately need some AI help 01:16:45 有数百个项目和数百个工作流程需要处理，is just hundreds of projects I need to handle and hundreds of workflow 01:16:49 这对人类来说已经变得不可能了。that is a human aim becoming impossible. 01:16:53 当然，我 And of course I have a little bit bandwidth 01:16:53 有两个小孩、一只狗还有家庭，时间精力有限。with two kids and one dog and family. 01:16:57 我只是迫切需要一点帮助 I just desperately need a little bit of help 01:16:59 在家庭方面。on my family side as well. 01:17:01 这就是全部的原因。That’s the whole reason. 01:17:02 那对我来说是个有趣的事实吧 That’ll be a fun fact for me I guess 01:17:04 我之前在 Tesla 工厂工作过 working at the Tesla factory before 01:17:07 有机会近距离两英尺看到 Elon Musk so did have opportunity to see Elon Musk like two feet away 01:17:12 但没想到他的职责竟然这么重大。but accident that duty is huge. 01:17:16 所以如果你没机会 So if you never get a chance 01:17:17 当面见到他，to see him in face to face, 01:17:18 别担心，他是可以被吓唬的。just be he can be intimidated. 01:17:21 所以这是真的。So that’s true. 01:17:23 关于我能为班级贡献什么，Regarding on what I can offer to the class, 01:17:26 我觉得我可以交流想法，I think I can bounce off ideas 01:17:30 并分享一些批判性思考。and also can share some critical thinking. 01:17:33 如果你在特斯拉工作期间有一些想法，If you have some ideas during time at Tesla, 01:17:36 请真正将第一性原理思维灌输到我的头脑中。really drill the first principle thinking into my head. 01:17:40 所以这也许会有所帮助。So maybe that could be helpful. 01:17:42 谢谢。Thanks. 01:17:43 很好。Nice. 01:17:43 谢谢。Thank you. 01:17:44 我们一定能帮到你。We can definitely help you. 01:17:46 楼。Lou. 01:17:47 大家好。Hi everyone. 01:17:49 我是 Lu，住在 San Jose。I’m Lu and based in San Jose. 01:17:52 我之前在一家 SaaS 公司从事产品分析工作 I worked in product analytics at a SaaS company 01:17:56 是的，这个行业目前正被 AI 变革冲击得非常迅速 and yeah it’s an industry getting hit by AI change very quickly right now 01:18:01 这也是我加入这个课程的一个主要原因。and that’s one big reason I joined this class. 01:18:05 我想更深入地了解为什么会出现这种情况 I want to better understand like why this is happening 01:18:09 ，以及 AI 将如何改变我的工作和生活。and how AI could change both my work and my life. 01:18:13 有趣的是，And a fun fact about me is that 01:18:15 我现在正和我的猫咪一起看这个课程。I’m watching this course with my cat by my side now 01:18:21 我很乐意和同学们一起讨论作业，and I would be very happy to support classmates 01:18:26 互相支持，一路共同学习。by discussing the homework together and also learning from each other along the way. 01:18:30 嗯。Yeah. 01:18:30 很好。Nice. 01:18:31 谢谢你，Bo。Thank you Bo. 01:18:36 大家好，我是 Bo。Hi everyone, I’m Bo. 01:18:38 我目前住在旧金山。I’m currently based in S.F. 01:18:40 加利福尼亚州，三藩市。San Francisco, California. 01:18:42 我在苹果公司做艺术总监 I work as an art director at Apple 01:18:44 的市场团队里工作。for the marketing team. 01:18:46 因此，我们为产品上市设计营销宣传活动。So we design marketing campaigns for the product launches. 01:18:52 关于我的一件趣闻：Fun fact about me: 01:18:54 我最初学的是软件工程，I actually started off studying software engineering, 01:18:57 但超级讨厌，I hated it, 01:18:58 后来没想到又绕回来了。and who would have known—full circle. 01:19:01 目前我能帮忙的，就是 Right now what I can offer is maybe 01:19:04 如果你有设计方面的问题，我可以提供协助。if you have some design questions I can help with that. 01:19:09 就这样了。So that’s it. 01:19:11 谢谢，Sasha。Thank you Sasha. 01:19:14 大家好，我是 Sasha。Hello everyone, I’m Sasha. 01:19:17 我是 I’m a data scientist 01:19:17 这家电力公司的一名 working at an electric company 01:19:21 数据科学家。called Pacific Gas and Electricity. 01:19:26 所以，在来这里工作之前 So and before working here 01:19:29 我在伯克利劳伦斯国家实验室做过博士后研究 I did my postdoc at the Berkeley Lawrence National Lab 01:19:35 主要专注于构建机器学习模型和数值模拟 focused on building machine learning models, numerical simulations 01:19:40 用来求解物理方程或耦合偏微分方程。basically to solve physical equations or coupled partial differential equations. 01:19:46 关于我的一些趣闻：Fun facts about me: 01:19:48 我关注 Kadabia 的 YouTube 频道 I have been following actually Kadabia’s YouTube channel 01:19:52 已经有一段时间了，目前我最喜欢的是 for a while, and so far my favorite one is 01:19:55 《Influence with Authority》 Influence with Authority. 01:20:00 虽然它很老 Even though it’s really old 01:20:01 ，而且之后有很棒的材料 and it had really great materials after that 01:20:03 ，但它仍然是我最喜欢的。but still that’s my favorite. 01:20:05 关于我还有个有趣的事实：Another fun fact about me is that 01:20:07 我有很多想法。I have lots of ideas. 01:20:08 我有点是个爱做白日梦的人。I kind of a daydreamer. 01:20:10 我脑子里有很多点子，I have tons of ideas, 01:20:12 但点子太廉价了，but ideas are cheap 01:20:13 所以我总缺乏足够的动力 so I always don’t have enough momentum 01:20:16 把它们付诸实现。to make the ideas come true. 01:20:18 所以我觉得我需要来上这堂课 So I’m just, I feel like I need to come to this class 01:20:21 汲取足够的能量，去开发一些 and get enough energy to create some apps, 01:20:24 有趣好玩的 app。some fun, interesting apps. 01:20:26 我能贡献些什么？What can I contribute? 01:20:30 一是，我很乐意 One is just I’m very happy 01:20:32 和同学们一起做作业。to engage and work on the homeworks with our classmates. 01:20:36 另外，如果你想了解更多 Another thing is if you’re interested in knowing more 01:20:38 关于能源公司的情况，我很乐意分享。about like Energy company. 01:20:43 很高兴交流，很高兴分享。Happy to talk, happy to share. 01:20:46 很好。Nice. 01:20:47 而这个方程式彻底颠倒了。And the equation completely flipped. 01:20:49 所以如今，想法不再廉价了。So now ideas are not cheap. 01:20:51 我们很多人羡慕那些 And a lot of us are envious of people 01:20:53 idea 丰富、能迅速将它们变为现实的人。who have a lot of ideas, and you can quickly turn those into reality. 01:20:58 好的，那应该就是所有想分享的人了，对吗？Great, so that’s probably everybody who wants to share, right?

幻灯片：

Module 1.pdf 8.41 MB

最后更新：2026 年 5 月 7 日

English Original

Slide deck:

Module 1.pdf 8.41 MB

Last updated: May 7th, 2026

Zayn's Notes

Explorer

Ch01-L01 模块1 理解生成式AI的力量

模块 #1：理解生成式 AI 的力量及其驾驭之道

我的理解

相关链接

原文

Graph View

Table of Contents

Backlinks