【AI思维】给我18分钟，让你聪明到危险 · 顶尖1%用AI
共 2 集
剧集目录
| 集 | 标题 | 时长 |
|---|---|---|
| P1 | 【AI思维】给我18分钟，让你聪明到危险 · 顶尖1%用AI | 18:57 |
| P2 | 【AI思维】给我18分钟，让你聪明到危险 · 顶尖1%用AI | 18:57 |
P1 · 【AI思维】给我18分钟，让你聪明到危险 · 顶尖1%用AI的4个秘密 p01 中英字幕 (P1)
字幕摘录
| 时间 | 英文 | 中文 |
|---|---|---|
| 0:00 | Most people are letting AI destroy their ability to think, | 大多数人都在让AI破坏他们思考的能力, |
| 0:04 | training AI to become their own replacement. | 训练AI成为他们自己的替补. |
| 0:06 | Tragic, because AI can make you dangerously intelligent. | 悲剧,因为AI可以让你危险地聪明. |
| 0:10 | I went from being homeless to an MIT grad | 从无家可归到麻省理工学院 |
| 0:13 | and running and advising AI companies worth billions. | 经营和向AI公司提供建议 价值数十亿美元。 |
| 0:16 | And here's what I've learned. | 且道我所学. |
| 0:18 | The top 1% use AI backwards. | 前1%使用AI反向. |
| 0:21 | They don't prompt to get answers. | 他们不会迅速得到答案。 |
| 0:23 | They use it to train their brain | 他们用它来训练大脑 |
| 0:25 | and outsmart almost any situation. | 和聪明的几乎任何情况。 |
展开字幕全文（408 条）
| 序号 | 英文 | 中文 |
|---|---|---|
| 1 | Most people are letting AI destroy their ability to think, | 大多数人都在让AI破坏他们思考的能力, |
| 2 | training AI to become their own replacement. | 训练AI成为他们自己的替补. |
| 3 | Tragic, because AI can make you dangerously intelligent. | 悲剧,因为AI可以让你危险地聪明. |
| 4 | I went from being homeless to an MIT grad | 从无家可归到麻省理工学院 |
| 5 | and running and advising AI companies worth billions. | 经营和向AI公司提供建议 价值数十亿美元。 |
| 6 | And here's what I've learned. | 且道我所学. |
| 7 | The top 1% use AI backwards. | 前1%使用AI反向. |
| 8 | They don't prompt to get answers. | 他们不会迅速得到答案。 |
| 9 | They use it to train their brain | 他们用它来训练大脑 |
| 10 | and outsmart almost any situation. | 和聪明的几乎任何情况。 |
| 11 | So in this video, | 所以在这段录像里 |
| 12 | I'll break down a counter-intuitive system | 我会打破一个反直觉系统 |
| 13 | the top 1% use to get smarter, faster with AI. | 前1%用来提高智能 更快 |
| 14 | Here is the four-step framework. | 以下是四步框架。 |
| 15 | Step one, intelligent laziness. | 第一步 聪明的懒惰 |
| 16 | A study in Harvard Business Review | 哈佛商业评论研究 |
| 17 | found that CEOs waste 72% of their time | 发现CEO浪费了72%的时间 |
| 18 | in meetings that don't move the needle. | 在不移动针头的会议。 |
| 19 | We've all experienced those meetings, haven't we? | 我们都经历过那些会议,不是吗? |
| 20 | The one-hour meeting that needed only 15 minutes | 只需15分钟的一小时会议 |
| 21 | to get to a decision. | 来决定 |
| 22 | But it's hard to stop. | 然其难止. |
| 23 | So why do some of the most accomplished folks | 所以,为什么做一些 最成就的人 |
| 24 | feel trapped this way? | 感觉被困住了吗? |
| 25 | Because we all suffer from this biological glitch | 因为我们都有这种生物缺陷 |
| 26 | called completion bias. | 名为完成偏差。 |
| 27 | Your brain is wired to seek an immediate dopamine hit | 你的大脑被电线连接到 寻求立即的多巴胺命中 |
| 28 | that you get from finishing a task. | 从你完成一个任务得到的。 |
| 29 | So we end up treating all tasks as equal | 所以,我们最终对待所有的任务是平等的 |
| 30 | because we're gonna get roughly the same amount of dopamine | 因为我们得到的多巴胺量大致相同 |
| 31 | when you spend time on redrafting an internal email | 当你花时间重新起草内部邮件时 |
| 32 | or a million dollar strategy document. | 或者一百万美元的战略文件。 |
| 33 | Everything is priority one. | 一切都是第一要务。 |
| 34 | So none of it is. | 故无一是. |
| 35 | So how do you avoid this priority blindness? | 那你如何避免这种优先失明? |
| 36 | A good way to think about tasks is to see two curves. | 思考任务的好方法就是看到两个曲线. |
| 37 | First curve has capped payoffs. | 第一曲线有上限的回报。 |
| 38 | This curve goes up and then flattens out | 这个曲线向上,然后平整 |
| 39 | once it reaches the zone of diminishing returns. | 一旦它到达了不断减少的回报区。 |
| 40 | So tasks like formatting slides or internal emails, | 因此,诸如格式化幻灯片或内部电子邮件等任务, |
| 41 | expense reports, FYI meetings, | 支出报告、财务会议、 |
| 42 | what happens if you spend additional effort | 如果你付出更多努力会怎么样? |
| 43 | to make the outcome of these tasks pitch perfect? | 使这些任务的结果完美无缺? |
| 44 | Nothing. | 没什么 |
| 45 | There's no upside here | 这里没有颠倒 |
| 46 | because the value flatlines after a point. | 因为值在一点后会平线。 |
| 47 | Nobody cares if you spend hours choosing better fonts | 没人在乎你有没有花几个小时选择更好的字体 |
| 48 | or breathtaking designs in internal slides | 或内部幻灯片中的呼吸设计 |
| 49 | that are seen for six minutes. | 6分钟内就能见到 |
| 50 | This curve shows you your zone of intelligent laziness. | 这个曲线显示你的智慧懒惰度 |
| 51 | There was a Nobel Prize winning economist | 有个诺贝尔奖得主经济学家 |
| 52 | and computer scientist and his name was Herbert Simon. | 电脑科学家叫赫伯特・西蒙 |
| 53 | And he came up with a concept called satisfying, | 他提出一个叫满足的概念, |
| 54 | which pretty much means stop when it's good enough. | 这几乎意味着停止 当它足够好。 |
| 55 | Satisfy and suffice, satisfies. | 满足和满足,满足。 |
| 56 | Now our second curve is the exact opposite. | 现在我们的第二个曲线正好相反。 |
| 57 | It has uncapped payoff. | 它有未封顶的回报。 |
| 58 | This curve stays flat for a long time, | 这个曲线很长一段时间保持平坦 |
| 59 | but then goes to the moon in a hurry. | 却匆匆赶到月亮上。 |
| 60 | These are tasks like customer interactions, | 这些是客户互动等任务 |
| 61 | product design, pricing model, | 产品设计、定价模式、 |
| 62 | finding a co-founder or a life partner. | 寻找共同创始人或生活伴侣. |
| 63 | Being 1% better here does not yield 1% better result. | 这里的1%好,不会产生1%好的结果。 |
| 64 | It actually solves the rest of the 99% of your problems. | 它能解决你99%的问题 |
| 65 | Pour your soul into this. | 把你的灵魂注入这个。 |
| 66 | Johnny Ive would obsess for many months | 强尼 我会迷恋好几个月 |
| 67 | on even the internal component design of iPhone. | 甚至连iPhone的内部组件设计。 |
| 68 | But you know, Steve Jobs never said, | 但是你知道,史蒂夫・乔布斯从来没有说过, |
| 69 | hey, this is costing us a lot of money | 嘿,这花了我们很多钱 |
| 70 | 谁会打开iPhone? | |
| 71 | But Steve knew this was the second curve. | 但史蒂夫知道这是第二个曲线。 |
| 72 | So if the first curve is your zone of laziness, | 所以如果第一个曲线是你的懒惰区 |
| 73 | your second curve is your zone of obsession. | 你的第二个弯曲是你的痴迷区域。 |
| 74 | Let's talk about how AI can help. | 让我们来谈谈AI可以帮助。 |
| 75 | The top 1% use AI on zone one or the zone of laziness. | 前1%在第一区或懒惰区使用AI. |
| 76 | The more they outsource zone one to AI, | 他们越是把第一区外包给AI |
| 77 | the more they can focus on zone two, the zone of obsession. | 他们越是专注于第二区, 痴迷区。 |
| 78 | So how do I decide what to outsource to AI and when? | 那么,我如何决定什么是外包给AI和何时? |
| 79 | So for that, I use a very simple framework | 因此,我用一个很简单的框架 |
| 80 | 称为 Drag Framework, D- R- A- G. | |
| 81 | Four categories of work you immediately should delegate | 四类工作你应立即委托 |
| 82 | to AI so you can stay in your zone of obsession. | 这样你就可以留在你的痴迷区域 |
| 83 | First, D equals drafting. | 第一,D等于起草。 |
| 84 | This is the blank page problem we all face. | 这是我们面临的空白页问题。 |
| 85 | It's hardest to get from zero to one sometimes. | 有时从零到一很难做到. |
| 86 | AI can help here, tremendously actually. | AI可以在这里帮助, 巨大的实际上。 |
| 87 | Give it a prompt using the aim protocol | 使用目标协议迅速启动 |
| 88 | that I've shared before. | 我之前分享过 |
| 89 | Hey AI, act in this role, use this input | 嘿,AI,做这个角色,使用这个输入 |
| 90 | and this is your mission, A-I-M. | 这是你的使命,A -I -M。 |
| 91 | And that way you get started very quickly | 那样你就能很快开始 |
| 92 | on that email or code or presentation. | 在电子邮件或代码 或演示文稿。 |
| 93 | And the first draft from AI will be crappy and atrocious, | AI的第一稿会很烂很凶残 |
| 94 | but that's fine. | 但没关系。 |
| 95 | Now you have a starting point. | 现在你有一个起点。 |
| 96 | You're not staring at a blank page anymore. | 你不再盯着空白的一页了 |
| 97 | Now it'll trigger something in your brain | 现在它会触发你脑中的东西 |
| 98 | and you're off to the races. | 你将去比赛。 |
| 99 | R equals research. | R等于研究. |
| 100 | This helps you solve the information overload problem. | 这有助于解决信息超载问题. |
| 101 | Today, if something requires deep research, | 今天,如果有什么需要深入的研究, |
| 102 | it can be dramatically accelerated using AI. | 使用人工智能可以大大加速。 |
| 103 | Summarization, extraction, competitive intel. | 总结,提取,竞争性情报. |
| 104 | You know, don't spend time doing that kind of research. | 你知道,不要花时间 做这种研究。 |
| 105 | Let your friendly neighborhood AI do it for you. | 让你的友好邻居AI为你做。 |
| 106 | When you use the deep research feature | 当你使用深层研究特征 |
| 107 | on Chad GPT or Gemini or Claude, | 在乍得GPT或双子座或克劳德, |
| 108 | it fires off hundreds of secondary search queries. | 它点燃了数百个二级搜索查询 |
| 109 | It goes out to the web like a spider | 像蜘蛛一样走进网络 |
| 110 | and finds hundreds of sites, consolidates the results, | 并发现数百个网站, |
| 111 | even checks his own work by asking what's missing | 连自己的作品都查过了 问出什么了 |
| 112 | and follows up on its own | 并独自跟进 |
| 113 | to finally deliver a rich document to you. | 终于给你送了一份丰富的文件 |
| 114 | It's like you just hired a consultant | 好像你刚雇了个顾问 |
| 115 | for a week long research project, | 为期一周的研究项目, |
| 116 | but instead you get there in 10 minutes. | 但你10分钟后就到了 |
| 117 | Third is A for analysis. | 第三是A分析. |
| 118 | Let AI take the first pass at analyzing, | 让大赦国际在分析时先通过 |
| 119 | summarizing, reasoning, | 总结,推理, |
| 120 | especially if it's all unstructured data | 特别是如果都是无结构的数据 |
| 121 | because AI is gonna find patterns | 因为AI会找到规律 |
| 122 | that we humans aren't gonna be able to. | 我们人类是无法做到的 |
| 123 | So use it for your advantage. | 所以用它来帮你 |
| 124 | And finally G is for all the grunt work. | 最后G是所有的咕噜声工作。 |
| 125 | Tasks like reformatting, translating, tabulating, | 重排、翻译、制表、 |
| 126 | cleaning data and on and on, the boring manual work. | 清理数据 继续 无聊的手工工作 |
| 127 | Just give it to AI. | 只要把它给AI。 |
| 128 | So what's the key principle behind drag? | 拖动背后的关键原则是什么? |
| 129 | Apply it only when you are in your zone one, | 只有在你进入第一区的时候才应用 |
| 130 | that first curve. | 第一个曲线。 |
| 131 | If it requires human interaction or judgment or intuition | 如果它需要人类的互动或判断或直觉 |
| 132 | or decision-making or tastes, that's curve two. | 或决策或口味, 这是曲线二。 |
| 133 | That you've gotta do it yourself. | 你得自己来 |
| 134 | But you know, I've found that 70 or 80% | 但你知道,我发现,70%或80% |
| 135 | of my repetitive tasks tend to be in zone one. | 我重复的任务往往在第一区 |
| 136 | And you might find that too. | 你也会找到的 |
| 137 | So be lazy when you can use drag. | 所以,懒惰 当你可以使用拖。 |
| 138 | Be obsessed for everything else. | 执着一切众生. |
| 139 | Step two, the intelligent hill. | 第二步,智慧山头. |
| 140 | For 300 years, Isaac Newton convinced us | 三百年来,艾萨克牛顿说服我们 |
| 141 | that universe was a clockwork machine, | 宇宙是时钟工作机, |
| 142 | predictable and certain. | 可预测和确定。 |
| 143 | But in 1927, another scientist named Heisenberg | 但是在1927年,另一位名叫海森堡的科学家 |
| 144 | shattered those classical beliefs. | 粉碎了这些古典信仰。 |
| 145 | He showed that our universe exists only | 他证明我们的宇宙只存在 |
| 146 | as a cloud of possibilities at quantum level. | 作为量子层面上的一团可能性 |
| 147 | It was a profound shift. | 这是一个深刻的转变。 |
| 148 | You and I have to make a similar shift | 咱俩得做个类似的转变 |
| 149 | when we use AI nowadays. | 当我们使用AI现在。 |
| 150 | The first trick is to stop treating AI like a calculator. | 第一个技巧是停止将AI当作计算器对待. |
| 151 | We like to live in a world with clear rules. | 我们喜欢生活在一个有明确规则的世界。 |
| 152 | You type two plus two into a calculator and you get four. | 把2加2打入计算器,然后得到4个. |
| 153 | Always, it's predictable. | 总是可以预测的 |
| 154 | But AI is not a calculator. | 但AI不是计算器. |
| 155 | It's a probability engine. | 这是一个概率引擎。 |
| 156 | If you ask the same question to AI again, | 如果你再次向大赦国际提出同样的问题, |
| 157 | it'll give you a completely different answer. | 它会给你一个完全不同的答案。 |
| 158 | It'll happily make things up for you | 这会给你带来幸福 |
| 159 | unless you ask it to verify. | 除非你要求它核实。 |
| 160 | AI is brilliant on some days, confused on others. | AI在一些日子很聪明,在另一些日子很困惑. |
| 161 | But on any given day, it refuses to admit | 但是在任何一天 它拒绝承认 |
| 162 | that it doesn't know the answer. | 说它不知道答案 |
| 163 | It loves to make things up. | 它喜欢编造东西。 |
| 164 | So you don't just ask AI the way you ask | 所以,你不只是问AI 你问的方式 |
| 165 | a normal human being. | 一个正常的人。 |
| 166 | You have to architect your questions very carefully. | 你得仔细设计你的问题 |
| 167 | Now, most people use a tactic called zero-shot prompting. | 现在,大多数人都使用一种叫做零镜头提示的战术. |
| 168 | So for example, they would ask, | 例如,他们会问, |
| 169 | give me the best new business idea. | 给我一个最好的新商业点子 |
| 170 | And of course, AI will dish out a response | 当然,AI会给出一个回应 |
| 171 | and tell you why it's the greatest idea in the world. | 告诉你为什么这是世界上最伟大的想法。 |
| 172 | But you're literally rolling the dice and looking to win. | 但你真的在掷骰子 想赢 |
| 173 | To get elite results though, | 为了获得精英成果 |
| 174 | you must climb the intelligent hill. | 你必须爬上智慧山 |
| 175 | There are four camps on the way. | 途中有四个阵营. |
| 176 | Each camp will show you a different way to work with AI. | 每个营地会给你们展示一个不同的与AI合作的方式. |
| 177 | Our first camp is called one-shot prompting. | 我们的第一个阵营叫做一弹一弹 |
| 178 | When you prompt, give one clear example | 提示时,请举一个明显的例子 |
| 179 | so the model doesn't guess blindly. | 所以模型不会盲目猜测 |
| 180 | So the prompt would look like, | 所以,提示会看起来像, |
| 181 | write a LinkedIn post about remote work. | 撰写关于远程工作的LinkedIn文章。 |
| 182 | Use this specific post as a style guide. | 使用此特定帖子作为样式引导 。 |
| 183 | And so give it a post, give it an example | 所以给它一个帖子,举个例子 |
| 184 | and paste that post in the prompt as a reference. | 贴在标签上作为参考 |
| 185 | And that simple act is already an upgrade | 那简单的行为已经升级了 |
| 186 | than rolling the dice blindly. | 而不是盲目地掷骰子。 |
| 187 | Second camp, few-shot prompting. | 第二营,几发子弹 |
| 188 | Now here you give AI three or more examples | 现在请举出三个或更多例子 |
| 189 | so it can find patterns of style and substance | 这样它就能找到风格和实质的规律 |
| 190 | and tone that you desire. | 和你想要的语气。 |
| 191 | Attach documents, links, data, or your prior work. | 附加文档、链接、数据或您先前的工作。 |
| 192 | This is called grounding the model. | 这叫做打地基模型。 |
| 193 | So basically it stops fantasizing and hallucinating | 所以基本上它停止了幻想和幻觉 |
| 194 | and gets grounded to reality. | 被现实所束缚 |
| 195 | Here's an example of a prompt. | 以下是一个提示的例子. |
| 196 | Here are the five of my previous presentations. | 以下是我先前介绍的五点。 |
| 197 | And now write a new presentation | 现在写一个新的演示文稿 |
| 198 | based on my tone of voice on topic XYZ. | 根据我对主题XYZ的声音 |
| 199 | And here's a pro tip. | 这是专业小费 |
| 200 | Ask the AI to explain the pattern back to you first. | 请AI先向您解释模式. |
| 201 | That way AI is forced to articulate what it's doing | 这样一来,AI就被迫阐述它在做什么 |
| 202 | and more importantly, you're forced to learn | 更重要的是,你被迫学习 |
| 203 | how your brain works. | 你的大脑如何工作。 |
| 204 | How did it come up with those patterns? | 怎么会有这种模式? |
| 205 | Now you're being smart about being smart. | 现在,你对聪明很聪明。 |
| 206 | Now let's move to the third camp. | 今移三营. |
| 207 | This one is called chain of thought reasoning. | 这个叫做思想推理链. |
| 208 | Again, fancy name, but the idea is simple. | 再说一遍 名气不错 但想法很简单 |
| 209 | Ask the model to think long and hard before it responds. | 问模型在回应前要长时间而努力思考. |
| 210 | Your job is to slow AI down and force explicit clarity | 你的工作就是减慢AI的速度 强迫它清晰清晰 |
| 211 | by asking it to show its work. | 要求它展示它的工作。 |
| 212 | That's all there is. | 仅此而已. |
| 213 | This is also a good way to reduce hallucinations, of course. | 这当然也是减少幻觉的好方法。 |
| 214 | So let's say you're working on some report | 假设你在做一些报告 |
| 215 | and so you attach it and write a prompt | 并写出一个提示 |
| 216 | that could look like this. | 看起来像这样 |
| 217 | Do not refine my research report yet. | 还没有完善我的调查报告。 |
| 218 | List the top three most impactful areas of improvement | 列出三个影响最大的改进领域 |
| 219 | after you analyze it. | 在你分析出来之后 |
| 220 | Tell me why you think so and suggest how we address each. | 告诉我你为什么这么认为 并提议我们如何讨论每个问题 |
| 221 | Think step by step, show me your thinking for each step. | 一步一步的思考, 给我看看你对每一步的想法。 |
| 222 | That last line is the most important one. | 最后一行是最重要的一条。 |
| 223 | And our fourth and final camp is agents. | 我们的第四个也是最后一个营地是特工 |
| 224 | According to Salesforce, AI agents help drive $67 billion | 根据销售部队,AI代理 帮助驱动670亿美元 |
| 225 | in global sales during cyber week alone. | 仅网络周全球销售量。 |
| 226 | So agents are already here. | 特工已经来了 |
| 227 | The best way to think about agents | 考虑特工的最佳方法 |
| 228 | is to think about who you would hire for a task. | 想想你会雇谁来做任务 |
| 229 | So let's say if you wanted to hire a researcher, | 这么说吧 如果你想雇个研究员 |
| 230 | an analyst, and a copywriter, | 分析师和抄写员 |
| 231 | you can do that with a single agentic prompt | 单个代理提示就可以做到这一点 |
| 232 | that looks like this. | 这看起来像这个。 |
| 233 | Do deep research on trends on topic XYZ, | 对主题XYZ的趋势进行深入研究, |
| 234 | analyze and cross-reference all the trends | 分析和对照所有趋势 |
| 235 | to find the three most important ones | 找到三个最重要的 |
| 236 | and draft a one-page memo summarizing the findings. | 并起草一份一页的备忘录,总结调查结果。 |
| 237 | Now, what is actionable? | 现在,什么是可行动? |
| 238 | Try this framework tonight. | 今晚试试这个框架 |
| 239 | Open your favorite AI app and take any prompt | 打开你最喜欢的 AI 应用程序,并迅速 |
| 240 | that you were about to use. | 你将要使用。 |
| 241 | Just try to get to the next camp. | 试着去下一个营地 |
| 242 | That's how you start climbing up the intelligent hill. | 这就是你开始爬上智慧山的方法. |
| 243 | Remember, when you were dealing with a drunk genius, | 记住,当你面对一个醉酒的天才, |
| 244 | make sure you were the one driving the car. | 确保开车的是你 |
| 245 | So now at this point, everything we have done | 所以现在,我们所做的一切 |
| 246 | has made you fast and efficient. | 使你迅速而高效。 |
| 247 | You're delegating better, you're prompting smarter, | 你授权的更好, 你激励聪明, |
| 248 | you're moving up the hill, | 你移动上山, |
| 249 | and there is less friction than before. | 而摩擦比以前少。 |
| 250 | And that's exactly where most people would stop. | 而这正是大多数人会停止的地方。 |
| 251 | But here's the plot twist. | 但这里的情节曲折。 |
| 252 | The top 1% go one step further. | 前1%更进一步。 |
| 253 | They slow things down deliberately. | 他们故意减慢速度 |
| 254 | Why is that important? | 为什么这很重要? |
| 255 | The trick that top 1% know is this. | 最高的1%知道这个 |
| 256 | They know when to shift the gear | 他们知道什么时候该换档 |
| 257 | because long-term intelligence | 因为长期智能 |
| 258 | isn't built through convenience, | 不是通过方便建造的, |
| 259 | it's built through resistance. | 它是通过抵抗建造的。 |
| 260 | And that's why we need to go to step three. | 这就是为什么我们需要去第三步。 |
| 261 | The intelligent gym. | 智能体育馆. |
| 262 | Most people use AI as a wheelchair for the mind. | 大多数人使用AI作为心灵的轮椅. |
| 263 | And if you sit in a wheelchair when you can still walk, | 如果你坐在轮椅上还能走路 |
| 264 | eventually your legs stop working. | 最终你的腿停止工作。 |
| 265 | Atrophy. | 萎缩. |
| 266 | And today is happening faster | 今天发生得更快 |
| 267 | than at any point in human history. | 比人类历史上的任何时刻都重要 |
| 268 | But the top 1% use a very different principle. | 但前1%使用非常不同的原理. |
| 269 | For information tasks, use AI to remove friction. | 对于信息任务,使用AI去除摩擦. |
| 270 | For transformation tasks, use AI to add friction. | 对于转化任务,使用AI添加摩擦. |
| 271 | When you go to a physical gym, | 当你去健身房时 |
| 272 | we all know how muscles are built, right? | 我们都知道肌肉是如何形成的,对吧? |
| 273 | Through resistance. | 通过抵抗。 |
| 274 | You lift increasingly heavier weights | 你举重越来越重 |
| 275 | to introduce wear and tear to your muscle fibers, | 介绍你的肌肉纤维磨损, |
| 276 | so they break and they grow back stronger. | 所以它们会断裂,会变强 |
| 277 | That is called progressive overload. | 这被称为逐步超载。 |
| 278 | But when it comes to our minds, | 但是,当涉及到我们的想法, |
| 279 | we do the exact opposite somehow. | 我们这样做完全相反。 |
| 280 | We avoid resistance. | 我们避免抵抗。 |
| 281 | We use AI to outsource our thinking. | 我们用人工智能外包我们的思维。 |
| 282 | Write my LinkedIn post, fix my resume, | 写我的LinkedIn帖子, 修复我的简历, |
| 283 | summarize this book. | 总结此书. |
| 284 | That's like going to the gym | 这就像去健身房 |
| 285 | and asking someone else to lift weights on your behalf. | 要求别人代表你举重 |
| 286 | You know, when astronauts spend months in zero gravity, | 你知道,当宇航员在零重力中度过数月, |
| 287 | their muscles and bones atrophy dramatically. | 他们的肌肉和骨骼急剧萎缩 |
| 288 | Up to 20%. | 高达20%. |
| 289 | AI is like zero gravity for your thinking. | AI就像零引力 对于你的思维。 |
| 290 | No friction, no load, no growth. | 无摩擦无负载无增长. |
| 291 | The intelligent gym is not about information. | 智能体育馆不是关于信息。 |
| 292 | It's about transformation. | 是为转变. |
| 293 | For things where you need to be smart and capable, | 对于你需要聪明和有能力的事情, |
| 294 | you can think of AI as your spotter. | 你可以把AI当成你的监视者 |
| 295 | In any gym, a spotter doesn't lift the weight for you. | 在任何体育馆,一个看门人都不会帮你举重. |
| 296 | They stand next to you and help you lift. | 他们站在你旁边 帮你抬起来 |
| 297 | They also make sure that you don't get crushed | 他们也保证你不会被压扁 |
| 298 | when you're lifting the weight. | 当你举起重量。 |
| 299 | So do the same with AI. | 对AI也一样 |
| 300 | Here's a concrete example. | 这里有一个具体的例子。 |
| 301 | If you want to learn a concept, | 如果你想学一个概念, |
| 302 | study it first yourself and then go to your spotter. | 先自己研究 然后去找你的观察者 |
| 303 | Your AI. | 你的人工智能。 |
| 304 | Paste the concept text and then prompt AI. | 粘贴概念文本然后提示AI. |
| 305 | I need to master this concept. | 我需要掌握这个概念。 |
| 306 | Quiz me on it. | 问其上. |
| 307 | And now comes the most important part | 现在最重要的部分来了 |
| 308 | of your intelligent gym. | 你聪明的健身房 |
| 309 | Ask AI to apply progressive overload. | 要求AI应用渐进超载. |
| 310 | Four levels. | 四关. |
| 311 | Level one, quiz me like I am a high school student. | 一级,像我是高中生一样测试我 |
| 312 | Level two, ask me questions like I am a college student. | 二级,问我的问题 像我是一个大学生。 |
| 313 | Level three, now grill me like you're interviewing me | 3级,现在像采访我一样折磨我 |
| 314 | for an executive job. | 执行职务。 |
| 315 | And level four, now challenge me like an irate boss | 四级,现在像老闆一样挑战我 |
| 316 | who thinks I'm unprepared. | 谁觉得我没有准备好。 |
| 317 | So that truly strengthens and deepens your understanding | 因此,这真正加强 和加深你的理解 |
| 318 | on that concept. | 关于这个概念。 |
| 319 | So now we have covered three key steps | 因此,我们现在已经涵盖了三个关键步骤 |
| 320 | to learn how the top 1% becomes smarter by using AI. | 以学习如何通过使用人工智能使前1%变得更加聪明。 |
| 321 | But there is one internal adjustment | 但有一个内部调整 |
| 322 | that changes everything. | 这改变了一切。 |
| 323 | And that is our final step. | 这是我们的最后一步。 |
| 324 | Step number four, the intelligent fool. | 第四步 聪明的傻瓜 |
| 325 | You know, the biggest obstacle to intelligence | 你知道,最大的障碍 智能 |
| 326 | isn't ignorance. | 不是无知。 |
| 327 | It's ego. | 这是自负。 |
| 328 | That's why the smartest people are obsessed | 所以最聪明的人才会着迷 |
| 329 | with what they don't know. | 他们不知道的事 |
| 330 | And this is what I call the fool's advantage. | 这就是我所说的傻瓜的优势。 |
| 331 | Let me give you an example. | 让我举个例子。 |
| 332 | Microsoft went from $300 billion to 300 trillion | 微软从3000亿美元到300万亿美元 |
| 333 | in market cap with just one mental cultural shift. | 在市场封顶,只有一个精神文化转变。 |
| 334 | When Satya Nadella became the CEO of Microsoft in 2014, | 2014年萨特雅·纳德拉成为微软CEO时, |
| 335 | they had missed two huge disruptions, search and mobile. | 他们错过了两次巨大的干扰 搜索和移动。 |
| 336 | The cloud race was ongoing, | 云彩赛正在进行中 |
| 337 | but it was slipping away from them | 却从他们那里溜走, |
| 338 | with Amazon becoming the 800 pound gorilla. | 亚马逊成了800磅的大猩猩 |
| 339 | And the culture inside the company was toxic and political. | 而公司内部的文化是有毒的和政治的. |
| 340 | And everyone was terrified to admit | 每个人都害怕承认 |
| 341 | that there were gaps in their knowledge. | 他们的知识有缺陷。 |
| 342 | Satya made one cultural move. | 萨特雅做了一个文化运动。 |
| 343 | He told the entire company, | 他告诉整个公司, |
| 344 | we're switching from a culture of know-it-alls | 我们从一种全知的文化转变 |
| 345 | to learn-it-alls. | 来学习一切。 |
| 346 | A complete reboot of Microsoft culture. | 全面重启微软文化. |
| 347 | The smartest people in the room | 房间里最聪明的人 |
| 348 | were finally given permission to say, | 终于被允许说, |
| 349 | I don't know, or I was wrong. | 我不知道,或者我错了。 |
| 350 | And to embrace that beginner's mind. | 并拥抱初学者的思想。 |
| 351 | Now, Wall Street was skeptical at first, | 现在,华尔街 起初怀疑, |
| 352 | but the market cap eventually went from 300 billion | 但最终市场上限从3000亿 |
| 353 | to over 3 trillion. | 超过3万亿 |
| 354 | Keeps growing. | 不断增长. |
| 355 | 10X growth in a decade. | 10X增长十年. |
| 356 | And here's why this matters. | 所以这才重要 |
| 357 | Neuroscience tells us that our brain | 神经科学告诉我们我们的大脑 |
| 358 | can rewire all the time. | 可以随时重接 |
| 359 | It's called neuroplasticity. | 称为神经弹性. |
| 360 | This rewiring happens only at the edge of your ability. | 这种复线只在你的能力边缘发生. |
| 361 | It happens when you are making errors. | 当你犯错时就会发生 |
| 362 | It happens when you're frustrated, | 当你感到沮丧的时候 |
| 363 | when you're feeling that discomfort. | 当你感到不舒服的时候 |
| 364 | And if you aren't feeling stupid, you aren't learning. | 如果你不觉得自己很愚蠢,你就不会学习。 |
| 365 | And aren't you glad that AI has just handed you | 你不高兴AI刚刚把你交给我 |
| 366 | the ultimate training ground to be a student again? | 终极训练基地 再次成为学生? |
| 367 | You can bring your beginner's mind to AI all day long. | 你可以整天把初学者的思想带给AI. |
| 368 | Ask questions you would never ask your colleagues | 问问题,你绝对不会问同事 |
| 369 | out of fear of embarrassment. | 因为害怕尴尬 |
| 370 | AI doesn't roll its eyes. | AI不卷其眼. |
| 371 | Pick one thing that you don't understand in your field. | 选一件你不懂的事 |
| 372 | Something that everyone else thinks you know, | 别人认为你知道的事 |
| 373 | but you know you don't. | 但你知道你没有。 |
| 374 | And then ask AI the most basic questions about that topic | 然后问AI关于这个话题的最基本的问题 |
| 375 | that you can think of. | 你可以想到这一点。 |
| 376 | And then ask, can you explain it to me in a simpler way? | 然后问 你能用更简单的方式解释给我听吗? |
| 377 | Teach me like I am 10 years old. | 教我10岁 |
| 378 | I ask these questions all the time. | 我总是问这些问题。 |
| 379 | In fact, I asked three times in a row | 其实我连问三次 |
| 380 | to simplify again and again. | 一次又一次地简化。 |
| 381 | And sure, I guarantee you, | 当然,我向你保证, |
| 382 | you'll feel ridiculous at first. | 你一开始会觉得很可笑 |
| 383 | I do, all the time. | 我总是这样 |
| 384 | But that's the whole point. | 但这就是重点 |
| 385 | Have the courage to play the fool today | 今天有勇气装傻 |
| 386 | so you can be the genius tomorrow. | 你明天就能成为天才了 |
| 387 | The trick to mastery is going back to simplicity itself. | 掌握的诡计是回到简单本身. |
| 388 | If you examine some of the greatest masters | 如果你检查一些最伟大的大师 |
| 389 | across human history, you'll see one consistent pattern. | 纵观人类历史 你会看到一个一致的模式 |
| 390 | Every master is a student for life. | 每师为终身弟子. |
| 391 | And you can be a genuine student | 你可以成为真正的学生 |
| 392 | if you're hiding behind a mask of mastery. | 如果你躲在一个控制面具后面 |
| 393 | You know, the biggest benefit of intelligence | 你知道,情报的最大好处 |
| 394 | is not the end of ignorance, is the end of pretending. | 这不是无知的终结, 是假装的终结。 |
| 395 | You know, we're surrounded by endless images | 我们被无尽的影像包围着 |
| 396 | of flawless people in their flawless poses, | 无瑕疵者, |
| 397 | flawlessly photoshopped. | 完美无瑕的相机处理。 |
| 398 | But in the end, all art is about asymmetry. | 但归根结底,所有艺术都是关于不对称的. |
| 399 | We're beautiful because we're broken. | 我们之所以美丽是因为我们被打破了. |
| 400 | Because the real purpose of intelligence | 因为情报的真正目的 |
| 401 | of this thing called life is to travel far and wide | 这个叫生命的东西 是远行和远行 |
| 402 | only to return to yourself and fully accept who you are. | 只能回到你自己 并完全接受你是谁 |
| 403 | That is your truest intelligence. | 这是你最真实的智慧 |
| 404 | If you like this video, don't forget to subscribe. | 如果你喜欢这段录像 别忘了订阅 |
| 405 | And if you want to use AI to start a business, | 如果你想用人工智能创业 |
| 406 | here's another video where I walk you through | 这是另一个视频 我带你通过 |
| 407 | exactly what I would do. | 这正是我会做的。 |
| 408 | Thank you, and I love you. | 谢谢,我爱你 |
该视频共有字幕 408 条。解锁更多字幕为会员功能，请移动到 价格
P2 · 【AI思维】给我18分钟，让你聪明到危险 · 顶尖1%用AI的4个秘密 p02 无字幕 (P2)
字幕摘录
| 时间 | 英文 | 中文 |
|---|---|---|
| 0:00 | Most people are letting AI destroy their ability to think. | 大多数人都让AI破坏他们的思考能力. |
| 0:04 | Training AI to become their own replacement. | 训练AI成为他们自己的替补. |
| 0:06 | Tragic, because AI can make you dangerously intelligent. | 悲剧,因为AI可以让你危险地聪明. |
| 0:10 | I went from being homeless to an MIT grad | 从无家可归到麻省理工学院 |
| 0:13 | and running and advising AI companies worth billions. | 经营和向AI公司提供建议 价值数十亿美元。 |
| 0:16 | And here's what I've learned. | 且道我所学. |
| 0:18 | The top 1% use AI backwards. | 前1%使用AI反向. |
| 0:21 | They don't prompt to get answers. | 他们不会迅速得到答案。 |
| 0:23 | They use it to train their brain | 他们用它来训练大脑 |
| 0:25 | and outsmart almost any situation. | 和聪明的几乎任何情况。 |
展开字幕全文（405 条）
| 序号 | 英文 | 中文 |
|---|---|---|
| 1 | Most people are letting AI destroy their ability to think. | 大多数人都让AI破坏他们的思考能力. |
| 2 | Training AI to become their own replacement. | 训练AI成为他们自己的替补. |
| 3 | Tragic, because AI can make you dangerously intelligent. | 悲剧,因为AI可以让你危险地聪明. |
| 4 | I went from being homeless to an MIT grad | 从无家可归到麻省理工学院 |
| 5 | and running and advising AI companies worth billions. | 经营和向AI公司提供建议 价值数十亿美元。 |
| 6 | And here's what I've learned. | 且道我所学. |
| 7 | The top 1% use AI backwards. | 前1%使用AI反向. |
| 8 | They don't prompt to get answers. | 他们不会迅速得到答案。 |
| 9 | They use it to train their brain | 他们用它来训练大脑 |
| 10 | and outsmart almost any situation. | 和聪明的几乎任何情况。 |
| 11 | So in this video, I'll break down a counter-intuitive system | 因此,在这个视频中,我将打破一个反直观系统 |
| 12 | the top 1% use to get smarter, faster with AI. | 前1%用来提高智能 更快 |
| 13 | Here is the four-step framework. | 以下是四步框架。 |
| 14 | Step one, intelligent laziness. | 第一步 聪明的懒惰 |
| 15 | A study in Harvard Business Review found that CEOs | 《哈佛商业评论》的研究发现 |
| 16 | waste 72% of their time in meetings | 72%的会议时间被浪费 |
| 17 | that don't move the needle. | 不动的针。 |
| 18 | We've all experienced those meetings, haven't we? | 我们都经历过那些会议,不是吗? |
| 19 | The one-hour meeting that needed only 15 minutes | 只需15分钟的一小时会议 |
| 20 | to get to a decision. | 来决定 |
| 21 | But it's hard to stop. | 然其难止. |
| 22 | So why do some of the most accomplished folks | 所以,为什么做一些 最成就的人 |
| 23 | feel trapped this way? | 感觉被困住了吗? |
| 24 | Because we all suffer from this biological glitch | 因为我们都有这种生物缺陷 |
| 25 | called completion bias. | 名为完成偏差。 |
| 26 | Your brain is wired to seek an immediate dopamine hit | 你的大脑被电线连接到 寻求立即的多巴胺命中 |
| 27 | that you get from finishing a task. | 从你完成一个任务得到的。 |
| 28 | So we end up treating all tasks as equal | 所以,我们最终对待所有的任务是平等的 |
| 29 | because we're gonna get roughly the same amount of dopamine | 因为我们得到的多巴胺量大致相同 |
| 30 | when you spend time on redrafting an internal email | 当你花时间重新起草内部邮件时 |
| 31 | or a million dollar strategy document. | 或者一百万美元的战略文件。 |
| 32 | Everything is priority one. | 一切都是第一要务。 |
| 33 | So none of it is. | 故无一是. |
| 34 | So how do you avoid this priority blindness? | 那你如何避免这种优先失明? |
| 35 | A good way to think about tasks is to see two curves. | 思考任务的好方法就是看到两个曲线. |
| 36 | First curve has capped payoffs. | 第一曲线有上限的回报。 |
| 37 | This curve goes up and then flattens out | 这个曲线向上,然后平整 |
| 38 | once it reaches the zone of diminishing returns. | 一旦它到达了不断减少的回报区。 |
| 39 | So tasks like formatting slides or internal emails, | 因此,诸如格式化幻灯片或内部电子邮件等任务, |
| 40 | expense reports, FYI meetings, | 支出报告、财务会议、 |
| 41 | what happens if you spend additional effort | 如果你付出更多努力会怎么样? |
| 42 | to make the outcome of these tasks pitch perfect? | 使这些任务的结果完美无缺? |
| 43 | Nothing. | 没什么 |
| 44 | There's no upside here | 这里没有颠倒 |
| 45 | because the value flatlines after a point. | 因为值在一点后会平线。 |
| 46 | Nobody cares if you spend hours choosing better fonts | 没人在乎你有没有花几个小时选择更好的字体 |
| 47 | or breathtaking designs in internal slides | 或内部幻灯片中的呼吸设计 |
| 48 | that are seen for six minutes. | 6分钟内就能见到 |
| 49 | This curve shows you your zone of intelligent laziness. | 这个曲线显示你的智慧懒惰度 |
| 50 | There was a Nobel Prize winning economist | 有个诺贝尔奖得主经济学家 |
| 51 | and computer scientist and his name was Herbert Simon | 电脑科学家叫赫伯特・西蒙 |
| 52 | and he came up with a concept called satisfying, | 他提出一个叫满足的概念, |
| 53 | which pretty much means stop when it's good enough. | 这几乎意味着停止 当它足够好。 |
| 54 | Satisfy and suffice, satisfies. | 满足和满足,满足。 |
| 55 | Now our second curve is the exact opposite. | 现在我们的第二个曲线正好相反。 |
| 56 | It has uncapped payoff. | 它有未封顶的回报。 |
| 57 | This curve stays flat for a long time | 此曲线长期保持平坦 |
| 58 | but then goes to the moon in a hurry. | 却匆匆赶到月亮上。 |
| 59 | These are tasks like customer interactions, | 这些是客户互动等任务 |
| 60 | product design, pricing model, | 产品设计、定价模式、 |
| 61 | finding a co-founder or a life partner. | 寻找共同创始人或生活伴侣. |
| 62 | Being 1% better here does not yield 1% better result. | 这里的1%好,不会产生1%好的结果。 |
| 63 | It actually solves the rest of the 99% of your problems. | 它能解决你99%的问题 |
| 64 | Pour your soul into this. | 把你的灵魂注入这个。 |
| 65 | Johnny Ive would obsess for many months | 强尼 我会迷恋好几个月 |
| 66 | on even the internal component design of iPhone. | 甚至连iPhone的内部组件设计。 |
| 67 | But you know, Steve Jobs never said, | 但是你知道,史蒂夫・乔布斯从来没有说过, |
| 68 | hey, this is costing us a lot of money | 嘿,这花了我们很多钱 |
| 69 | 谁会打开iPhone? | |
| 70 | But Steve knew this was the second curve. | 但史蒂夫知道这是第二个曲线。 |
| 71 | So if the first curve is your zone of laziness, | 所以如果第一个曲线是你的懒惰区 |
| 72 | your second curve is your zone of obsession. | 你的第二个弯曲是你的痴迷区域。 |
| 73 | Let's talk about how AI can help. | 让我们来谈谈AI可以帮助。 |
| 74 | The top 1% use AI on zone one or the zone of laziness. | 前1%在第一区或懒惰区使用AI. |
| 75 | The more they outsource zone one to AI, | 他们越是把第一区外包给AI |
| 76 | the more they can focus on zone two, the zone of obsession. | 他们越是专注于第二区, 痴迷区。 |
| 77 | So how do I decide what to outsource to AI and when? | 那么,我如何决定什么是外包给AI和何时? |
| 78 | So for that, I use a very simple framework | 因此,我用一个很简单的框架 |
| 79 | 称为 Drag Framework, D- R- A- G. | |
| 80 | Four categories of work you immediately should delegate | 四类工作你应立即委托 |
| 81 | to AI so you can stay in your zone of obsession. | 这样你就可以留在你的痴迷区域 |
| 82 | First, D equals drafting. | 第一,D等于起草。 |
| 83 | This is the blank page problem we all face. | 这是我们面临的空白页问题。 |
| 84 | It's hardest to get from zero to one sometimes. | 有时从零到一很难做到. |
| 85 | AI can help here, tremendously actually. | AI可以在这里帮助, 巨大的实际上。 |
| 86 | Give it a prompt using the aim protocol | 使用目标协议迅速启动 |
| 87 | that I've shared before. | 我之前分享过 |
| 88 | Hey AI, act in this role, use this input | 嘿,AI,做这个角色,使用这个输入 |
| 89 | and this is your mission, A-I-M. | 这是你的使命,A -I -M。 |
| 90 | And that way you get started very quickly | 那样你就能很快开始 |
| 91 | on that email or code or presentation | 关于电子邮件或代码或演示文稿 |
| 92 | and the first draft from AI will be crappy and atrocious, | 而AI的第一稿将是 肮脏和残暴的, |
| 93 | but that's fine. | 但没关系。 |
| 94 | Now you have a starting point. | 现在你有一个起点。 |
| 95 | You're not staring at a blank page anymore. | 你不再盯着空白的一页了 |
| 96 | Now it'll trigger something in your brain | 现在它会触发你脑中的东西 |
| 97 | and you're off to the races. | 你将去比赛。 |
| 98 | R equals research. | R等于研究. |
| 99 | This helps you solve the information overload problem. | 这有助于解决信息超载问题. |
| 100 | Today, if something requires deep research, | 今天,如果有什么需要深入的研究, |
| 101 | it can be dramatically accelerated using AI. | 使用人工智能可以大大加速。 |
| 102 | Summarization, extraction, competitive intel. | 总结,提取,竞争性情报. |
| 103 | You know, don't spend time doing that kind of research. | 你知道,不要花时间 做这种研究。 |
| 104 | Let your friendly neighborhood AI do it for you. | 让你的友好邻居AI为你做。 |
| 105 | When you use the deep research feature | 当你使用深层研究特征 |
| 106 | on chat GBT or Gemini or Claude, | 在聊天GBT或双子座或克劳德, |
| 107 | it fires off hundreds of secondary search queries. | 它点燃了数百个二级搜索查询 |
| 108 | It goes out to the web like a spider | 像蜘蛛一样走进网络 |
| 109 | and finds hundreds of sites, consolidates the results, | 并发现数百个网站, |
| 110 | even checks his own work by asking what's missing | 连自己的作品都查过了 问出什么了 |
| 111 | and follows up on its own to finally deliver | 并独自跟进 最终交付 |
| 112 | a rich document to you. | 给您一个丰富的文件。 |
| 113 | It's like you just hired a consultant | 好像你刚雇了个顾问 |
| 114 | for a week long research project, | 为期一周的研究项目, |
| 115 | but instead you get there in 10 minutes. | 但你10分钟后就到了 |
| 116 | Third is A for analysis. | 第三是A分析. |
| 117 | Let AI take the first pass at analyzing, | 让大赦国际在分析时先通过 |
| 118 | summarizing, reasoning, | 总结,推理, |
| 119 | especially if it's all unstructured data | 特别是如果都是无结构的数据 |
| 120 | because AI is gonna find patterns | 因为AI会找到规律 |
| 121 | that we humans aren't gonna be able to. | 我们人类是无法做到的 |
| 122 | So use it for your advantage. | 所以用它来帮你 |
| 123 | And finally G is for all the grunt work. | 最后G是所有的咕噜声工作。 |
| 124 | Tasks like reformatting, translating, tabulating, | 重排、翻译、制表、 |
| 125 | cleaning data and on and on, the boring manual work. | 清理数据 继续 无聊的手工工作 |
| 126 | Just give it to AI. | 只要把它给AI。 |
| 127 | So what's the key principle behind drag? | 拖动背后的关键原则是什么? |
| 128 | Apply it only when you are in your zone one, | 只有在你进入第一区的时候才应用 |
| 129 | that first curve. | 第一个曲线。 |
| 130 | If it requires human interaction or judgment or intuition | 如果它需要人类的互动或判断或直觉 |
| 131 | or decision-making or tastes, that's curve two. | 或决策或口味, 这是曲线二。 |
| 132 | That you've gotta do it yourself. | 你得自己来 |
| 133 | But you know, I've found that 70 or 80% | 但你知道,我发现,70%或80% |
| 134 | of my repetitive tasks tend to be in zone one. | 我重复的任务往往在第一区 |
| 135 | And you might find that too. | 你也会找到的 |
| 136 | So be lazy when you can use drag. | 所以,懒惰 当你可以使用拖。 |
| 137 | Be obsessed for everything else. | 执着一切众生. |
| 138 | Step two, the intelligent hill. | 第二步,智慧山头. |
| 139 | For 300 years, Isaac Newton convinced us | 三百年来,艾萨克牛顿说服我们 |
| 140 | that universe was a clockwork machine, | 宇宙是时钟工作机, |
| 141 | predictable and certain. | 可预测和确定。 |
| 142 | But in 1927, another scientist named Heisenberg | 但是在1927年,另一位名叫海森堡的科学家 |
| 143 | shattered those classical beliefs. | 粉碎了这些古典信仰。 |
| 144 | He showed that our universe exists | 他证明我们的宇宙存在 |
| 145 | only as a cloud of possibilities at quantum level. | 仅作为量子级可能性的云. |
| 146 | It was a profound shift. | 这是一个深刻的转变。 |
| 147 | You and I have to make a similar shift | 咱俩得做个类似的转变 |
| 148 | when we use AI nowadays. | 当我们使用AI现在。 |
| 149 | The first trick is to stop treating AI like a calculator. | 第一个技巧是停止将AI当作计算器对待. |
| 150 | We like to live in a world with clear rules. | 我们喜欢生活在一个有明确规则的世界。 |
| 151 | You type two plus two into a calculator and you get four. | 把2加2打入计算器,然后得到4个. |
| 152 | Always, it's predictable. | 总是可以预测的 |
| 153 | But AI is not a calculator. | 但AI不是计算器. |
| 154 | It's a probability engine. | 这是一个概率引擎。 |
| 155 | If you ask the same question to AI again, | 如果你再次向大赦国际提出同样的问题, |
| 156 | it'll give you a completely different answer. | 它会给你一个完全不同的答案。 |
| 157 | It'll happily make things up for you | 这会给你带来幸福 |
| 158 | unless you ask it to verify. | 除非你要求它核实。 |
| 159 | AI is brilliant on some days, confused on others. | AI在一些日子很聪明,在另一些日子很困惑. |
| 160 | But on any given day, it refuses to admit | 但是在任何一天 它拒绝承认 |
| 161 | that it doesn't know the answer. | 说它不知道答案 |
| 162 | It loves to make things up. | 它喜欢编造东西。 |
| 163 | So you don't just ask AI | 所以,你不只是问AI |
| 164 | the way you ask a normal human being. | 你问一个正常人的方式。 |
| 165 | You have to architect your questions very carefully. | 你得仔细设计你的问题 |
| 166 | Now, most people use a tactic called zero-shot prompting. | 现在,大多数人都使用一种叫做零镜头提示的战术. |
| 167 | So for example, they would ask, | 例如,他们会问, |
| 168 | give me the best new business idea. | 给我一个最好的新商业点子 |
| 169 | And of course, AI will dish out a response | 当然,AI会给出一个回应 |
| 170 | and tell you why it's the greatest idea in the world. | 告诉你为什么这是世界上最伟大的想法。 |
| 171 | But you're literally rolling the dice and looking to win. | 但你真的在掷骰子 想赢 |
| 172 | To get elite results though, | 为了获得精英成果 |
| 173 | you must climb the intelligent hill. | 你必须爬上智慧山 |
| 174 | There are four camps on the way. | 途中有四个阵营. |
| 175 | Each camp will show you a different way to work with AI. | 每个营地会给你们展示一个不同的与AI合作的方式. |
| 176 | Our first camp is called one-shot prompting. | 我们的第一个阵营叫做一弹一弹 |
| 177 | When you prompt, give one clear example | 提示时,请举一个明显的例子 |
| 178 | so the model doesn't guess blindly. | 所以模型不会盲目猜测 |
| 179 | So the prompt would look like, | 所以,提示会看起来像, |
| 180 | write a LinkedIn post about remote work. | 撰写关于远程工作的LinkedIn文章。 |
| 181 | Use this specific post as a style guide. | 使用此特定帖子作为样式引导 。 |
| 182 | And so give it a post, give it an example | 所以给它一个帖子,举个例子 |
| 183 | and paste that post in the prompt as a reference. | 贴在标签上作为参考 |
| 184 | And that simple act is already an upgrade | 那简单的行为已经升级了 |
| 185 | than rolling the dice blindly. | 而不是盲目地掷骰子。 |
| 186 | Second camp, few-shot prompting. | 第二营,几发子弹 |
| 187 | Now here you give AI three or more examples | 现在请举出三个或更多例子 |
| 188 | so it can find patterns of style and substance | 这样它就能找到风格和实质的规律 |
| 189 | and tone that you desire. | 和你想要的语气。 |
| 190 | Attach documents, links, data, or your prior work. | 附加文档、链接、数据或您先前的工作。 |
| 191 | This is called grounding the model. | 这叫做打地基模型。 |
| 192 | So basically it stops fantasizing and hallucinating | 所以基本上它停止了幻想和幻觉 |
| 193 | and gets grounded to reality. | 被现实所束缚 |
| 194 | Here's an example of a prompt. | 以下是一个提示的例子. |
| 195 | Here are the five of my previous presentations. | 以下是我先前介绍的五点。 |
| 196 | And now write a new presentation | 现在写一个新的演示文稿 |
| 197 | based on my tone of voice on topic XYZ. | 根据我对主题XYZ的声音 |
| 198 | And here's a pro tip. | 这是专业小费 |
| 199 | Ask the AI to explain the pattern back to you first. | 请AI先向您解释模式. |
| 200 | That way AI is forced to articulate what it's doing | 这样一来,AI就被迫阐述它在做什么 |
| 201 | and more importantly, you're forced to learn | 更重要的是,你被迫学习 |
| 202 | how your brain works. | 你的大脑如何工作。 |
| 203 | How did it come up with those patterns? | 怎么会有这种模式? |
| 204 | Now you're being smart about being smart. | 现在,你对聪明很聪明。 |
| 205 | Now let's move to the third camp. | 今移三营. |
| 206 | This one is called chain of thought reasoning. | 这个叫做思想推理链. |
| 207 | Again, fancy name, but the idea is simple. | 再说一遍 名气不错 但想法很简单 |
| 208 | Ask the model to think long and hard before it responds. | 问模型在回应前要长时间而努力思考. |
| 209 | Your job is to slow AI down and force explicit clarity | 你的工作就是减慢AI的速度 强迫它清晰清晰 |
| 210 | by asking it to show its work. | 要求它展示它的工作。 |
| 211 | That's all there is. | 仅此而已. |
| 212 | This is also a good way to reduce hallucinations, of course. | 这当然也是减少幻觉的好方法。 |
| 213 | So let's say you're working on some report | 假设你在做一些报告 |
| 214 | and so you attach it and write a prompt | 并写出一个提示 |
| 215 | that could look like this. | 看起来像这样 |
| 216 | Do not refine my research report yet. | 还没有完善我的调查报告。 |
| 217 | List the top three most impactful areas of improvement | 列出三个影响最大的改进领域 |
| 218 | after you analyze it. | 在你分析出来之后 |
| 219 | Tell me why you think so and suggest how we address each. | 告诉我你为什么这么认为 并提议我们如何讨论每个问题 |
| 220 | Think step by step, show me your thinking for each step. | 一步一步的思考, 给我看看你对每一步的想法。 |
| 221 | That last line is the most important one. | 最后一行是最重要的一条。 |
| 222 | And our fourth and final camp is agents. | 我们的第四个也是最后一个营地是特工 |
| 223 | According to Salesforce, AI agents help drive $67 billion | 根据销售部队,AI代理 帮助驱动670亿美元 |
| 224 | in global sales during cyber week alone. | 仅网络周全球销售量。 |
| 225 | So agents are already here. | 特工已经来了 |
| 226 | The best way to think about agents | 考虑特工的最佳方法 |
| 227 | is to think about who you would hire for a task. | 想想你会雇谁来做任务 |
| 228 | So let's say if you wanted to hire a researcher, | 这么说吧 如果你想雇个研究员 |
| 229 | an analyst, and a copywriter, | 分析师和抄写员 |
| 230 | you can do that with a single agentic prompt | 单个代理提示就可以做到这一点 |
| 231 | that looks like this. | 这看起来像这个。 |
| 232 | Do deep research on trends on topic XYZ, | 对主题XYZ的趋势进行深入研究, |
| 233 | analyze and cross-reference all the trends | 分析和对照所有趋势 |
| 234 | to find the three most important ones | 找到三个最重要的 |
| 235 | and draft a one-page memo summarizing the findings. | 并起草一份一页的备忘录,总结调查结果。 |
| 236 | Now, what is actionable? | 现在,什么是可行动? |
| 237 | Try this framework tonight. | 今晚试试这个框架 |
| 238 | Open your favorite AI app | 打开你最喜欢的 AI 应用程序 |
| 239 | and take any prompt that you were about to use. | 并使用任何时间。 |
| 240 | Just try to get to the next camp. | 试着去下一个营地 |
| 241 | That's how you start climbing up the intelligent hill. | 这就是你开始爬上智慧山的方法. |
| 242 | Remember, when you were dealing with a drunk genius, | 记住,当你面对一个醉酒的天才, |
| 243 | make sure you were the one driving the car. | 确保开车的是你 |
| 244 | So now, at this point, everything we have done | 所以,现在,我们所做的一切 |
| 245 | has made you fast and efficient. | 使你迅速而高效。 |
| 246 | You're delegating better, you're prompting smarter, | 你授权的更好, 你激励聪明, |
| 247 | you're moving up the hill, | 你移动上山, |
| 248 | and there is less friction than before. | 而摩擦比以前少。 |
| 249 | And that's exactly where most people would stop. | 而这正是大多数人会停止的地方。 |
| 250 | But here's the plot twist. | 但这里的情节曲折。 |
| 251 | The top 1% go one step further. | 前1%更进一步。 |
| 252 | They slow things down deliberately. | 他们故意减慢速度 |
| 253 | Why is that important? | 为什么这很重要? |
| 254 | The trick that top 1% know is this. | 最高的1%知道这个 |
| 255 | They know when to shift the gear | 他们知道什么时候该换档 |
| 256 | because long-term intelligence | 因为长期智能 |
| 257 | isn't built through convenience, | 不是通过方便建造的, |
| 258 | it's built through resistance. | 它是通过抵抗建造的。 |
| 259 | And that's why we need to go to step three. | 这就是为什么我们需要去第三步。 |
| 260 | The intelligent gym. | 智能体育馆. |
| 261 | Most people use AI as a wheelchair for the mind. | 大多数人使用AI作为心灵的轮椅. |
| 262 | And if you sit in a wheelchair when you can still walk, | 如果你坐在轮椅上还能走路 |
| 263 | eventually your legs stop working. | 最终你的腿停止工作。 |
| 264 | Atrophy. | 萎缩. |
| 265 | And today is happening faster than at any point | 今天发生的速度比任何时刻都快 |
| 266 | in human history. | 在人类历史上。 |
| 267 | But the top 1% use a very different principle. | 但前1%使用非常不同的原理. |
| 268 | For information tasks, use AI to remove friction. | 对于信息任务,使用AI去除摩擦. |
| 269 | For transformation tasks, use AI to add friction. | 对于转化任务,使用AI添加摩擦. |
| 270 | When you go to a physical gym, | 当你去健身房时 |
| 271 | we all know how muscles are built, right? | 我们都知道肌肉是如何形成的,对吧? |
| 272 | Through resistance. | 通过抵抗。 |
| 273 | You lift increasingly heavier weights | 你举重越来越重 |
| 274 | to introduce wear and tear to your muscle fibers | 介绍你的肌肉纤维磨损 |
| 275 | so they break and they grow back stronger. | 所以它们会断裂,会变强 |
| 276 | That is called progressive overload. | 这被称为逐步超载。 |
| 277 | But when it comes to our minds, | 但是,当涉及到我们的想法, |
| 278 | we do the exact opposite somehow, you know? | 我们做完全相反的,你知道吗? |
| 279 | We avoid resistance. | 我们避免抵抗。 |
| 280 | We use AI to outsource our thinking. | 我们用人工智能外包我们的思维。 |
| 281 | Write my LinkedIn post, fix my resume, | 写我的LinkedIn帖子, 修复我的简历, |
| 282 | summarize this book. | 总结此书. |
| 283 | That's like going to the gym and asking someone else | 这就像去健身房 问别人 |
| 284 | to lift weights on your behalf. | 代表你举重。 |
| 285 | You know, when astronauts spend months in zero gravity, | 你知道,当宇航员在零重力中度过数月, |
| 286 | their muscles and bones atrophy dramatically. | 他们的肌肉和骨骼急剧萎缩 |
| 287 | Up to 20%. | 高达20%. |
| 288 | AI is like zero gravity for your thinking. | AI就像零引力 对于你的思维。 |
| 289 | No friction, no load, no growth. | 无摩擦无负载无增长. |
| 290 | The intelligent gym is not about information. | 智能体育馆不是关于信息。 |
| 291 | It's about transformation. | 是为转变. |
| 292 | For things where you need to be smart and capable, | 对于你需要聪明和有能力的事情, |
| 293 | you can think of AI as your spotter. | 你可以把AI当成你的监视者 |
| 294 | In any gym, a spotter doesn't lift the weight for you. | 在任何体育馆,一个看门人都不会帮你举重. |
| 295 | They stand next to you and help you lift. | 他们站在你旁边 帮你抬起来 |
| 296 | They also make sure that you don't get crushed | 他们也保证你不会被压扁 |
| 297 | when you're lifting the weight. | 当你举起重量。 |
| 298 | So do the same with AI. | 对AI也一样 |
| 299 | Here's a concrete example. | 这里有一个具体的例子。 |
| 300 | If you want to learn a concept, | 如果你想学一个概念, |
| 301 | study it first yourself and then go to your spotter. | 先自己研究 然后去找你的观察者 |
| 302 | You're AI. | 你是爱尔 |
| 303 | Paste the concept text and then prompt AI. | 粘贴概念文本然后提示AI. |
| 304 | I need to master this concept. | 我需要掌握这个概念。 |
| 305 | Quiz me on it. | 问其上. |
| 306 | And now comes the most important part | 现在最重要的部分来了 |
| 307 | of your intelligent gym. | 你聪明的健身房 |
| 308 | Ask AI to apply progressive overload. | 要求AI应用渐进超载. |
| 309 | Four levels. | 四关. |
| 310 | Level one, quiz me like I am a high school student. | 一级,像我是高中生一样测试我 |
| 311 | Level two, ask me questions like I am a college student. | 二级,问我的问题 像我是一个大学生。 |
| 312 | Level three, now grill me like you're interviewing me | 3级,现在像采访我一样折磨我 |
| 313 | for an executive job. | 执行职务。 |
| 314 | And level four, now challenge me like an irate boss | 四级,现在像老闆一样挑战我 |
| 315 | who thinks I'm unprepared. | 谁觉得我没有准备好。 |
| 316 | So that truly strengthens and deepens your understanding | 因此,这真正加强 和加深你的理解 |
| 317 | on that concept. | 关于这个概念。 |
| 318 | So now we have covered three key steps | 因此,我们现在已经涵盖了三个关键步骤 |
| 319 | to learn how the top 1% becomes smarter by using AI. | 以学习如何通过使用人工智能使前1%变得更加聪明。 |
| 320 | But there is one internal adjustment | 但有一个内部调整 |
| 321 | that changes everything. | 这改变了一切。 |
| 322 | And that is our final step. | 这是我们的最后一步。 |
| 323 | Step number four, the intelligent fool. | 第四步 聪明的傻瓜 |
| 324 | You know, the biggest obstacle to intelligence | 你知道,最大的障碍 智能 |
| 325 | isn't ignorance, it's ego. | 不是无知,而是自负 |
| 326 | That's why the smartest people are obsessed | 所以最聪明的人才会着迷 |
| 327 | with what they don't know. | 他们不知道的事 |
| 328 | And this is what I call the fool's advantage. | 这就是我所说的傻瓜的优势。 |
| 329 | Let me give you an example. | 让我举个例子。 |
| 330 | Microsoft went from $300 billion to 300 trillion | 微软从3000亿美元到300万亿美元 |
| 331 | in market cap with just one mental cultural shift. | 在市场封顶,只有一个精神文化转变。 |
| 332 | When Satya Nadella became the CEO of Microsoft in 2014, | 2014年萨特雅·纳德拉成为微软CEO时, |
| 333 | they had missed two huge disruptions. | 他们错过了两次巨大的干扰。 |
| 334 | Search and mobile. | 搜索和移动。 |
| 335 | The cloud race was ongoing, | 云彩赛正在进行中 |
| 336 | but it was slipping away from them | 却从他们那里溜走, |
| 337 | with Amazon becoming the 800 pound gorilla. | 亚马逊成了800磅的大猩猩 |
| 338 | And the culture inside the company was toxic and political. | 而公司内部的文化是有毒的和政治的. |
| 339 | And everyone was terrified to admit | 每个人都害怕承认 |
| 340 | that there were gaps in their knowledge. | 他们的知识有缺陷。 |
| 341 | Satya made one cultural move. | 萨特雅做了一个文化运动。 |
| 342 | He told the entire company, | 他告诉整个公司, |
| 343 | we're switching from a culture of know-it-alls | 我们从一种全知的文化转变 |
| 344 | to learn-it-alls. | 来学习一切。 |
| 345 | A complete reboot of Microsoft culture. | 全面重启微软文化. |
| 346 | The smartest people in the room | 房间里最聪明的人 |
| 347 | were finally given permission to say, | 终于被允许说, |
| 348 | I don't know, or I was wrong. | 我不知道,或者我错了。 |
| 349 | And to embrace that beginner's mind. | 并拥抱初学者的思想。 |
| 350 | Now, Wall Street was skeptical at first, | 现在,华尔街 起初怀疑, |
| 351 | but the market cap eventually went from 300 billion | 但最终市场上限从3000亿 |
| 352 | to over 3 trillion, and it keeps growing. | 超过3万亿, 它不断增长。 |
| 353 | 10X growth in a decade. | 10X增长十年. |
| 354 | And here's why this matters. | 所以这才重要 |
| 355 | Neuroscience tells us that our brain | 神经科学告诉我们我们的大脑 |
| 356 | can rewire all the time. | 可以随时重接 |
| 357 | It's called neuroplasticity. | 称为神经弹性. |
| 358 | This rewiring happens only at the edge of your ability. | 这种复线只在你的能力边缘发生. |
| 359 | It happens when you are making errors. | 当你犯错时就会发生 |
| 360 | It happens when you're frustrated, | 当你感到沮丧的时候 |
| 361 | when you're feeling that discomfort. | 当你感到不舒服的时候 |
| 362 | And if you aren't feeling stupid, you aren't learning. | 如果你不觉得自己很愚蠢,你就不会学习。 |
| 363 | And aren't you glad that AI has just handed you | 你不高兴AI刚刚把你交给我 |
| 364 | the ultimate training ground to be a student again? | 终极训练基地 再次成为学生? |
| 365 | You can bring your beginner's mind to AI all day long. | 你可以整天把初学者的思想带给AI. |
| 366 | Ask questions you would never ask your colleagues | 问问题,你绝对不会问同事 |
| 367 | out of fear of embarrassment. | 因为害怕尴尬 |
| 368 | AI doesn't roll its eyes. | AI不卷其眼. |
| 369 | Pick one thing that you don't understand in your field. | 选一件你不懂的事 |
| 370 | Something that everyone else thinks you know, | 别人认为你知道的事 |
| 371 | but you know you don't. | 但你知道你没有。 |
| 372 | And then ask AI the most basic questions about that topic | 然后问AI关于这个话题的最基本的问题 |
| 373 | that you can think of. | 你可以想到这一点。 |
| 374 | And then ask, can you explain it to me in a simpler way? | 然后问 你能用更简单的方式解释给我听吗? |
| 375 | Teach me like I am 10 years old. | 教我10岁 |
| 376 | I ask these questions all the time. | 我总是问这些问题。 |
| 377 | In fact, I asked three times in a row | 其实我连问三次 |
| 378 | to simplify again and again. | 一次又一次地简化。 |
| 379 | I guarantee you, you'll feel ridiculous at first. | 我向你保证 你一开始会觉得很可笑 |
| 380 | I do all the time. | 我总是这样 |
| 381 | But that's the whole point. | 但这就是重点 |
| 382 | Have the courage to play the fool today | 今天有勇气装傻 |
| 383 | so you can be the genius tomorrow. | 你明天就能成为天才了 |
| 384 | The trick to mastery is going back to simplicity itself. | 掌握的诡计是回到简单本身. |
| 385 | If you examine some of the greatest masters | 如果你检查一些最伟大的大师 |
| 386 | across human history, you'll see one consistent pattern. | 纵观人类历史 你会看到一个一致的模式 |
| 387 | Every master is a student for life. | 每师为终身弟子. |
| 388 | And you can be a genuine student | 你可以成为真正的学生 |
| 389 | if you're hiding behind a mask of mastery. | 如果你躲在一个控制面具后面 |
| 390 | You know, the biggest benefit of intelligence | 你知道,情报的最大好处 |
| 391 | is not the end of ignorance, is the end of pretending. | 这不是无知的终结, 是假装的终结。 |
| 392 | You know, we're surrounded by endless images | 我们被无尽的影像包围着 |
| 393 | of flawless people in their flawless poses, | 无瑕疵者, |
| 394 | flawlessly photoshopped. | 完美无瑕的相机处理。 |
| 395 | But in the end, all art is about asymmetry. | 但归根结底,所有艺术都是关于不对称的. |
| 396 | We're beautiful because we're broken. | 我们之所以美丽是因为我们被打破了. |
| 397 | Because the real purpose of intelligence, | 因为情报的真正目的 |
| 398 | of this thing called life, is to travel far and wide, | 这叫做生命的事物,是漫游的, |
| 399 | only to return to yourself and fully accept who you are. | 只能回到你自己 并完全接受你是谁 |
| 400 | That is your truest intelligence. | 这是你最真实的智慧 |
| 401 | If you like this video, don't forget to subscribe. | 如果你喜欢这段录像 别忘了订阅 |
| 402 | And if you want to use AI to start a business, | 如果你想用人工智能创业 |
| 403 | here's another video where I walk you through | 这是另一个视频 我带你通过 |
| 404 | exactly what I would do. | 这正是我会做的。 |
| 405 | Thank you, and I love you. | 谢谢,我爱你 |
该视频共有字幕 405 条。解锁更多字幕为会员功能，请移动到 价格