AI时代最值钱的技能是什么？与Matt Pocock聊聊编程
字幕摘录
| 时间 | 英文 | 中文 |
|---|---|---|
| 0:00 | How do you stand out in a world where everyone can build things so quick? | 在一个每个人都能这么快地建造东西的世界里,你如何突出? |
| 0:03 | I think the people who are really successful are the people who are the best communicators, right? | 我认为真正成功的人是 最好的沟通者,对不对? |
| 0:07 | So in my opinion, the best way to get the best work out of agents is to tell them what you care about. | 所以在我看来 最好的办法就是 告诉他们你关心什么 |
| 0:13 | So the skill there is being able to just get it all out of your head. | 所以,那里的技巧是能够 得到这一切从你的脑海。 |
| 0:17 | The skill is being aware of unknown unknowns, right? | 技能是知道未知的,对不对? |
| 0:20 | Could be security, could be how you deploy it, could be just things you haven't thought of. | 可能是安全,可能是你如何部署它, 可能是你没有想到的事情。 |
| 0:24 | And if you're not careful about those, then the agent will just make stuff up. | 如果你不注意这些,那么代理会只是编造的东西。 |
| 0:29 | What would you see as the big opportunities to be looking at to help launch your career in the right spot? | 你会看到什么是大 机会看 帮助启动你的职业生涯 在正确的位置? |
| 0:34 | Taking agents, figuring out processes for knowledge work, and then bringing those for organizations. | 吸收代理人,找出知识工作的程序,然后将这些流程带给各组织。 |
| 0:39 | Because every organization does things slightly differently. | 因为每个组织的行为都略有不同。 |
展开字幕全文（932 条）
| 序号 | 英文 | 中文 |
|---|---|---|
| 1 | How do you stand out in a world where everyone can build things so quick? | 在一个每个人都能这么快地建造东西的世界里,你如何突出? |
| 2 | I think the people who are really successful are the people who are the best communicators, right? | 我认为真正成功的人是 最好的沟通者,对不对? |
| 3 | So in my opinion, the best way to get the best work out of agents is to tell them what you care about. | 所以在我看来 最好的办法就是 告诉他们你关心什么 |
| 4 | So the skill there is being able to just get it all out of your head. | 所以,那里的技巧是能够 得到这一切从你的脑海。 |
| 5 | The skill is being aware of unknown unknowns, right? | 技能是知道未知的,对不对? |
| 6 | Could be security, could be how you deploy it, could be just things you haven't thought of. | 可能是安全,可能是你如何部署它, 可能是你没有想到的事情。 |
| 7 | And if you're not careful about those, then the agent will just make stuff up. | 如果你不注意这些,那么代理会只是编造的东西。 |
| 8 | What would you see as the big opportunities to be looking at to help launch your career in the right spot? | 你会看到什么是大 机会看 帮助启动你的职业生涯 在正确的位置? |
| 9 | Taking agents, figuring out processes for knowledge work, and then bringing those for organizations. | 吸收代理人,找出知识工作的程序,然后将这些流程带给各组织。 |
| 10 | Because every organization does things slightly differently. | 因为每个组织的行为都略有不同。 |
| 11 | I think that's a massive opportunity. | 我认为这是一个巨大的机会。 |
| 12 | I would love to hear a war story. | 我很想听一个战争故事。 |
| 13 | I tried doing spectrum development, I suppose, and I just saw the code getting worse and worse and worse. | 我尝试了频谱开发,我想, 我刚刚看到代码越来越糟糕。 |
| 14 | And the agent would produce worse and worse results. | 而毒剂会产生越来越糟糕的结果. |
| 15 | And I tried this for months and I just thought about it again and again and again. | 我尝试了几个月,我一次又一次地思考。 |
| 16 | It wasn't working. | 没用的 |
| 17 | So I realized. | 所以我意识到了。 |
| 18 | All right, Matt, I'm super excited to be sitting here with you today. | 好吧,马特,我超兴奋 今天坐在这里和你在一起。 |
| 19 | You know, I am a frequent user of your skills, of course. | 你知道,我是一个经常使用 你的技能,当然。 |
| 20 | And it's really interesting to me that you taught yourself to code. | 这对我来说很有趣,你教自己 密码。 |
| 21 | And when we first hopped on the call right before we started recording, you said that you think learning how to code is a lot easier than people make it out to be. | 当我们在开始录制之前第一次接听电话时, 你说,你认为学习密码比人们想的容易得多。 |
| 22 | So you taught yourself how to code. | 所以你教自己如何编码 |
| 23 | You were doing it by hand for about 10 years. | 你亲手做了大约10年。 |
| 24 | And then we entered this whole era of being able to use coding agents and things like that. | 然后我们进入了能够使用编码剂和类似事物的整个时代。 |
| 25 | So I'm really curious to hear your opinion on all this today. | 所以我很想知道你今天对这一切的看法。 |
| 26 | But I'd love to just kick it off real quick to you and understand what was that transition like for you when you had learned all these fundamentals of coding and you were doing it by hand. | 但我想尽快地把它踢开, 并理解当你学会了这些编码的基本原理 并且你正在用手做的时候,你是如何转变的。 |
| 27 | And then you can sort of like, you know, the term vibe coding pops up and people are just building things in one day. | 然后,你可以有点像,你知道, 气温编码一词出现, 人们只是在一天内建造东西。 |
| 28 | What did that transition look like for you? | 那个转变对你来说是什么样子的? |
| 29 | Well, thank you. | 谢谢 |
| 30 | Thank you for using my skills as well. | 谢谢你也利用了我的技巧 |
| 31 | I appreciate it. | 谢谢 |
| 32 | That's very kind. | 赞曰. |
| 33 | I really started. | 我真的开始。 |
| 34 | I was a skeptic for a long time. | 久之为疑. |
| 35 | I had about 10 years of coding, like as a sort of amateur and then professionally for about eight years before vibe coding, before stuff got really good. | 我做了大约10年的编码, 像一种业余的, 然后专业的大约8年 之前的氛围编码, 之前的东西真的很好。 |
| 36 | I consider like when Opus 4.5 got released, when OpenClaw got built around, you know, December last year, 2025. | 我认为,当Opus 4.5 被释放的时候, 当OpenClaw 建造在周围, 你知道,去年12月,2025年。 |
| 37 | I consider that to be the moment, the sort of threshold where I went, OK, I'm never going to write a line of code myself anymore. | 我认为这是那一刻, 那种门槛 我去了,OK,我再也不会写 一条代码线我自己。 |
| 38 | You know, so that's the moment when agents got really, really good. | 你知道,这就是一个时刻 特工得到了真正,真的好。 |
| 39 | And it was the kind of year leading up to that where I was going, OK, yeah, I can see the benefits. | 在那之前的一年里,我要去的地方, OK,是的,我可以看到好处。 |
| 40 | I can see that everyone's very excited about this, but it hasn't quite clicked for me. | 我可以看到每个人都对此感到非常兴奋,但是它并没有为我点击。 |
| 41 | And as soon as I, I don't know, for me, I've always been very excited by the parts of my job that were like trying to make my colleagues better, right? | 当我,我不知道,对我来说, 我一直非常兴奋 我的工作部分,就像试图让我的同事更好,对不对? |
| 42 | Trying to build things that helped my colleagues ship stuff faster, trying to communicate to other devs and like and teach them how to do their job better, even if they were probably maybe senior to me. | 试图建造一些东西来帮助我的同事更快地运送东西,试图与其他的虔诚者沟通并喜欢,并教他们如何更好地完成任务,即使他们可能比我年长。 |
| 43 | So I'm always interested in like the environment in which my colleagues operated and the environment in which I operated. | 所以,我总是对喜欢 我的同事在什么环境中工作 和我在什么环境中工作感兴趣。 |
| 44 | And so when agents came around and they actually were really good, I started thinking, right, I can use these skills that I've developed over my career. | 所以当特工们出现的时候, 他们其实是非常好的, 我开始思考,对,我可以使用这些技能, 这些技能是我职业生涯中发展出来的。 |
| 45 | And my interest in this stuff to really help these agents get a lot, lot better, a lot faster. | 我对这东西很感兴趣 来帮助这些特工们获得更多,更好的,更快的。 |
| 46 | And that hadn't really been working previously before Opus four point five. | 这在奥普斯四点五之前并没有真正发挥作用。 |
| 47 | And then December last year, it started working and I realized, wow, OK, I can really put my skills to work here and actually pump out some pretty good stuff. | 然后去年12月,它开始工作, 我意识到,哇,好吧,我真的可以把我的技能 工作在这里,实际上抽出一些相当好的东西。 |
| 48 | And it's been working so well so far. | 并且一直运作至今. |
| 49 | That's super cool. | 这太酷了 |
| 50 | You know, one of the things that I've kind of taken away from the stuff that you're putting out there online is that you believe the software engineering fundamentals and principles are even more important now because AI is being slapped into all of these | 我从你在网上展示的东西中 拿走了一件事, 就是你相信软件工程的基础和原则 现在更重要,因为AI被打入了所有这些开发过程。 |
| 51 | And but also on the same point, you don't necessarily believe that someone has to be super, super technical or have a huge coding background in order to take advantage of the stuff. | 但也在同一点上,你不一定认为某人必须具有超超,超技术性或巨大的编码背景才能利用这些东西. |
| 52 | So what does that look like for someone who's not technical? | 那对不技术的人来说是什么样子? |
| 53 | Because I know that your audience is primarily pretty technical. | 因为我知道你的观众 主要是很专业的 |
| 54 | What does it look like for someone who's not technical? | 对一个不技术的人来说是什么样子? |
| 55 | How do they actually use software engineering foundations and principles to build in the right way? | 他们实际上如何利用软件工程基础和原则以正确的方式建设? |
| 56 | Right. | 对 |
| 57 | So in my opinion, the best way to get the best work out of agents is to tell them what you care about. | 所以在我看来 最好的办法就是 告诉他们你关心什么 |
| 58 | Right. | 对 |
| 59 | Because agents will tend to like people underestimate how people underestimate the communication barrier there is between you, the human and the agent. | 因为特工会倾向于喜欢人们低估人们如何低估你,人和特工之间的沟通障碍. |
| 60 | Right. | 对 |
| 61 | And it doesn't matter how smart that model is, there's always going to be that gap between you. | 不管那个模型有多聪明, 你之间总会有这种差距。 |
| 62 | And the thing about that gap is that the agent doesn't know precisely what you want. | 而那个缺口的事实是, 经纪人并不知道你想要什么。 |
| 63 | It doesn't know what you care about. | 它不知道你关心什么。 |
| 64 | And so you have to tell it again and again every time. | 所以,你必须一次又一次地说出来。 |
| 65 | And so I think you're going to get better results out of agents if the environment in which they operate is useful for them. | 所以我认为如果特工操作的环境 对他们有用的话 你会得到更好的结果 |
| 66 | Right. | 对 |
| 67 | If they're in like a garbage, vibe coded, crappy code base that has never had any care put into it, no one cares about the code base. | 如果他们像一个垃圾, 气温编码, 破烂的代码基础 从来没有在它的任何小心, 没有人关心 代码基础。 |
| 68 | They only care about the outputs. | 他们只关心产出。 |
| 69 | Then you're going to get garbage in, garbage out. | 然后,你要去得到垃圾 垃圾出来。 |
| 70 | Right. | 对 |
| 71 | The agent is just going to produce rubbish because it's in a rubbish environment. | 经纪人只是要生产垃圾 因为它在垃圾环境中。 |
| 72 | And so you have to start caring about that environment. | 所以,你必须开始关心环境。 |
| 73 | And the way you start caring about it is you start thinking about how, let's say I was running a huge coding organization and I had a hundred humans, human coders. | 你开始关心它的方式是,你开始思考如何, 比如说,我经营着一个巨大的编码组织, 我拥有一百个人类,人类的编码员。 |
| 74 | I know that sounds like a strange idea to actually do my work for me. | 我知道这听起来像一个奇怪的主意 实际上为我做我的工作。 |
| 75 | Right. | 对 |
| 76 | Or push work out. | 或推出业. |
| 77 | How would I raise the floor so that the worst human coder that I had access to was going to perform better? | 我该如何提高地板,使我接触到的最糟糕的人类编解码器能更好地工作? |
| 78 | What strategies would I use? | 我会用什么策略? |
| 79 | And it turns out people have been thinking about this stuff for 20 years. | 结果人们已经思考了20年了。 |
| 80 | Right. | 对 |
| 81 | And there are books written about all of these concepts. | 还有一些关于这些概念的书。 |
| 82 | And it turns out if you read those books, they feel like they were written last week. | 如果你读了那些书,他们就会觉得是上周写的。 |
| 83 | They feel so relevant for right now because they're talking about how you make good tests, how you add checks to make sure that your code base gets better and better over time to raise the floor, to stop classes of bugs, to prevent mistakes. | 他们现在觉得很贴切 因为他们在谈论你如何做好测试 如何增加支票 以确保你的密码基础 随着时间的推移会越来越好 |
| 84 | And that's what everyone is trying to do now. | 现在大家都想这么做 |
| 85 | And so I think the first step in that is to care about that and care about learning it and get interested in it because agents are pretty good at teaching you stuff as well. | 所以,我认为第一步是 关心这一点, 关心学习,并对此产生兴趣 因为特工们也非常善于教你的东西。 |
| 86 | So if you're interested in it, if you're asking the right questions, then the agents will be able to understand that you care about that and start pushing you in that direction. | 所以如果你感兴趣,如果你问对了问题,那么特工们就能理解你关心这一点,开始把你推向那个方向. |
| 87 | That's super interesting because I do, of course, I would say 80% of the stuff that I've learned in the past year, I've had my agents teach it to me. | 这非常有趣,因为我当然会说, 百分之八十的东西,我在过去的一年里学到, 我让我的代理人教我。 |
| 88 | And I know that you have a skill, the slash teach. | 我知道你有本事 刀锋教书 |
| 89 | And, you know, one thing that I'd be curious about is how do you think about trusting what they're teaching you? | 你知道,我想知道的一件事 是,你如何思考相信 他们教你的东西? |
| 90 | You shouldn't, right? | 你不应该,对不对? |
| 91 | You shouldn't. | 你不应该。 |
| 92 | You should always be skeptical over anything that anyone tells you. | 你应该总是怀疑别人告诉你的任何事 |
| 93 | I think I'm like a born skeptic. | 我想我就像一个天生的怀疑者。 |
| 94 | I've always like, it took me quite a while to get interested in AI because I was like, this thing is rubbish. | 我一直喜欢,它花了我很长时间 开始对AI感兴趣 因为我喜欢,这个东西是垃圾。 |
| 95 | This thing is like the new Web3 crypto bubble, you know, BS, whatever. | 这东西就像新的Web3密码泡, 你知道,BS,什么的。 |
| 96 | Am I allowed to swear on this? | 我可以发誓吗? |
| 97 | I'm not sure. | 我不太确定 说不定 |
| 98 | Yeah, okay. | 是啊,没关系。 |
| 99 | Okay, good. | 好吧,不错。 |
| 100 | I tend to swear a lot on podcasts. | 我经常在播客身上发誓 |
| 101 | I don't know, it loosens me up. | 我不知道,这让我放松。 |
| 102 | So I was very skeptical for a long time and I still remain skeptical unless I can read the primary source, right? | 所以我很怀疑,很长一段时间,我仍然怀疑,除非我能读到主要来源,对不对? |
| 103 | So when I designed the teach skill, the teach skill is essentially all about going out, | 所以当我设计教学技巧时, 教学技巧基本上都是关于外出, |
| 104 | fetching a high trust, high quality primary sources, and then breaking those down for people with quotes, | 获取一个高信任,高质量的原始来源, 然后打破那些为人们 引用, |
| 105 | with like attribution, with saying exactly where it links to it. | 并说出它与它的确切关联 |
| 106 | And over time, the hallucinations of agents have gone way down. | 随着时间流逝 特工们的幻觉逐渐下降 |
| 107 | You know, they will tend to not make stuff up so much anymore. | 你知道,他们会 倾向于不编 这么多的东西了。 |
| 108 | They still do. | 他们仍然如此。 |
| 109 | Fable still does it. | 福尔仍然这样做。 |
| 110 | They still do it, but it's much less frequent. | 他们仍然这样做,但频率要低得多. |
| 111 | And I found that what I mostly use those teach skill sessions for is I will just get it to | 我发现我主要使用这些 教授技能课程 是为了,我会得到它 |
| 112 | introduce me to a topic and then I'll just go and read the primary source, | 向我介绍一个话题 然后我就去读一下主要来源 |
| 113 | because that primary source has the actual information I need. | 因为那个主线有我需要的信息 |
| 114 | It's just maybe I didn't quite have the language to interpret it before the teach session. | 只是也许在教书之前 我还没有语言来解释它 |
| 115 | All right, real quick, guys, a huge thanks to Hostinger for sponsoring this part of the video. | 非常感谢Hostinger赞助了这段录像 |
| 116 | So I know that a lot of you guys are obviously learning to build with AI. | 所以,我知道,你们很多人显然 学习与AI一起建设。 |
| 117 | And at some point, what you build has to live somewhere other than your local laptop, | 在某个时候,你建造的东西 得生活在你本地的笔记本电脑以外的地方, |
| 118 | because once other people use it, you need a server that stays on | 因为一旦其他人使用它, 你需要一个服务器保持 |
| 119 | and lets you change things without breaking what's actually working. | 并让你改变事物 不打破实际工作。 |
| 120 | So Hostinger VPS is your own server with dedicated CPU and RAM and full root access. | 因此Hostinger VPS是你自己的服务器,拥有专用的CPU和RAM以及完整的根访问. |
| 121 | And it makes the setup the easy part. | 这让设置变得容易 |
| 122 | So you can actually spend your time building instead of on the plumbing, | 所以你可以把时间用在管道上 |
| 123 | because I know a lot of us aren't from like a DevOps background. | 因为我知道我们很多人 不像DevOps的背景 |
| 124 | So I'd go with the KVM2 plan. | 所以我会去KVM2计划。 |
| 125 | For the price, you get two vCPUs, 8 gigs of RAM and 100 gigs of storage, | 为了价格,你得到两个vCPU, 8个工作 RAM和100工作 存储, |
| 126 | which is plenty for a first project. | 足够做第一个项目了 |
| 127 | And you can always upgrade later if you need to. | 如果你需要,你随时可以升级 |
| 128 | Then you can pick the cloud code template, which comes with cloud code already installed. | 然后你可以选择云码模板,它与已经安装的云码同时出现. |
| 129 | So you can start building as soon as you log in. | 你一登录就可以开始建设了 |
| 130 | And to keep what you build safe, you can snapshot your server before you change anything. | 并且为了保持安全,你可以在改变任何事情之前将你的服务器拍下来. |
| 131 | And you can roll back if it breaks with weekly backups included. | 如果每周有备份的话,你可以滚回去。 |
| 132 | And if you get stuck, Hostinger agent in the web console helps with server tasks, errors, | 如果你被卡住,网络控制台上的托管人代理 帮助服务器的任务,错误, |
| 133 | and deployments just by using your natural plain English. | 使用自然而然的英语进行部署。 |
| 134 | 这样您就可以在Hostinger.com slash nathercvps上抓取KVM2. | |
| 135 | The link is in the description and you can use code natherc at checkout, | 链接在描述中,你可以使用代码Natherc 在离职, |
| 136 | which will take an extra 10% off any yearly plan on top of the current pricing. | 在目前的定价基础上,任何年度计划都将额外减少10%。 |
| 137 | So thanks again, Hostinger for sponsoring this part of the video. | 再次感谢主持人赞助了这段视频。 |
| 138 | And now let's get back to it. | 而今复说. |
| 139 | Gotcha. | 抓到他们了 |
| 140 | Okay. | 摆 |
| 141 | So what would you say is, you know, you're seeing a lot of different people | 所以,你会说什么是, 你知道,你看到 很多不同的人 |
| 142 | coming into your ecosystem and they're learning how to teach themselves things with AI, | 进入你的生态系统 他们正在学习如何 教自己的东西与AI, |
| 143 | how to build products or automations. | 如何制造产品或自动化。 |
| 144 | And they probably have a lot of different motivations in there. | 他们可能有很多不同的动机。 |
| 145 | What are you seeing as the different opportunities that are popping up for people? | 你看到什么是不同的 机会为人们? |
| 146 | That want to make sure maybe they are a little bit skeptical, | 想确定他们是不是有点怀疑 |
| 147 | like you were about this whole AI thing, but now we're seeing this, right? | 就像你刚才在说艾尔的事 但现在我们看到了 对吧? |
| 148 | Like it's not going away. | 譬如不退转. |
| 149 | What would you see as the big opportunities to be looking at | 你会看到什么是 大机会看 |
| 150 | to help, you know, launch your career in the right spot? | 帮助,你知道, 启动你的职业生涯 在正确的位置? |
| 151 | Well, I mean, in any period of massive change, | 我的意思是,在任何大规模改变的时期, |
| 152 | there are always going to be opportunities available, right? | 总是会有机会的,对吧? |
| 153 | As everything shifts, then people are going to be adopting things at a slower pace. | 随着一切的转变,人们将会以更慢的速度接受事物。 |
| 154 | For me, I'm seeing a ton of like teaching opportunities, right? | 对我来说,我看到了一吨 像教学机会,对不对? |
| 155 | People want to learn this stuff because it is so useful. | 人们想学习这些东西,因为它是如此有用。 |
| 156 | It is so powerful. | 它是如此强大。 |
| 157 | And every organization I talk to is having resistance with this stuff, | 和我交谈的每个组织 都有抵抗这些东西, |
| 158 | trying to get it adopted. | 试图让它被收养。 |
| 159 | And let me make it clear. | 让我说清楚 |
| 160 | I do think that there are parts about the AI world that are unethical or have dubious ethics. | 我确实认为AI世界有些地方是不道德的,或者有可疑的道德. |
| 161 | And I think there are attitudes out there that are absolutely correct. | 我认为外面有些态度是完全正确的。 |
| 162 | You know, it is a bit weird, the fact that, you know, | 你知道,这有点奇怪,事实上,你知道, |
| 163 | Anthropic just nicked a bunch of novels and then just sort of paid them back later. | Anthropic刚刚偷了一堆小说, 之后又付了钱。 |
| 164 | You know, it is weird that it is like, | 你知道,这很奇怪 它就像, |
| 165 | it is going to strain the energy ecosystem to have all of this AI out there. | 它会给能源生态系统带来压力 让这些人工智能都在那里 |
| 166 | But you're right. | 但你是对的 |
| 167 | It is here. | 它在这里。 |
| 168 | It is not going away because there is an economic incentive to have it. | 它不会消失,因为有获得它的经济动力。 |
| 169 | So in terms of opportunities, I mean, they're just everywhere you look, really. | 所以就机会而言,我的意思是, 它们只是所有你看起来,真的。 |
| 170 | Like the ability to produce different types of applications that embed AI, | 如能产生嵌入AI的不同类型的应用程序, |
| 171 | to go into companies to say, | 进入公司说, |
| 172 | I know how to use AI and I can help your systems improve and systems develop. | 我知道如何使用人工智能 我可以帮助你的系统改善和系统发展。 |
| 173 | It seems like everyone has this buzzword on their lips. | 似乎每个人都嘴唇上有这个叫声 |
| 174 | And I think the opportunities are there if you look for them. | 如果你找的话,我想机会是存在的。 |
| 175 | Totally. | 当然 |
| 176 | So let us take a practical example. | 因此,让我们举一个实际的例子。 |
| 177 | Let us say someone right now works a full-time job and they are not technical. | 让我们说,现在有人做全职工作,他们不是技术性的。 |
| 178 | Let us say maybe they are a salesperson. | 让我们说,也许他们是推销员。 |
| 179 | So not technical, they work a full-time job, | 所以,不是技术性的, 他们是全职工作, |
| 180 | but they want to figure out with AI, | 但是他们想和AI一起解决 |
| 181 | should I learn how to build this stuff and then start up a consulting business? | 我该学学怎么制造这些东西 然后开始一个咨询业务吗? |
| 182 | Or should I try to use AI at work myself and just get promotions | 或者,我应该尝试使用AI 在工作,只是得到升迁 |
| 183 | and try to work into a leadership position at my job? | 并试图工作到领导职位 在我的工作? |
| 184 | How would you yourself think about the different paths | 你怎么想的 不同的路径 |
| 185 | and trying to figure out which one is right for you? | 想知道哪个适合你吗? |
| 186 | So what I'd say is, I think of this as kind of, | 所以,我想说的是, 我认为这是一种, |
| 187 | I take a sort of narrower view, | 我从狭义的角度看 |
| 188 | which is what are the skills that are going to be useful to me going forward? | 哪些技能对我前进有帮助? |
| 189 | And if you have the skills that I'm about to describe, | 如果你有我要描述的技能 |
| 190 | then you will be able to find your way to something useful, | 然后你们就能找到有用的东西, |
| 191 | whether it's building your own thing | 无论它是否在建造你自己的东西 |
| 192 | or whether it's getting promoted within your own organization. | 还是在自己的组织内晋升 |
| 193 | And the skills I think are useful are, | 我认为有用的技能是, |
| 194 | people think that agents, there's no learning curve to them. | 人们认为特工们, 他们没有学习曲线。 |
| 195 | They think you just talk to it and you just get stuff back. | 他们认为你只是说话 它,你只是拿回来的东西。 |
| 196 | And what I've found is that is absolutely not the case. | 我发现的绝对不是这种情况。 |
| 197 | And as models improve, they don't necessarily lessen that learning curve. | 随着模型的改进,它们并不一定会降低学习曲线。 |
| 198 | The learning curve is still there. | 学习曲线还在那里。 |
| 199 | It's just maybe a little bit hidden from most people. | 只是对大多数人有点隐瞒 |
| 200 | So the concept of what can you trust from an agent? | 你从经纪人那里可以相信什么? |
| 201 | How do you organize agents? | 你怎么组织特工? |
| 202 | How do you get agents to behave in ways that follow what you want? | 你怎么让特工们按照自己的意愿行事? |
| 203 | How do you control the information that agents receive? | 你怎么控制特工们收到的信息? |
| 204 | How do you design the information that you want the agents to receive? | 你如何设计你想要特工收到的信息? |
| 205 | What forms and data structures should those be in? | 这些表格和数据结构应采用何种形式和结构? |
| 206 | How do you then take those agents and turn them into reusable processes? | 那你怎么把这些特工 变成可重复使用的程序? |
| 207 | Or how do you take your processes | 或者你如何处理你的流程 |
| 208 | and turn them into something that agents can follow? | 把他们变成特工可以跟踪的东西? |
| 209 | For instance, I've been using agents for code, obviously, | 举例来说,我一直在使用代理代码,很明显, |
| 210 | and agents are really good at code because code is very easy to verify. | 并且特工非常擅长代码 因为代码很容易被验证 |
| 211 | Just like you see agents solving all these math problems | 就像你看到特工解决这些数学问题一样 |
| 212 | because math is very far. | 因为数学很遥远 |
| 213 | You can create these proofs that you just generate over time | 您可以创建这些证明, 这些证明是您随时间而生成的 |
| 214 | and you can solve these amazing math problems. | 你可以解决这些惊人的数学问题。 |
| 215 | But real-world work is much harder than coding for agents | 但现实世界的工作比代理商的编码难得多 |
| 216 | because it's much harder to verify. | 因为很难核实 |
| 217 | And so when I've had my biggest struggles trying to adopt agents for my day-to-day work | 所以当我最大的斗争 试图收养代理 我的日常工作 |
| 218 | and it's taken a long time and trying out innumerable processes | 花了很长时间 尝试了无数个过程 |
| 219 | until I finally found something I'm pretty happy with | 直到我终于找到了我很高兴的东西 |
| 220 | and I'm slowly improving it over time. | 我慢慢地改进了它 |
| 221 | So I think those are the big opportunities | 所以我认为这些是大机会 |
| 222 | is actually taking agents, figuring out processes for knowledge work, | 正在寻找代理 找出知识工作的程序 |
| 223 | and then bringing those for organizations | 然后带给各组织 |
| 224 | because every organization does things slightly differently | 因为每个组织的工作都略有不同 |
| 225 | and there's all of these opportunities out there to get that working. | 和所有这些机会 在那里得到工作。 |
| 226 | And that's a skill, that's really hard. | 那是个技巧,真的很难 |
| 227 | Even me having fiddled about with agents for a year, year and a half now, | 连我和特工们共事一年半 |
| 228 | I'm still finding it very difficult and I think that's a massive opportunity. | 我仍然觉得这很困难 我认为这是一个巨大的机会 |
| 229 | So you said that the fundamentals are important | 所以你说基础很重要 |
| 230 | and you've also said that explaining the context behind why you want something done, | 你也说过 解释你为什么想要做一些事情的背景 |
| 231 | so sort of like intent-driven prompting and development. | 就像意图驱动的激励和发展 |
| 232 | Now, how do those things come together? | 现在,这些事情是如何结合在一起的? |
| 233 | Where does that break where a non-technical person | 在非技术人员的情况下,这种突破在哪里? |
| 234 | that might not have knowledge of how to set up the right code base | 可能不知道如何建立正确的代码基础 |
| 235 | or how to create the right scoped API keys or whatever the issue might be, | 或如何创建合适的 API 密钥或任何问题, |
| 236 | how does that person drive with intent | 这个人是怎样驾驶的 |
| 237 | while still making sure that things don't fall through? | 还要确保事情不会发生吗? |
| 238 | And I'm not sure if you have any war stories of, | 我不确定你有没有战争故事 |
| 239 | and I'm sure this is where your whole vibe coding pivot away comes into place, | 我敢肯定,这就是你 整个气温编码中心 到达的地方, |
| 240 | but I'd love to hear if you've seen specific examples of that falling short | 但我很想听听你有没有看到 具体的例子 |
| 241 | and how someone can think about. | 以及人们如何思考。 |
| 242 | I guess the point I'm trying to make is what I've seen is you can have enough awareness | 我想说的是 我所看到的就是 你有足够的意识 |
| 243 | to make your agent think about the problem | 让你的经纪人考虑这个问题 |
| 244 | and you might not know exactly how to solve the problem, | 你也许不知道如何解决问题, |
| 245 | but if you can identify that this is something that needs to be looked into, | 但是如果你能确认这是需要研究的 |
| 246 | a lot of times that can go a long way. | 很多时候,这可以很长一段时间。 |
| 247 | So I guess that's the point I'm trying to make is how do you think about | 所以我想说的是 你在想什么 |
| 248 | a regular person being able to identify those opportunities? | 能够识别这些机会的普通人? |
| 249 | I mean, you're talking about a regular person versus a coder, right? | 我的意思是,你说的是一个普通人 对一个编码员,对不对? |
| 250 | But everything you're saying there is so familiar to me as a developer, right? | 但你所说的一切 我作为开发商都很熟悉 对吧? |
| 251 | There's no divide there. | 那儿没有分裂 |
| 252 | There's a great quote in the Pragmatic Programmer which says, | "实用程序员"中有一句好话说 |
| 253 | no one knows exactly what they want, right? | 没人知道他们到底想要什么 对吧? |
| 254 | No one knows what they want. | 没有人知道他们想要什么。 |
| 255 | Everyone has these unknown unknowns in their head. | 每个人都有这些未知数 在他的头上。 |
| 256 | Whenever they're specifying anything, whenever they're telling, | 无论何时,无论何时,无论何时,只要他们告诉, |
| 257 | let's say we're talking to a dev, whenever I as a normal person talk to a developer, | 假设我们是在跟一个德夫说话 我作为普通人跟开发商说话 |
| 258 | which is kind of like a normal person talking to an agent, | 就像普通人跟经纪人说话 |
| 259 | there is that communication barrier, right? | 有那个沟通障碍,对不对? |
| 260 | And the dev is going to listen to everything and try to work out what I care about. | 和dev会去听 一切,并尝试 解决我关心的。 |
| 261 | And so one of my skills in particular, | 因此,我的一个技能,特别是, |
| 262 | which is one of the most commonly used ones is called Grill Me. | 它是最常用的一种叫做Grill Me。 |
| 263 | And what that does is the agent essentially tells the agent to interview me relentlessly | 关键是那个特工要他无情地采访我 |
| 264 | about the thing that we're trying to build. | 关于我们试图建造的东西。 |
| 265 | And the reason that is so effective is because | 之所以如此有效是因为 |
| 266 | it then gets more context from you about what you care about | 然后从你那里得到更多的背景 关于你关心的 |
| 267 | and the thing you're actually trying to build | 而实际上你试图建造的东西 |
| 268 | and flushes out some of those unknown unknowns where, | 并冲出一些未知的未知数 其中, |
| 269 | sure, some things you don't know that you don't know | 当然,有些事情你不知道 你不知道 |
| 270 | and some things that you do know that you don't know. | 和一些你确实知道的东西 你不知道。 |
| 271 | And that's very confusing, but essentially it's stuff | 这很困惑,但本质上是东西 |
| 272 | you're not aware of that you need to care about, right? | 你不知道你需要关心,对吧? |
| 273 | Could be security, could be how you deploy it, | 这可能是安全, 可能是如何部署它, |
| 274 | could be just things you haven't thought of. | 可能是你没想过的事 |
| 275 | And if you're not careful about those, then the agent will just make stuff up | 如果你不小心,那探员会编造出来的 |
| 276 | or it will say, obviously that's not important because they didn't mention it. | 或者会说 这显然不重要 因为他们没提到 |
| 277 | And you'll end up with weird stuff happening | 你最后会遇到奇怪的事情 |
| 278 | or you'll just end up with something that's totally different from how you actually imagined it. | 否则你最后会遇到 和你想象的完全不同的事情 |
| 279 | So the skill there is being able to just get it all out of your head. | 所以,那里的技巧是能够 得到这一切从你的脑海。 |
| 280 | Yes, the skill is, I think, knowing that you're... | 是的,技能是,我认为,知道你是... |
| 281 | What is the skill? | 技术是什么? |
| 282 | The skill is being aware of unknown unknowns, right? | 技能是知道未知的,对不对? |
| 283 | Of knowing that you could be missing things in your design | 知道你在设计中会漏掉一些东西 |
| 284 | and not trusting that you've come up with a complete design. | 也不相信你会想出一个完整的设计 |
| 285 | And whenever you create code based on some idea, it's really hard, right? | 每当你根据某种想法创建代码, 这真的很难,对不对? |
| 286 | That's a hard thing to do because by turning it into code, | 这很难做 因为把它变成代码 |
| 287 | you realize the weaknesses in the design. | 你知道设计中的弱点 |
| 288 | And sometimes those things just can't be made clear until you've reached code | 有时候这些事情是无法澄清的 直到你到达密码 |
| 289 | because it's a kind of translation from the fuzzy human brain | 因为它是一种翻译 从模糊的人类大脑 |
| 290 | into executable statements, right? | 变成可执行声明,对吧? |
| 291 | Into things that have to link to each other. | 入诸物中须相通. |
| 292 | And there's always difficulty in translation there. | 在那里翻译总是有困难的. |
| 293 | Absolutely. | 当然 |
| 294 | I mean, I remember when I first started building automations, | 我的意思是,我记得当我开始 建立自动化, |
| 295 | I remember doing the best I could, trying to predict all the edge cases | 我记得我尽力了 试图预测所有的边缘情况 |
| 296 | and building guardrails to protect against those. | 并建造护栏来防御这些 |
| 297 | But then you put it in production and things happen, | 但你把它放在生产 事情发生, |
| 298 | especially when the inputs to the system are human | 特别是当对系统的投入是人类的 |
| 299 | because humans are just as unpredictable, | 因为人类同样无法预料 |
| 300 | if not more unpredictable than agents are. | 如果不是比特工更难预测的话 |
| 301 | So I remember at least being aware of being able to communicate that to my client | 所以,我记得至少知道 能够告诉我的客户 |
| 302 | and saying, hey, I've built all these protections, | 并说,嘿,我已经建立了所有这些保护, |
| 303 | but there are still some things that might happen in the real world. | 但现实世界中仍有一些可能发生的事情. |
| 304 | And that's why I'm going to be here to monitor this and see what happens | 所以我会来监视这个 看看会发生什么 |
| 305 | and make sure that all of these failures are still happening in a safe way. | 并确保所有这些失败仍然以安全的方式发生。 |
| 306 | Now, a lot of my audience is looking to build solutions | 现在,我的许多观众 正在寻找解决方案 |
| 307 | for their clients or their stakeholders at work. | 为其客户或在工作中的利益攸关方提供服务。 |
| 308 | And one thing that I feel like I've been seeing a lot is a bit of a trust gap | 有一点我觉得我看到了很多 有点信任差距 |
| 309 | because it's so amazing how much you can build in one day | 因为有一天你能造出多少 |
| 310 | and how expert you can appear if you lock yourself in a room for a weekend | 如果你把自己锁在房间里过周末 你会有多厉害? |
| 311 | and just have a portfolio when you come out on Monday. | 周一你出来的时候有份工作 |
| 312 | How would you think about positioning yourself | 你会如何思考定位自己 |
| 313 | about your experience, your expertise, your portfolio projects? | 关于你的经验,你的专长,你的组合项目? |
| 314 | How would you position yourself to a cold prospect | 你将如何面对冷淡的前景 |
| 315 | to give them confidence that you are the guy or girl to get the job done, | 让他们相信 你是完成这项工作的男是女 |
| 316 | to be able to look at all the other, | 能观察其他一切 |
| 317 | maybe 20 vendors they're looking at or 20 people they're considering promoting. | 也许有20个卖家在看 或者有20个人在考虑晋升 |
| 318 | How do you stand out in a world where everyone can build things so quick? | 在一个每个人都能这么快地建造东西的世界里,你如何突出? |
| 319 | Real quick, guys, I've got this completely free kit | 伙计们,快点,我有一个完全免费的包 |
| 320 | I want to give to you all on helping you build your own AIOS. | 我想给大家 关于帮助你建立自己的AIOS。 |
| 321 | Whether you need to build it or whether you need to scale it up, | 无论你需要建造,还是需要扩大, |
| 322 | this is the full kit. | 这是完整的包。 |
| 323 | It's based on my actual framework. | 它基于我的实际框架。 |
| 324 | I call it the Herc loop, context, connections, capabilities, and cadence. | 我称之为赫克环路, 上下文,连接,能力, 和cadence。 |
| 325 | This thing comes with six skills. | 此一物随六种技艺而来. |
| 326 | It helps you set up a brain. | 它帮助你建立大脑。 |
| 327 | It helps you set up your priorities in your business and it interviews you. | 它能帮助你设定你的优先 业务和采访你。 |
| 328 | It is exactly how I built my own actual AI operating system | 这正是我如何建造自己的人工智能操作系统 |
| 329 | and it's how I'm able to move so fast. | 这就是我能如此快速移动的方式。 |
| 330 | So if you guys want this, the link for this is down in the description. | 所以如果你们想要这个,这个的链接 在描述中下降。 |
| 331 | Anyways, let's get back to the video. | 无论如何,让我们回到视频。 |
| 332 | Honestly, I've never been in that position, right? | 老实说,我从没这样过,对吧? |
| 333 | I've never been a freelance consultant. | 我从未做过自由职业顾问 |
| 334 | I suppose I have been a freelance consultant. | 我想我是自由职业的顾问 |
| 335 | I just happened to have a huge audience while I was doing it, right? | 我只是碰巧有一个巨大的观众 当我做它,对不对? |
| 336 | So I had the kind of reputation that came before. | 所以我有那种声誉 之前。 |
| 337 | I think what people really want in those situations | 我觉得在这种情况下人们真正想要的 |
| 338 | is they want someone they can work with | 他们想要一个可以合作的人 |
| 339 | and they want someone who can understand what they care about, right? | 他们想要一个能理解他们所关心的人,对吧? |
| 340 | Who can try to cross the communication gap. | 谁能尝试跨越沟通差距. |
| 341 | And so because agents allow you to ship prototypes to just ship... | 因为特工允许你运送原型船... |
| 342 | I would say at this point, vibe coding is actually a good thing, right? | 我想说,此时此刻, 气温编码实际上是好事,对不对? |
| 343 | You're just trying to get to an endpoint quick enough. | 你只是想尽快到达终点 |
| 344 | If you can ship something that demonstrates to them | 如果你能送些东西给他们看 |
| 345 | that you understand their business, | 你了解他们的生意, |
| 346 | that you care about what they care about | 你关心他们关心的 |
| 347 | and that you can demonstrate real acumen | 并且你可以表现出真正的聪明 |
| 348 | in building something real for them very quickly, | 很快地为他们打造了真实的东西 |
| 349 | I think they're going to like that. | 我想他们会喜欢的 |
| 350 | So I feel like I can't offer concrete advice for that | 所以,我觉得我不能提供 具体的建议 |
| 351 | because I've never been in that competitive situation | 因为我从来没有 在竞争的情况下 |
| 352 | bidding for roles and stuff. | 竞争角色和东西。 |
| 353 | But use AI to your advantage. | 但利用人工智能对你有利 |
| 354 | And if it freaks people out that you're able to churn out | 如果它吓人 人,你可以发抖 |
| 355 | work in a day, then just charge them for five days | 一天工作,然后就收费五天 |
| 356 | and just sit on your bum for four. | 坐在你的屁股四个。 |
| 357 | People don't realize how quickly you can churn out | 人们不知道你有多快就能出来 |
| 358 | high quality stuff with agents yet, | 高级的东西与代理还没有, |
| 359 | especially people who are not AI build. | 特别是那些不是人工智能的人 |
| 360 | And so I think you might just have to adjust | 所以我觉得你可能只需要调整一下 |
| 361 | what you're doing for them. | 你为他们做了什么。 |
| 362 | What I think is interesting about what you said there | 我对你说的有意思 |
| 363 | is it sounds like it's the humanness | 听起来像是人性 |
| 364 | that helps build that trust | 有助于建立信任 |
| 365 | rather than look at all these cool things I've built | 而不是看我建造的这些酷东西 |
| 366 | or look at all these frameworks I understand. | 或者看看这些我理解的框架 |
| 367 | It's the humanness of, | 这是人性, |
| 368 | does this person care about what I care about | 这个人在乎我关心的是什么 |
| 369 | and are they thinking about protecting my business, | 他们想保护我的生意, |
| 370 | not just getting a paycheck out of me? | 不只是从我身上拿到薪水? |
| 371 | What about with your students? | 你的学生呢? |
| 372 | I'm sure there's been some things that you've seen | 我敢肯定,有些事情 你见过 |
| 373 | where it's like this person is, | 就像这个人一样 |
| 374 | there's potential here, very talented. | 这里有潜力 非常有才华 |
| 375 | And you might've also seen some traits where it's like, | 你可能也看到了一些特质 比如说, |
| 376 | this person might need more work a couple extra weeks. | 这个人可能需要多工作 几个星期。 |
| 377 | What are you seeing as far as like, | 你看见什么,像, |
| 378 | what sticks out about your students to you? | 你的学生对你有什么看法? |
| 379 | I think the people who are really successful | 我认为那些真正成功的人 |
| 380 | are the people who are the best communicators, right? | 人谁是最好的传播者,对不对? |
| 381 | That's always been true for developers | 对开发者来说,这一直是事实 |
| 382 | and for non-developers too. | 对非发展中国家来说也是 |
| 383 | Everyone I've seen who's been like the top, top tier | 我见过的每个人 谁像顶级,顶级 |
| 384 | of developer has always been incredibly good at communicating. | 开发者总是非常善于沟通 |
| 385 | That means that when they make a code change, | 这意味着当他们修改代码时 |
| 386 | they are writing a really detailed report | 他们正在写一份非常详细的报告 |
| 387 | about what changed. | 关于什么改变了。 |
| 388 | It's easy to read, easy to understand, | 易读易识. |
| 389 | very clear from their thinking. | 从他们的思维中非常清楚。 |
| 390 | In other words, the clarity of their thinking | 换句话说,他们思想的清晰 |
| 391 | ends up in the product they create. | 最终在他们创造的产品。 |
| 392 | And that is what I notice about folks who are not doing that | 这就是我注意到的人 谁不这样做 |
| 393 | is they get worse results | 他们得到的结果更糟 |
| 394 | because they can't communicate things to the agent | 因为他们不能跟特工沟通 |
| 395 | and then they can't communicate what is wrong. | 然后他们无法沟通什么是错的。 |
| 396 | They're not thinking in clear terms and clear boundaries. | 他们的思维方式并不明确,界限也不明确. |
| 397 | And so actually I think the quality of your thinking | 所以实际上我觉得你的思维质量 |
| 398 | and the quality of your language is key, right? | 你语言的质量是关键 对吧? |
| 399 | Not only for when you're talking to the client, | 不只是当你和客户说话时 |
| 400 | but also when you're talking to the agent | 当你和经纪人说话的时候 |
| 401 | about what the client wants. | 关于客户想要什么。 |
| 402 | I have a few skills for this, which is, | 我有一些技能 这一点,即, |
| 403 | and there's sort of a coding idea | 有一个编码的想法 |
| 404 | that comes into this as well, | 也涉及到这个, |
| 405 | which is there's a concept called DDD, | 有个叫DDD的概念 |
| 406 | which is Domain Driven Design. | 这是域驱动设计。 |
| 407 | Sounds very complicated. | 听起来很复杂 |
| 408 | But it has this idea at the core of it, | 但它有这个想法的核心, |
| 409 | which is that you have when you're building a project, | 当你在建一个项目时 |
| 410 | a, what they call a ubiquitous language, | 他们称之为无所不在的语言, |
| 411 | where the language in the code, | 在代码中的语言, |
| 412 | the terms you use in the code should be the same | 您在代码中使用的术语应该相同 |
| 413 | as the ones the developers use to talk about the product, | 由于开发者用来谈论产品, |
| 414 | which is the same as the client | 这与客户端相同 |
| 415 | asking you to build the thing in the product, | 让你在产品中制造这个东西, |
| 416 | which is the same as the users using the product. | 它与使用该产品的用户相同。 |
| 417 | So if you have a concept of an order, right? | 所以,如果你有一个概念 一个命令,对不对? |
| 418 | What is an order that should look at the same in the code | 密码里应该看一样的顺序是什么? |
| 419 | as it does in the database, as it does in the UI, | 就像在数据库里一样 就像在UI里一样 |
| 420 | as it does when you and the client are talking about it. | 就像你和客户在谈论的时候一样 |
| 421 | And actually iterating on that and creating that | 并创造出 |
| 422 | and the preciseness of that language, | 和该语言的准确性, |
| 423 | again, helps you align with what they actually want | 再来一次,帮助你适应他们真正想要的东西 |
| 424 | and what you're actually building. | 以及你真正建造的东西 |
| 425 | I can't recommend this technique enough. | 我无法推荐这种技巧 |
| 426 | And actually I have a modified version of GrillMe, | 事实上我有一个改版的GrillMe, |
| 427 | which does this. | 是谁干的? |
| 428 | So it helps you come up with these terms | 因此,它帮助你 提出这些条件 |
| 429 | called Grill with Docs. | 和医生一起叫格里尔 |
| 430 | Grill和Docs就是这样做的 | |
| 431 | It helps you build this domain language as you go. | 它帮助你建立这个域语言,当你去。 |
| 432 | I mean, I don't know. | 我是说,我不知道。 |
| 433 | I'm sort of shilling DDD there, like, | 我有点先令DDD在那里,喜欢, |
| 434 | but it's quite an advanced, you know, | 但它是一个相当高级,你知道的, |
| 435 | you talk to a lot of coders, | 你和很多密码员说话, |
| 436 | they won't have heard of DDD, | 他们不会听说过DDD, |
| 437 | but I've just found it extraordinary | 但我刚刚发现很特别 |
| 438 | for building with agents. | 和特工一起建造的 |
| 439 | Interesting. | 有意思,这很有趣 |
| 440 | What do you think about JDD? | 你觉得JDD怎么样? |
| 441 | Have you ever heard of that? | 你听说过吗? |
| 442 | No, what's JDD? | 没有,什么是JDD? |
| 443 | Jealousy-driven development? | 嫉妒驱动的发展? |
| 444 | Telling Claude, hey, Astro's doing this way better than you. | 告诉克劳德,嘿,Astro这样做比你更好。 |
| 445 | What's your problem? | 你有什么问题吗? |
| 446 | What's going on here? | 这是怎么回事? |
| 447 | There's spectrum development. | 有谱系发展. |
| 448 | There's ReadMe-driven development is a fun one. | 有ReadMe驱动的开发是有趣的. |
| 449 | Behavioral-driven development. | 行为驱动发展. |
| 450 | I mean, it's everywhere. | 我是说 到处都是 |
| 451 | And I just saw that Anthropic announced something about, | 我刚看到Anthropic宣布了一些事情 |
| 452 | there's going to be some sort of consequences | 会有某种后果 |
| 453 | or it will prohibit you from being cruel or mean to Claude. | 或者禁止你对克劳德残忍或刻薄 |
| 454 | I don't know if you saw that. | 我不知道你有没有看到 |
| 455 | I just saw that like last night. | 我昨晚就看见了 |
| 456 | So we'll see what happens there. | 所以,我们会看到发生了什么。 |
| 457 | But you have talked a lot on YouTube, | 但你在YouTube上说过很多话 |
| 458 | here on stages about your skills. | 在这里,关于你的技巧的阶段。 |
| 459 | And there's a reason why I use them. | 我使用它们是有原因的 |
| 460 | There's a reason why so many people use them. | 如此众多人使用它们是有原因的. |
| 461 | And what I would like to hear about from you is, | 我想从你那里听到的是, |
| 462 | before all those, | 在所有这些之前, |
| 463 | I'm going to put you on the spot a little bit here. | 我要把你放在这一点点。 |
| 464 | I would love to hear a war story. | 我很想听一个战争故事。 |
| 465 | I really like to learn from people | 我真的很喜欢从别人那里学 |
| 466 | by hearing about failures they've had | 他们听到失败的消息 |
| 467 | or why they've learned a lesson. | 或者为什么他们学到教训 |
| 468 | And I'm sure there was a specific moment | 我敢肯定,有一个特定的时刻 |
| 469 | where you decided, okay, | 你决定的地方,好吧, |
| 470 | I need to build a skill around this | 我需要在这件事上培养出一种技巧 |
| 471 | so I don't run into that issue again. | 所以我不会再遇到那个问题了 |
| 472 | Or you had some sort of big problem | 或者你有个大问题 |
| 473 | inside of your projects or with your students. | 在你的项目或与你的学生。 |
| 474 | I would love to just see if we could look inside your brain | 我很想看看你的大脑里 |
| 475 | a little bit and understand one of these stories | 稍微了解一下这些故事 |
| 476 | that you learned a big lesson from. | 你从中吸取了很大的教训。 |
| 477 | Well, I was really intrigued by a talk that I saw | 嗯,我真的很感兴趣 一个谈话,我看到了 |
| 478 | given by a guy from OpenAI. | 由OpenAI的一个家伙提供。 |
| 479 | It was on the AI engineer, Sean Grove, I think, | 这是在AI工程师,肖恩・格罗夫,我想, |
| 480 | saying that specs are the new code, right? | 说Specs是新的代码,对不对? |
| 481 | Specs are the new code. | 斑点是新代码. |
| 482 | And this theory was that you don't need to write code anymore | 这个理论是 你不再需要写代码了 |
| 483 | because what you can do is you can write a specification, | 因为你能做的是写一个规格 |
| 484 | just write a very detailed document, | 只要写一个非常详细的文件, |
| 485 | and agents can help you do that, | 探员可以帮你 |
| 486 | on how you want the code to be. | 关于你希望代码是什么。 |
| 487 | And then what you do is you then get an agent | 然后你就会得到一个经纪人 |
| 488 | to turn that specification into code. | 将这个规格变成代码。 |
| 489 | And this is still a very, very popular idea. | 这仍然是一个非常,非常流行的想法。 |
| 490 | People love this thing. | 人们喜欢这个东西。 |
| 491 | And there are lots of frameworks built around it. | 周围有许多框架。 |
| 492 | I would say this is spec-driven development, basically. | 我想说,这基本上是由光谱驱动的发展。 |
| 493 | That's a pretty muddy term, | 这是一个相当泥巴的术语, |
| 494 | but that's how I define it at least, | 但我至少是这样定义的 |
| 495 | which is you keep the specs around | 这是你保持的规格周围 |
| 496 | and you just sort of refine them as you go. | 你只是细化了一下 |
| 497 | And I tried doing this. | 我试过这样做。 |
| 498 | I really tried. | 我真的试过了 |
| 499 | This was before I came up with my skills and stuff. | 这是之前,我得到了我的技能和东西。 |
| 500 | I have a repo that is my pride and joy, | 我有一个回扣,是我的骄傲和喜悦, |
| 501 | which is a custom video editor. | 这是自定义视频编辑器。 |
| 502 | It's like a sort of version of Descript. | 类似一种描述的版本. |
| 503 | It's got AI writing, design thumbnails on it, | 它有AI写作, 设计缩略图在上面, |
| 504 | I do fricking everything on this thing. | 我对这玩意一无所有 |
| 505 | And I tried turning that into a big spec | 我试图把它变成一个大谱 |
| 506 | and I tried doing spec-driven development, I suppose. | 我想,我试图做 光谱驱动开发。 |
| 507 | And what I found was I just kept wanting to look at the code, | 我发现的只是我一直想看密码 |
| 508 | first of all, because I was curious, | 首先,因为我很好奇 |
| 509 | and I just saw the code getting worse and worse and worse. | 我刚看到密码越来越差 |
| 510 | And the agent would produce worse and worse results. | 而毒剂会产生越来越糟糕的结果. |
| 511 | And I tried this for months | 我试了好几个月 |
| 512 | because I thought this guy was open AI, right? | 因为我觉得这家伙是开放AI,对不对? |
| 513 | He knows what he's talking about. | 其人知其义. |
| 514 | I'll be able to get this concept down. | 我会把这个概念弄下来的 |
| 515 | And I just thought about it again and again and again, | 我只是一次又一次地想着 |
| 516 | and it wasn't working. | 它没有工作。 |
| 517 | So I realized, no, I'm going to ditch these specs. | 所以我意识到,不,我要放弃这些规格。 |
| 518 | I'm actually going to focus more on the code. | 我将更专注于代码。 |
| 519 | And the thing that triggered it for me was | 触发它的东西对我来说 |
| 520 | code is the environment that the agent operates in, right? | 代码是代理操作的环境,对不对? |
| 521 | Like what I was talking about before. | 如前所说. |
| 522 | If you have an agent that's basically that dog | 如果你有个特工 基本上就是那条狗 |
| 523 | sitting there with a house on fire around it, | 坐在那里,房子着火了, |
| 524 | it's not going to produce good code, right? | 它不会产生好的代码,对不对? |
| 525 | The dog is in a bad environment. | 狗在恶劣的环境中。 |
| 526 | You know, what I wanted was good code | 你知道,我想要的是好代码 |
| 527 | and like code that made sense. | 和像代码 这有道理。 |
| 528 | And if you think about what good code means, | 如果你想到什么是好的代码, |
| 529 | a good code base is a code base that's easy to make changes in. | 良好的代码基础是易于更改的代码基础. |
| 530 | A change is not going to screw up something | 改变是不会搞砸的 |
| 531 | that the agent didn't think about. | 经纪人没想过 |
| 532 | A code base where everything is easy to modify. | 代码基础,所有东西都很容易修改. |
| 533 | And that's what I started focusing on. | 这就是我开始关注的。 |
| 534 | And when I started focusing on that | 当我开始专注于 |
| 535 | and started focusing back on the code, | 开始关注密码 |
| 536 | that's when I felt, okay, | 那时我感觉,好吧, |
| 537 | I'm actually starting to get somewhere | 我开始有所进展了 |
| 538 | and I can kick things into gear. | 我可以把东西踢到装备。 |
| 539 | But the spectrum development just did not work for me. | 但光谱的开发对我并不起作用。 |
| 540 | The models are getting better every time. | 模特们每次都变好 |
| 541 | Every time there's a new number iteration, | 每当有一个新的数字迭代, |
| 542 | they're getting better. | 他们越来越好。 |
| 543 | Why do you think that these models | 为什么你认为这些模型 |
| 544 | don't already understand a lot of these concepts | 不懂这些概念 |
| 545 | or don't natively do them? | 还是不是本地人做的? |
| 546 | And when do you think that'll happen, I guess? | 我猜你认为这什么时候会发生? |
| 547 | So it's about the communication barrier, right? | 所以这是关于通讯障碍,对不对? |
| 548 | Models can't know what you care about until you tell them. | 模特儿不能知道你在乎什么 除非你告诉他们 |
| 549 | And skills are one way of telling the model | 技能是告诉模特的一种方式 |
| 550 | what you care about. | 你关心的是什么 |
| 551 | Now, I think that, you know, models are getting better | 现在,我认为,你知道, 模型越来越好 |
| 552 | and harnesses are getting better. | 绳子越来越好 |
| 553 | I think people underrate the impact of harnesses on models. | 我认为人们低估了驾驭模型的影响。 |
| 554 | 你知道, 利用是Claw Code, Claw.ai, | |
| 555 | 和Codex,随便。 | |
| 556 | And I think then models are obviously going to get better, | 我觉得模型显然会好转 |
| 557 | but they're never going to be good enough | 但是他们永远也不够好 |
| 558 | to read your mind, right? | 读你的思想,对不对? |
| 559 | Never going to be good enough | 永远不够好 |
| 560 | to align with you perfectly every time. | 每次都跟你完全一致 |
| 561 | And so you need some framework around that, | 因此,你需要 一些框架围绕, |
| 562 | some process to help this genius, genius developer | 帮助这个天才开发者的过程 |
| 563 | understand what you want. | 了解你想要什么吗? |
| 564 | Because it takes a really good product manager | 因为它需要一个非常好的产品经理 |
| 565 | and a really good developer to make a great product. | 和真正的好开发者 制造一个伟大的产品。 |
| 566 | You have to care about things. | 你要关心的事情。 |
| 567 | You have to look at the outputs and like add your taste to it. | 你必须看产出 喜欢加入你的品味。 |
| 568 | So I always think there's going to be some layer of process | 所以我总觉得会有一些过程 |
| 569 | over the top of that to just get over | 翻到顶部,只为了结束 |
| 570 | that communication barrier. | 那个沟通障碍 |
| 571 | That makes sense. | 这有道理。 |
| 572 | How much faster do you think that you can move now, | 你觉得你能移动多快? |
| 573 | that you've learned these fundamentals | 你学会了这些基础 |
| 574 | from when you were doing it by hand? | 从你用手做起? |
| 575 | I mean, it's outrageous the difference, right? | 我的意思是,这是令人愤慨的区别,对不对? |
| 576 | Like, it is outrageous the amount of stuff I'm able to ship. | 喜欢,这是令人愤慨的 数量的东西我可以运送。 |
| 577 | Like, it's just not comparable. | 喜欢,它只是不可比。 |
| 578 | Like, I've got like four agents right now | 喜欢,我已经得到 像四个特工现在 |
| 579 | working on different projects. | 从事不同的项目。 |
| 580 | I've given them all long-term tasks. | 我给了他们所有长期的任务。 |
| 581 | I've said, okay, do this and do this and do this. | 我说了,好,做这个,做这个,做这个。 |
| 582 | Here are my long-term goals. | 这是我的长期目标。 |
| 583 | Go and figure it out. | 去想办法 |
| 584 | I've got a sort of new couple of skills | 我有一套新技能 |
| 585 | that I'm playing with like a kind of chief of staff skill, | 我玩的就像 幕僚长的技巧, |
| 586 | which does sort of like sub-agent management. | 有点像副代理管理 |
| 587 | And that's outrageous. | 这太过分了 |
| 588 | That's like a team of eight developers or something | 这就像一个团队 八开发者什么的 |
| 589 | that I've just got access to that are shipping for me. | 我刚刚得到的渠道 是运我。 |
| 590 | And I check in with them | 我向他们报到 |
| 591 | and I make sure their plans are good | 我确定他们的计划是好的 |
| 592 | and I grill and all that stuff. | 我烧烤和所有的东西。 |
| 593 | And I don't know. | 则不知. |
| 594 | I mean, you could look at my GitHub | 我的意思是,你可以 看我的GitHub |
| 595 | and just look at the number of commits that I'm pushing in. | 看看我承诺多少 |
| 596 | And it's in, you know, 200, 300 a day or something. | 一天有200,300个 |
| 597 | Whereas before, you know, 20 a day | 而以前,你知道,一天20个 |
| 598 | I think would have been a good day. | 我认为这将是一个好日子。 |
| 599 | But again, you can look back at my stuff. | 但你还是可以回头看看我的东西 |
| 600 | So it's not comparable, not remotely. | 因此,它不是可比性的,也不是远程的. |
| 601 | I have to ask the question | 我得问个问题 |
| 602 | about your favorite harness and model. | 关于你最喜欢的绳索和模型 |
| 603 | 目前我一直在使用Claw Code和Opus 5.5 | |
| 604 | Opus 5.5 is an excellent model. | Opus 5.5是一个优秀的模型. |
| 605 | I use it on a much lower effort level than most people do. | 我用在比大多数人低得多的努力水平上。 |
| 606 | And this is kind of one of my hot takes, | 这也是我最热的一招 |
| 607 | which is I think most people are using | 我认为大多数人在使用 |
| 608 | way too high an effort level. | 工作水平太高了 |
| 609 | I think people look at effort and they just think, | 我认为人们看 努力,他们只是认为, |
| 610 | yeah, I'll crank that up, right? | 是啊,我会调谐,对不对? |
| 611 | Like I want it to put in effort, you know. | 就像我想让它努力,你知道的。 |
| 612 | And I use it on Opus 5.5 medium effort basically. | 我基本上用在了Opus 5.5中功上。 |
| 613 | And I'm debating going to low effort actually | 而我在辩论 低努力实际上 |
| 614 | because you get much more token efficiency. | 因为你得到更多的象征性效率。 |
| 615 | It replies much faster. | 它的反应快得多。 |
| 616 | And because I've kind of raised the floor on my agents, | 并且因为我有点升 地板在我的代理人, |
| 617 | like they don't need to work that hard | 好像他们不用那么努力 |
| 618 | to produce good code, you know. | 要产生良好的代码,你知道的。 |
| 619 | They just sort of, they just write stuff | 他们只是有点,他们只是写的东西 |
| 620 | that passes the tests and the tests catch it. | 测试通过 测试赶上它。 |
| 621 | So that's what I like at the moment. | 这就是我现在喜欢的。 |
| 622 | But I'm not that much of a polyglot. | 但我不是个多民族 |
| 623 | Like I don't go around like trying | 喜欢我不到处转 喜欢尝试 |
| 624 | all of these models and harnesses | 所有这些模型和手段 |
| 625 | because I think if they're like the model | 因为我认为如果他们像模特儿 |
| 626 | {\fn黑体\fs22\bord1\shad0\3aHBE\4aH00\fscx67\fscy66\2cHFFFFFF\3cH808080}这把绳子很好 {\fn黑体\fs22\bord1\shad0\3aHBE\4aH00\fscx67\fscy66\2cHFFFFFF\3cH808080}很显然 | |
| 627 | to creating good work. | 创造良好的工作。 |
| 628 | But I also think like, you know, | 但我也认为,你知道, |
| 629 | just improve the environment that your agents operate in. | 只是改善你的特工们在其中的环境 |
| 630 | That's where I find the most gains. | 那是我发现最有收获的地方 |
| 631 | I recently moved a project from before. | 我最近从以前搬来一个项目 |
| 632 | I sort of hadn't figured out how to get this project | 我还没想好怎么完成这个项目 |
| 633 | to a place where I could get an agent | 到一个地方,我可以得到一个经纪人 |
| 634 | to like look in the browser and click through it. | 喜欢浏览器并点击它。 |
| 635 | It was a complicated project. | 这是一个复杂的项目。 |
| 636 | It relied on lots of third-party dependencies. | 它依赖许多第三方依赖。 |
| 637 | Like it had OBS running locally. | 就像OBS在本地运行一样 |
| 638 | This was my video editor. | 这是我的视频编辑。 |
| 639 | Like how do I make sure that's available | 譬如我如何确保有这个机会 |
| 640 | and can do that while I'm also using it to record. | 也可以用它来记录 |
| 641 | And, you know, it was quite complicated. | 而且,你知道,这是相当复杂的。 |
| 642 | But I fixed that. | 但我已经修好了 |
| 643 | And now the agent can click through it, | 现在代理可以点击它, |
| 644 | can create screenshots, | 可以创建截图, |
| 645 | can actually run the application properly. | 可实际正确运行应用程序。 |
| 646 | And just that difference has led to a massive increase | 就因为这个差异,才有大幅增长 |
| 647 | in the quality of the stuff I'm able to produce | 在质量 的东西,我可以生产 |
| 648 | because the agent has such better data to access. | 因为特工有更好的数据可以访问。 |
| 649 | And I think I could get, you know, | 我想我可以得到,你知道, |
| 650 | just as good result like with Opus 4.5 and that setup. | 就像Opus 4.5和那套设定一样 |
| 651 | You know what I mean? | 你知道我的意思吗? |
| 652 | Like it's amazing the difference that has made. | 就像它惊人的区别 已经产生了。 |
| 653 | And so honestly, I think working on the environment | 老实说 我想在环境方面工作 |
| 654 | your agent's operating, it just pays you back so much. | 你的经纪人在做手术 只是还你这么多钱 |
| 655 | And I know it's kind of a cop out from the question, right? | 我知道这有点像警察,对吧? |
| 656 | Because you want me to give a hot take of, you know, | 因为你希望我给一个热拍,你知道, |
| 657 | Codex sucks and Claude Cody's great and stuff. | Codex烂透了 克劳德·科迪也很棒 |
| 658 | But I just, I didn't care that much | 但我只是,我不在乎这么多 |
| 659 | about the harness really. | 关于带子,真的。 |
| 660 | I really don't. | 我真的没有。 |
| 661 | You know, it's improving and it's great. | 你知道,它正在改善 它的伟大。 |
| 662 | And Opus 5.5 especially is a much better model than 5 was. | 而Opus 5.5尤其比5型要好得多。 |
| 663 | Much more efficient. | 效率要高得多。 |
| 664 | But really I think of it as like | 但我真的觉得 |
| 665 | just being employees in my company. | 只是我公司的雇员。 |
| 666 | And I want my company to be as good as it can be. | 我希望我的公司尽可能好起来 |
| 667 | So that the base level of employee is better. | 这样员工的基础水平就更好了. |
| 668 | Yeah. | 对 |
| 669 | You know, and I actually really love that answer | 我真的很喜欢这个答案 |
| 670 | because I think if anything that helps people realize | 因为我想如果有什么能帮助人们意识到 |
| 671 | you don't need to be stressing out so much | 你不用那么紧张 |
| 672 | about the newest model drop | 关于最新模式下降 |
| 673 | and making sure you fully understand | 和确保你完全理解 |
| 674 | what it's best at and all these benchmarks. | 最好的标准是什么? |
| 675 | Because, you know, if you think back to | 因为,你知道,如果你想回 |
| 676 | before we even had Opus and we had Sonnet | 之前我们甚至有奥普斯和我们有索内 |
| 677 | and it was a really, really good model. | 这是一个非常,非常非常好的模型。 |
| 678 | Like you could automate tons of stuff. | 就像你可以自动化很多东西 |
| 679 | You can do tons of knowledge work | 你可以做很多知识工作 |
| 680 | and you could do a lot with that model. | 你可以用那个模型做很多事 |
| 681 | So even Opus 5.5 on low is better than that model was. | 所以即使是低点的Opus 5.5都比那个模型要好。 |
| 682 | And I think a lot of people don't realize that | 我觉得很多人不知道 |
| 683 | just because of how fast everything moves. | 只是因为一切动得如此之快 |
| 684 | And, you know, I've always believed that, | 而且,你知道, 我一直相信, |
| 685 | you know, someone like you, Matt, | 像你这样的人 马特 |
| 686 | with a very old model could still produce better output | 用一个很老的模型仍能产生更好的输出 |
| 687 | than someone with Opus 5.5 | 比一个有Opus 5.5的人 |
| 688 | that has no idea of how to communicate. | 根本不知道怎么沟通 |
| 689 | So that's what I think is really cool about that. | 所以我觉得这很酷。 |
| 690 | And I do love that answer. | 我喜欢这个答案。 |
| 691 | Now, one thing that I'm curious about for you as well is | 现在,有件事我也好奇你 |
| 692 | someone that might have been listening | 可能是有人在听 |
| 693 | to this whole podcast so far | 到现在为止,这个播客 |
| 694 | and maybe they're still feeling overwhelmed | 也许他们仍然感觉不堪重负 |
| 695 | about a lot of the stuff that we're talking about. | 关于很多我们谈论的东西。 |
| 696 | But, you know, there's a lot of benefits | 但是,你知道,有很多好处 |
| 697 | and you've talked about how much faster you've been able to move | 你说过你能移动多快 |
| 698 | and we've talked about how the barrier entry is so much lower. | 我们讨论过障碍物的进入程度有多低 |
| 699 | What would be kind of like your final motivational message, | 你最后的动机信息是什么? |
| 700 | your final push to get someone to start taking action | 你最后的冲动是找人开始行动 |
| 701 | and just getting in there, getting their hands dirty | 刚进去 弄脏了他们的手 |
| 702 | and learning more? | 和学习更多? |
| 703 | So I think if you feel the urge to want to get better at this stuff, | 所以我想如果你觉得有冲动 想要改善这些东西, |
| 704 | I have to say that my skills are a great place to start. | 我不得不说,我的技能是一个伟大的起点。 |
| 705 | If you're looking for, just give me the basic idea | 如果你在找,给我一个基本的想法 |
| 706 | of just do this and this and this, | 只要做这个和这个, |
| 707 | then downloading my skills, downloading any harness, | 然后下载我的技能 下载任何绳索 |
| 708 | like just choose from the main ones, | 就像从主要选择, |
| 709 | 从Claw Code, Codex, GitHub CodePilot中选择, 随便什么. | |
| 710 | Don't spend too much time thinking about the harness, | 别花太多时间去想那把带子 |
| 711 | just try it out. | 试试看 |
| 712 | And then from there, it's about, | 然后,从那里,它是关于, |
| 713 | 我最近采访了Lauren Tan, | |
| 714 | potato on Twitter who works at GrokBot. | 在GrokBot工作的推特上的土豆。 |
| 715 | And she talks about climbing the trust ladder, right? | 她谈到爬上信任梯子,对吧? |
| 716 | Climbing the trust ladder. | 攀登信托梯. |
| 717 | Because you are not going to trust your agent | 因为你不会相信你的经纪人 |
| 718 | the first time you see it work. | 你第一次看到它的工作。 |
| 719 | And that is good. | 这是很好的。 |
| 720 | That's a good impulse. | 这是一个很好的冲动。 |
| 721 | You shouldn't be trusting it blindly. | 你不应该盲目相信它。 |
| 722 | You should, you know, when you learn to drive, | 你应该,你知道, 当你学会开车, |
| 723 | what's the first pedal you learn? | 你学的第一个踏板是什么? |
| 724 | Do you learn the accelerator or do you learn the brake? | 你学加速器还是学刹车? |
| 725 | You learn the brake first, right? | 你先学刹车,对吗? |
| 726 | You learn how to stop because you don't want to just push on, | 你学会如何停止 因为你不想只是推, |
| 727 | push on, push on because you're going to burn | 推啊,推啊,因为你快烧焦了 |
| 728 | through your rate limits. | 通过你的利率限制。 |
| 729 | You're going to produce stuff that you didn't want to see. | 你会生产一些你不想看到的东西 |
| 730 | And you're not going to get a feel for the thing | 而你不会得到一个感觉的东西 |
| 731 | that you're using. | 你在用这个 |
| 732 | So what I recommend is just to try opening up the model, | 所以我建议你试着打开模型 |
| 733 | giving it some prompts, try just open up an empty folder, | 给它一些提示,尝试打开一个空文件夹, |
| 734 | use VS Code as well because it is sometimes nice | 也使用 VS 代码,因为它有时很好 |
| 735 | just to be able to look inside the code. | 只为了能够查看密码内部 |
| 736 | And from there, you can start working on some idea. | 然后,你可以开始研究一些想法。 |
| 737 | It really helps to have an idea of the thing | 这真的有助于有一个想法的事情 |
| 738 | that you want to build and to see, have something cool | 你想建造和看到, 有很酷的东西 |
| 739 | in your mind that you can actually build towards. | 在你的脑海里,你可以 实际建设向上。 |
| 740 | Because if you don't have like a reason | 因为如果你没有这样的理由 |
| 741 | or like a mission for doing this, | 或者像一个任务做这个, |
| 742 | then it's a lot harder to get your motivation | 那你的动机就更难了 |
| 743 | and the feedback loop won't be as strong | 反馈循环不会那么强 |
| 744 | because if you're learning stuff | 因为如果你正在学习的东西 |
| 745 | and you're seeing a cool idea come to life, | 你正看到一个很酷的主意诞生, |
| 746 | that's just, it's a multiplier. | 那只是,这是一个倍数。 |
| 747 | And so from there, you start to watch the agent like a hawk, | 因此,从那里开始,你开始 看着经纪人像鹰, |
| 748 | understand everything that it's doing, | 理解它所做的一切, |
| 749 | understand everything that a harness gives you. | 了解所有你得到的东西 |
| 750 | And you can start to feel when you're trusting the agent | 当你相信特工时,你就能感觉到 |
| 751 | and when you understand the agent. | 当你了解特工时 |
| 752 | And when you're at that point, | 当你在那个时候, |
| 753 | then you can start maybe using two agents | 那你可以开始使用两个探员 |
| 754 | or you can start maybe writing a couple of skills | 或者你可以开始写一些技能 |
| 755 | to make the agent do custom things. | 让经纪人做自定义的事 |
| 756 | You can start connecting it up to different data sources. | 你可以开始连接到不同的数据源。 |
| 757 | You can plug it into your Slack, to your Twitter, | 你可以把它插进你的Slack,你的推特, |
| 758 | to your whatever, to your Gmail. | 敬你的东西 敬你的Gmail |
| 759 | And from there, you can start, | 从那里开始,你可以开始 |
| 760 | I've got an agent that handles my to-do list for me. | 我有个探员帮我处理待办事宜 |
| 761 | So I've got, I dictate something to my watch | 所以,我有,我向我的手表下点什么 |
| 762 | and it goes and pings a system and it goes in my to-do list | 它去和拨打一个系统 它去在我的待办事宜列表 |
| 763 | and then the agent triages it and whatever. | 然后毒剂分解它和什么的。 |
| 764 | So you start integrating it in your life | 所以你开始把它融入你的生活 |
| 765 | and integrating it into your work. | 并把它融入你的工作。 |
| 766 | So that's it. | 故曰此义. |
| 767 | Climb the truss ladder, start small, | 爬上梯子,开始小 |
| 768 | start with my skills if you want to or don't. | 如果你愿意或不愿意,就从我的技能开始 |
| 769 | And from there, you can just go nuts with it. | 从那里开始,你可以疯狂地去。 |
| 770 | And it's amazing how if you just pay attention to it | 如果你只注意它,那真是不可思议 |
| 771 | and try to think on its level, | 并且想一想, |
| 772 | it's amazing how quickly you can climb that truss ladder. | 爬梯子的速度太快了 |
| 773 | I love that answer. | 我喜欢这个答案。 |
| 774 | I kind of call that in a lot of my videos | 我有很多视频都这么说 |
| 775 | or that theory, I refer to that sometimes | 或者这个理论,我指的是有时 |
| 776 | as teaching a kid to ride a bike, | 教小孩骑自行车 |
| 777 | like the bike method of how you have to earn that trust | 就像自行车方法 你如何赢得信任 |
| 778 | and you don't start with trying to teach 10 kids | 你不从教10个孩子开始 |
| 779 | at the same time. | 同时。 |
| 780 | You have to watch them closely, you adjust them | 你一定要仔细观察,调整 |
| 781 | and you're still at the end of the day, | 你仍然在最后时刻, |
| 782 | you're not just gonna have 20 kids go ride their bikes | 你不会只有20个孩子去骑自行车吧? |
| 783 | on the highway and then go inside and take a nap. | 在高速公路上,然后进去睡觉。 |
| 784 | You're still very much in the process. | 你仍然很在意这个过程. |
| 785 | So I love that answer. | 所以我喜欢这个答案。 |
| 786 | And I've seen Elon, Boris and Sam, I believe, | 我见过伊隆,鲍里斯和山姆,我相信, |
| 787 | and probably many more people | 也许还有更多的人 |
| 788 | basically say some version of this, | 基本上说这个版本, |
| 789 | give X model your most ambitious tasks | 给 X 模型您最雄心勃勃的任务 |
| 790 | or your most ambitious goals. | 或者你最雄心勃勃的目标。 |
| 791 | And it's really interesting to me | 这对我来说真的很有趣 |
| 792 | that they are coming out and saying this | 他们出来说这个 |
| 793 | because I mean, I totally believe in that, | 因为我相信这一点 |
| 794 | but I would love to hear from you. | 但我很想听到你的声音。 |
| 795 | Why do you think in their minds | 你为什么在他们的脑海里想 |
| 796 | they had to make that a point to tell the world, | 他们不得不把这一点告诉世界, |
| 797 | hey, try our model, | 嘿,试试我们的模型, |
| 798 | but try your most ambitious tasks on this model? | 尝试你最雄心勃勃的任务 在这个模型? |
| 799 | I don't know, because I think the opposite, right? | 我不知道,因为我认为相反,对不对? |
| 800 | I think give it your most menial boring work, | 给它你最无聊的工作, |
| 801 | give it the stuff that you hate doing. | 给它你讨厌的东西。 |
| 802 | Get it to draft replies to people | 写给大家听 |
| 803 | who are asking you for things | 他们问你的东西, |
| 804 | and give it simple achievable tasks | 并赋予它简单可实现的任务 |
| 805 | where you can measure the output | 您可以在其中测量输出 |
| 806 | and measure how good it is, right? | 并衡量它有多好,对不对? |
| 807 | Now, of course, you can use it to brainstorm. | 现在,当然,你可以用它来集思广益。 |
| 808 | You can use it to think about ideas | 你可以用它来思考想法 |
| 809 | and think about how you might turn those ideas into execution. | 想想你如何把这些想法变成处决 |
| 810 | And it's pretty good at that. | 而且它相当擅长这个。 |
| 811 | It's useful to use grill me for that actually. | 用烤肉来烤我有用 |
| 812 | But I would say that the real utility that I get from AI | 但我会说,我从AI得到的真正效用 |
| 813 | and the reason that I use it | 以及我使用它的原因 |
| 814 | is because it helps automate the boring stuff. | 因为它帮助自动化 无聊的东西。 |
| 815 | It helps me focus on the strategic layer | 帮助我专注于战略层面 |
| 816 | and the stuff that I want to focus on. | 以及我想专注的东西 |
| 817 | You know, all of the like, | 你知道,所有的类似, |
| 818 | I have video content that I produce | 我有我制作的视频内容 |
| 819 | and then written content that I produce. | 然后写我制作的内容 |
| 820 | And writing that written content | 写那些书面内容 |
| 821 | has got so much easier since AI came around | 自从人工智能出现后 事情变得简单多了 |
| 822 | because I plug it into Sonic 5.5 | 因为我把它插进Sonic 5.5 |
| 823 | and it produces a first draft. | 并产生初稿。 |
| 824 | I iterate on it. | 我从头到尾都看出来了 |
| 825 | And, you know, that has been amazing for me | 而且,你知道, 这一直是惊人的我 |
| 826 | because it's made my work more interesting to me | 因为这让我的工作更有趣 |
| 827 | because I didn't like writing those things | 因为我不喜欢写那些东西 |
| 828 | before it was a real pain. | 之前,这是一个真正的痛苦。 |
| 829 | And I even used to pay people to do it for me. | 我甚至还经常花钱给别人做 |
| 830 | So, yeah, give it your boring stuff. | 所以,是的,给它你无聊的东西。 |
| 831 | Give it your low-level menial jobs, | 给它你低级的体温工作, |
| 832 | the stuff that you don't need to worry | 你不需要担心的东西 |
| 833 | if it gets wrong or not | 如果它得到错误与否 |
| 834 | because you can just go in and correct it. | 因为你可以进去纠正它 |
| 835 | So, and let it do that | 所以,让它做 |
| 836 | because that's what it's really good at. | 因为那是它真正擅长的。 |
| 837 | So I want to close off here with one question | 所以我想在这里结束一个问题 |
| 838 | to just kind of learn a little bit more | 来学习更多 |
| 839 | about you and where you're headed. | 关于你,你要去哪里。 |
| 840 | So I'd love to hear, | 所以,我很乐意听到, |
| 841 | when you think about the next, | 当你想到下一个, |
| 842 | let's just say two to three years, | 就说两三年吧 |
| 843 | which in the AI world is a decade, | 在AI世界是十年, |
| 844 | what are you really excited about? | 你真的很兴奋什么? |
| 845 | What are you building? | 你在建什么楼? |
| 846 | What are you working on? | 你在忙什麽? |
| 847 | What's something in your life | 你生活中有什么东西 |
| 848 | that's got you excited, | 这让你兴奋, |
| 849 | especially as you think about | 尤其是当你想到 |
| 850 | how AI will evolve that? | AI会如何演变? |
| 851 | So I'm going to prefix this by saying | 所以我要先说 |
| 852 | I'm a teacher. | 师曰. |
| 853 | I'm not a pundit, right? | 我不是一个专家,对不对? |
| 854 | I have no idea. | 我不知道,我还不知道 |
| 855 | - 我在牛津郡 - 是的 - I'm in Oxfordshire. | |
| 856 | Like I'm not in SF. | 就像我没在SF。 |
| 857 | I don't have the ear of anyone important. | 我没有任何人的重要耳朵。 |
| 858 | Like no one is whispering the future into my ear. | 就像没人在我耳边悄悄地说出未来 |
| 859 | I actually reject most offers | 其实我拒绝了大多数报价 |
| 860 | to like preview things | 喜欢预览 |
| 861 | because I just don't care actually. | 因为我根本不在乎 |
| 862 | I just want to make it work now, right? | 我只是想让它工作,现在,对不对? |
| 863 | And that has been working well for me so far | 这对我来说一直很好 |
| 864 | in terms of my sanity too | 就我的理智而言 |
| 865 | because it's easy to get swept up | 因为很容易被扫荡 |
| 866 | in what's going to happen. | 将会发生什么。 |
| 867 | And your question was | 你的问题是 |
| 868 | what is going to happen in three years? | 三年后会发生什么? |
| 869 | Or like what am I excited about? | 或者说我兴奋什么? |
| 870 | Yeah, what are you excited about? | 是啊,你兴奋什么? |
| 871 | I'm excited about finding the right way | 我对找到正确的方式感到兴奋 |
| 872 | of turning my ideas into work, right? | 我的想法变成工作,对不对? |
| 873 | And into code because I still don't think | 并输入代码 因为我仍然不认为 |
| 874 | I've been trying to like, | 我一直在尝试喜欢, |
| 875 | I suppose what I'm excited about right now | 我想,我现在很兴奋 |
| 876 | are software factories, | 是软件厂, |
| 877 | ways that you can automate | 自动化的方法 |
| 878 | initiating work | 启动工作 |
| 879 | and improving code bases over time by themselves. | 并逐步完善代码基础 |
| 880 | I think agents currently | 我觉得现在的特工 |
| 881 | are not very good at thinking long term about a code base. | 不太善于长期思考代码基础。 |
| 882 | And what I'm excited about right now | 我现在很兴奋 |
| 883 | is turning agents into that annoying developer | 将代理商变成那个烦人的开发商 |
| 884 | who's always on the team who says, | 谁总是在团队谁说, |
| 885 | no, we should do it the right way, | 不,我们应该做正确的方式, |
| 886 | not the fast way, right? | 不是这么快,对吧? |
| 887 | We should do it for the benefit | 我们应该为了好 |
| 888 | of the people who come after us. | 那些追随我们的人 |
| 889 | Agents are really bad at that right now. | 特工们现在很坏 |
| 890 | And I think there is a way to make them good at it. | 我认为有办法让他们擅长 |
| 891 | That's what I'm specifically excited about. | 这就是我特别兴奋的。 |
| 892 | And I think that we haven't found the right way | 我觉得我们没有找到正确的路 |
| 893 | of organising lots of agents yet. | 组织了很多特工 |
| 894 | I still think there's huge potential in that | 我仍然认为这有巨大的潜力 |
| 895 | and no one, nothing I've seen has made me go, | 没有人,我所看到的一切 都让我去, |
| 896 | yes, that is the right way yet. | 是的,这是正确的方式了。 |
| 897 | So, and I'm also excited that open models | 所以,我也很兴奋开放的模型 |
| 898 | are going to get better. | 会好起来的 |
| 899 | That's another thing I'm really excited about | 这是另一件事 我真的很兴奋 |
| 900 | because they are getting better. | 因为他们越来越好。 |
| 901 | And eventually we will have an open model | 最终我们将有一个开放的模型 |
| 902 | that's as good as Opus 5.5. | 这和Opus 5.5一样好 |
| 903 | And I'd be delighted with that. | 我会很高兴的。 |
| 904 | That's fantastic. | 这太棒了。 |
| 905 | I don't need the models to be that good, right? | 我不需要模特这么好,对吧? |
| 906 | Because I'm pretty good already, you know? | 因为我已经相当好,你知道吗? |
| 907 | Like I can sort of make up for them. | 就像我能弥补他们一样 |
| 908 | Just having a Opus 5.5 level of open model would be amazing. | 只是有一个Opus 5.5级的开放模型是惊人的。 |
| 909 | And even beyond that, | 更何况, |
| 910 | just sort of getting more token efficiency, | 只是得到更多的象征性效率, |
| 911 | getting a nice plain speaking model | 找到一个很好的普通的演讲模式 |
| 912 | who can ship work quickly. | 他们能迅速运货 |
| 913 | I don't know. | 师曰. |
| 914 | It's cool. | 这很酷。 |
| 915 | It's an exciting time. | 这是一个令人兴奋的时刻。 |
| 916 | It absolutely is. | 这是绝对的。 |
| 917 | Well, Matt, thank you so much for coming on today. | 马特,谢谢你今天能来 |
| 918 | I really enjoyed this conversation | 我真的很喜欢这个谈话 |
| 919 | and thank you for all of the golden nuggets | 并感谢你所有的金色金币 |
| 920 | you dropped on us today. | 今天你撞上我们了 |
| 921 | Where can people find you | 人们在哪里能找到你 |
| 922 | if they want to learn more | 如果他们想学习更多 |
| 923 | and see what else you're up to? | 看看你还想干什么? |
| 924 | AIhero.dev is the place. | 爱荷华德夫就是这里 |
| 925 | I've got a newsletter there. | 我在那里有一份通讯。 |
| 926 | I'm on Twitter. | 我在推特上说 |
| 927 | I'm on YouTube. | 我上YouTube了 我上电视了 |
| 928 | Those are the main three. | 这些是主要的三个。 |
| 929 | Awesome. | 厉害 |
| 930 | Well, thank you so much again, Matt. | 非常感谢你,马特 |
| 931 | And hopefully we'll see you again soon. | 希望我们能很快再见到你 |
| 932 | See you, man. | 再见,伙计。 |
该视频共有字幕 934 条。解锁更多字幕为会员功能，请移动到 价格