本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
设计师用 Claude 的时候，效果通常比工程师用好。
Designers, when using Claude, as opposed to engineers using Claude, would consistently get better results.
这是因为他们用的语言不一样。
And it's because of the language that they use.
我们
We have
得记住，欧洲语境下的设计源自皇室，他们追求独特，因为当时材料很稀缺。
to remember that design in the European sense came from royalty and the desire to be distinctive because they were working with scarce materials.
这个时代的有趣之处在于，当所有人都能接触到各种材料时，那种关于品味的观念就不太适用了。
What's interesting about this era is that this idea of taste doesn't fit when all the materials are available to everyone.
现在大家都在琢磨，大语言模型到底有没有品味。
Right now, everybody's trying to solve whether LLMs have taste.
这些模型里塞满了数百万种对品味的定义。
These models have millions of definitions of taste.
大语言模型是基于人类产出的内容训练的，而不是基于输入。
LLMs have been trained on the output of humanity, not on the input.
所以，促使做出设计决策的那些原因，大模型其实是不知道的。
So what led to a design decision is not something that the LLMs know.
或许可以给在场的设计工程师们一些建议。
Maybe advice for our design engineers in the room.
你们过去是怎么有效传达直觉的价值的？尤其是在面对截止日期，或者另一个可能不够成熟的直觉时？
How have you in the past effectively communicated the value of an instinct versus a deadline or even another instinct which may be less informed?
这确实是个难题，因为人工智能让软件开发变得前所未有的简单。
Yeah, that's a tough one because AI is making it easier than ever to create software.
但当每个人都能通过一句话就生成界面、网站或应用程序时，设计又该何去何从？
But what happens to design when anyone can generate an interface, a website, or an application with a prompt?
有人认为，人工智能会让设计变得商品化。
Some argue that AI will commoditize design.
也有人觉得，它会通过自动化重复性工作，让人有更多空间去专注更高阶的创意决策，从而让优秀的设计变得更加珍贵。
Others believe it will make great design even more valuable by automating routine work and freeing people to focus on higher order creative decisions.
在这场对话中，a16z的普通合伙人阿尼什·阿查里亚与微软设计副总裁约翰·马达，以及Impeccable的创始人兼首席执行官保罗·巴考斯，共同探讨了设计、软件、创造力，以及当人工智能融入创作过程后会发生什么。
In this conversation, a16z general partner Anish Acharya sits down with Microsoft VP of Design, John Maeda, and Impeccable founder and CEO, Paul Bakaus, to discuss design, software, creativity, and what happens when AI becomes part of the creative process.
我想先聊聊设计和技术之间的关系。
I wanna actually talk a little bit about, perhaps to begin, with the relationship between design and technology.
有些人觉得设计更接近艺术，也有人把它看得很实用，比如日常用品的设计。
I think there's some people who may view design as sort of more spiritually close to art and others who see it perhaps in a more utilitarian fashion, you know, the design of everyday things.
约翰，你能不能先说说，你怎么看这两者之间的互动？
So perhaps, John, I can sort of ask you to kick us off and talk to us about how you see the interplay between the two.
很高兴能来这里。
Well, first of all, glad to be here.
我以前在凯鹏华盈的时候，经常想这个问题：为什么在2014年、2015年，设计变得这么重要？
I used to think about this a lot when I was at Kleiner Perkins and thinking like, why is it that design is important in twenty fourteen, fifteen?
到底是什么让它变得重要？
Like, why was it important?
那是因为一家叫Airbnb的公司特别成功。
It was because this weird company called Airbnb was unusually successful.
而设计之所以重要，根本原因在于移动互联网的兴起。
And if you connect to why design became important, it was because of mobile.
在移动设备出现之前，桌面体验差一点也没关系，毕竟大家用得不多。
Before mobile, desktop experiences could be crappy and it was okay because you didn't use them very often.
但移动设备使用频率高，如果体验不好，那可就天天受罪了。
But mobile had high usage and therefore it was bad all the time, it would be painful.
所以，正是从那时起，情况开始发生变化。
So that's when it sort of started to happen.
不过说到现在两者的关系，我特别兴奋能和保罗一起讨论，因为我一直期待着能实现自动设计的这一刻。
In terms of the relationship now, however, I'm so excited to be on this with Paul, because I've been a fan of this moment when we'd be able to auto design.
这个想法其实从九十年代我在麻省理工的时候就有了。
And that's since the nineties when I was at MIT.
那时候我们就觉得总有一天能做到，现在真的实现了。
We thought it was gonna be possible one day, and now it's very possible.
谢谢。
Thanks.
而保罗，我认为正是走在自动设计最前沿的人
And Paul, I think, is someone who's really at the cutting edge of
所有这些。
all that.
太惊人了。
Incredible.
能给我们详细说说吗？
So elaborate a bit for us.
你所说的自动设计指的是什么？
What do you mean by auto design?
现在的实现方式，和当年在麻省理工时你的设想一致吗？
And is the way that it works today, does that match the sort of idea that you had back in MIT?
如果回溯到上世纪80年代，有一位叫穆里尔·库珀的女士，她很特别，因为她是受过专业训练的平面设计师。
Well, I mean, if you go back to the 1980s, there was a woman named Muriel Cooper, who was unusual because she was a trained print graphic designer.
她为麻省理工学院出版社设计过那些书籍封面。
She designed those book fans on MIT Press.
麻省理工学院出版社的标志非常漂亮。
There's a beautiful logo at MIT Press.
她还设计了MIT出版史上其他许多重要作品。
She designed that among other pretty major things in MIT's history publication.
她最早构想出一个世界，在那里人们不想盯着终端屏幕，而是想用Helvetica这样的字体在屏幕上阅读内容。
She was the one who imagined a world where people wouldn't want to use the terminal and might want to use this thing called Helvetica to look at things on the screen.
她预见了整个桌面出版和电子出版的变革。
So she predicted the whole desktop publishing revolution, electronic publishing revolution.
当时，她身边还有一群来自MIT人工智能实验室的人。
And at the time, she also had a bunch of people from the MIT's AI lab.
MIT人工智能实验室当时在思考：如何让机器完成人类擅长的事情。
And MIT AI lab was asking, how do we create things by machines that humans are good at.
于是她的团队专注于如何用当时所谓的‘人工智能’自动设计东西。
And so her lab focused on how to automatically design things with quote unquote AI at the time.
太了不起了。
Incredible.
保罗，说说你的看法
Paul, give us your view on
设计与技术之间的关系和互动。
the relationship and interplay between design and technology.
结果和你当初想的一样吗？
Has it turned out the way you expected?
说实话，我觉得我们还在探索这个问题。
I think we're still working that out, to be honest.
但我确实看到它们正以一些有趣的方式逐渐融合。
But I do see a lot of it converging in interesting ways.
很久以前，我在一次会议上看过约翰关于算法设计的演讲，那时候我一直觉得自己是个怪人，既是设计师又是工程师，从没想过要把自己局限在其中任何一个领域。
I remember seeing one of John's talks about algorithmic design a long time ago at some conference, And I always thought that I was always this oddball that was both designer and engineer, and I never wanted to debox in into any of those areas.
但现实是，现在要为自己打造辅助工具变得容易多了，这些工具能帮你用算法生成设计，用技术大幅提升设计能力。
But I think the reality is that it's now much easier to build yourself tools that help you with that, right, that help you algorithmically create design and supercharge your design with technology.
我觉得这真的非常令人兴奋。
And I think that's really exciting.
所以对我来说，它们根本就是一回事。
So to me, they're one and the same.
它们一直就是一体的。
They've always have been.
约翰管这叫自动设计。
I mean, John calls it auto design.
用约翰的话说，我们其实是在提升那些可以机械化部分的门槛。
I think we're basically, like to use John's words as well, we're kind of, like, raising the floor in areas that can be mechanical.
这特别让人兴奋，因为设计中太多时间都花在了可以自动化的事务上，而不是去思考真正该做什么、该构建什么这些高阶问题。
And I think that's really exciting because so much of design is spent with the things that we can automate as opposed to doing the really high level thinking of what actually needs to be done and what needs to be built.
这正是我特别期待的方向。
And that's something that I'm really excited about.
保罗，你觉得设计中的手工技艺和自动化之间存在张力吗？
Paul, do you think there's a tension between craft and automation in design?
是的。
Yeah.
我觉得确实存在。
I think so.
我觉得最终每个人都要自己决定，希望把多少工作交给自动化。
Think I there's ultimately everybody has to think for themselves how much they want to automate away.
但做出决策、判断该做什么、不该做什么，这些依然离不开人的判断。
There is definitely still the intent and the judgment that is needed from a human side, I would say, on figuring out the what and what not to do.
我特别敬佩像史蒂夫·乔布斯这样的人，他可以说是世界上最好的编辑之一。
I actually really greatly respect people like Steve Jobs because, I mean, Steve was one of the greatest editors in the world.
对吧？
Right?
他一生中拒绝的东西，远比答应的要多得多。
He said no to so many things as opposed to saying yes to so many things.
我觉得，拥有自己的立场依然非常重要。
And I think having a viewpoint is still important.
所以我认为，机器最终是为你服务的。
So I think, ultimately, the machine can work for you.
就像你用一台好相机拍照，关键还是你得清楚自己想拍什么。
But the same way, you take a photo on a good camera, ultimately, you figure out what you want to take a photo of.
现在的相机里头做了大量后期处理。
There's a lot of post processing in the camera.
对吧？
Right?
你看现在的iPhone相机，设备上运行的处理简直多得不得了。
There's so much I mean, if you look at an iPhone camera today, there's so much happening on device.
你拍出来的那张照片，早就不是真实世界的样子了。
That photo that you took is literally, like, not reality anymore.
对吧？
Right?
但还是能看得出人的审美和意图。
But there's still some sort of human viewpoint.
约翰，说到你这边，你觉得现在像Cloud Code、Codex这些智能代理在设计表达上有哪些局限？
John, maybe to bring it to you, what do you think are the limitations of the sort of the Cloud Code, Codex, and all of the agents today when it comes to design expression?
天啊。
Oh, wow.
说实话，大家都说这玩意儿无限可能，什么寒武纪大爆发啊，乱七八糟的。
Well, I mean, everyone's saying it's limitless, and it's that it's not the whatever, the Cambrian period and blah, blah, blah.
一堆形容词堆出来，哎呀妈呀，太厉害了。
There's so many adjectives that say, like, oh my gosh, amazing.
我觉得，这其实早就该来了。
I think the reality is that this has been a long time coming.
就像以前文案自动化出现之前，营销素材早就开始自动化了。
In the same way we saw writing being automated before LLMs, and we saw marketing assets being automated as well.
现在不过是把这种趋势往上推，推到了设计这个环节——就像保罗说的，这需要克制。
So we would just spread that to the kind of we went upstream on this idea of designing things, which to Paul's point, it requires restraint.
需要很多别的东西，跟工程思维完全相反。
It requires all kinds of things, which is the opposite of engineering.
我本身是工程师，说白了，只要我能做出来，我就想把它建起来。
Speaking as a engineer as well, if I can make it, I'm a build it.
哦，这个我也能搞。
Oh, I'll do this.
我会一直这么做，这么做，再这么做。
I'll keep doing this and this and this and this.
除非你是架构师，否则克制根本不在考虑范围内。
And so restraint isn't part of the equation unless you're an architect.
设计师总是像设计人类体验的架构师一样思考。
Designers always think like architects of the human experience.
现在，工具和模型结合在一起，再加上评估机制，我们就有了完整的反馈循环。
Now with the harness and the models coming together, we have a nice combo platter, plus the evals, we have the feedback loop.
这其实是全新的东西。
That's actually really brand new.
所以我才这么兴奋。
That's why I'm so excited.
我们一直在和保罗做实验。
We've been experimenting with Paul.
我们有个叫 GitHub Copilot 的应用。
We have this thing called the GitHub Copilot app.
这是一个新应用。
It's a new app.
难点在于如何保留人类的品味与匠心，同时融合所有高科技元素。
And the challenge with that has been to keep the human taste and craft and all the high-tech goodies.
正如保罗所说，提升底线之外，人类也应该能拓展自己能力的上限。
And to Paul's point about raising the floor, humans should be able to raise the ceiling of what they can do.
人类动画师负责创作和最终决策，处理那些极致精细的部分，而代码代理则承担那些繁琐重复的工作。
So we have human animators making the craft and the final decisions around some of the super duper high extreme polish, but we want to have the coding agents do the part that have the drudgery.
是的。
Yeah.
这完全说得通。
That makes perfect sense.
我想深入聊聊这个新的 Git 产品。
I wanna pull the thread about the new Git product.
保罗，不如你先退一步，给我们讲讲 Impeccable 吧。
Paul, why don't we actually take a step back and you can talk to us about Impeccable?
当初的初衷是什么？
What was the intention?
现在它是什么样子？
What does it look like today?
接下来聊聊你和约翰是怎么合作的吧。
And then let's talk a bit about how you and John are working together.
是的。
Yeah.
当然。
Of course.
Impeccable 最初是我自己需要的一个开源项目。
So Impeccable started as an open source project that I needed for myself.
过去一年，我给几个朋友做外包，想看看能把 AI 工程推到什么地步，能不能让自己效率提升十倍、五倍，或者更多。
So I used to contract for a couple of my friends in the last year to figure out how far can I push AI engineering and how much can I 10x or 5x or whatever myself?
我知道现在很多人都在经历类似的探索。
And I know many are going through that journey right now.
在这个过程中，我为自己开发了很多工具，其中一个是 Impeccable。
In the process, I had to build a whole lot of tooling for myself, and one of those tools was impeccable.
后来我发现，每次我让 Claude 或 Codex 改进代码时，它们都做得特别差，因为它们根本不知道‘更好’在设计语境里到底指的是什么。
And ultimately, I noticed that whenever I would tell Claude or Codex, make this better, it would do a really terrible job because it didn't know what angle of better, what does better mean in the design context.
对吧？
Right?
我还注意到，设计师用 Claude 时，比工程师用的效果要好得多。
And I also noticed that designers, when using Claude, opposed to engineers using Claude, would consistently get better results.
这是为什么呢？
So why is that?
对吧？
Right?
模型明明是一样的。
The model is literally the same.
网站也完全一样。
The website is the same.
设计系统可能是一样的，但设计师用同样的机器提示，出来的结果就是更好。
The design system might be the same, but the designers still get better output even though they're prompting a machine.
这背后的原因在于他们使用的语言。
And it's because of the language that they use.
很多时候，工程师根本不会用那些词。
It's because oftentimes engineers don't use the words.
比如垂直节奏、留白、让这个更醒目一点，或者更低调一点。
Things like vertical rhythm or negative space or make this bolder or quieter.
他们缺乏像资深设计师那样长期积累的术语体系。
They don't have the same vocabulary as a designer who's been in the game for a long time.
所以我第一个想法就是，把这些术语直接整合到智能体的框架里。
So the first instinct that I had was to bring that vocabulary to the actual harness, to the agent harness.
这就是Impeccable的第一个版本，效果立刻就提升了很多。
And that was the first iteration of Impeccable, and that already made a huge difference.
有意思的是，我本来以为这会把设计师推开，主要吸引工程师，但事实完全不是这样。
And interestingly, I thought that it would alienate designers, and it would mainly appeal to engineers, but that was not the case.
现在很多人都在要求设计师更贴近代码，我觉得这很合理，也许设计师自己也开始意识到这一点，因为大家都在往代码靠拢。
I think a lot of designers are being asked right now to move closer to code, and I think that makes a lot of sense, and maybe they're waking up themselves to that idea because everybody's moving closer to code.
代码当然是智能体所依赖的基础。
Code is the substrate that the agents use, of course.
而Impeccable提供了一种更快抵达目标的方式，而且用的是设计师本来就熟悉的语言。
And Impeccable feels like a way to get there much more quickly and with the language that they use.
如果现在看目标用户，大概一半是工程师，一半是设计师，还有很多产品经理、创始人和独立开发者，Impeccable就是这么慢慢成型的。
So I would say it's actually like, you know, if I look at the target audience right now, it's half engineers, half designers, a whole bunch of PMs as well, founders, solopreneurs, but that's how Impeccable came together.
它是一个完全开源的智能体技能工具。
It is an agent skill that's completely open source.
它内置了大量指令来精准控制输出。
It has lots of subcommands to steer.
我在上面又加了一个可视化操作模式，还搭了一层质量过滤机制，专门清理那些粗糙的细节。
And then I kinda built a visual iteration mode on top of it, and a quality layer that removes slop.
当然，这些我们也可以继续聊。
And of course, we can talk about that too.
但它确实能防止模型过度拟合，所以不是所有东西都像我所说的‘克劳德米色’那样，或者那些乱用斜体和紫色渐变的工具。
But it does stop the overfitting of the models, so not everything is Claude Beige, as I call it, or, you know, instruments that riff italics or purple gradients.
不过，归根结底还是这三样东西。
But but, yeah, it's ultimately these three pieces.
所以，词汇库是一个质量层，再加上一个能在你自己的代码库中运行的视觉迭代模式。
So vocabulary, it's a quality layer, and it's a visual iteration mode that runs in your own code base.
掌握了这套词汇，就像掌握了一组魔法咒语。
And knowing the vocabulary, it's sort of like a set of magic spells.
你知道的。
You know?
要么你会，要么你不会。
Either you know them or you don't.
我很喜欢Impeccable让这些技能变得更普及的方式。
And I love how Impeccable makes them more broadly available.
约翰，你能不能聊聊保罗用Impeccable做的这件事，跟凯的工具之间有什么相似之处？
John, talk to us a bit about the analog between what Paul's doing with Impeccable and Kai's power tools.
上世纪80年代末，有个叫Photoshop的软件特别厉害，一下子让人能画画、摆弄照片，想干嘛都行。
Well, in the 1980s, late 1980s, there was this application that was called Photoshop that was amazing because it suddenly let let you draw pictures and play with photographs, etcetera, whatever.
那时候摄影师和相关行业的人特别喜欢用它。
It was popular among photographers and people like that.
但等Kai's Power Tools一出来，Photoshop的用户群就被炸开了，因为大家能用算法搞出各种稀奇古怪的脸部效果和滤镜。
But when something called Kai's Power Tools came out, it blew up the TAM for Photoshop because people could do weird algorithmic things with their faces or all kinds of filters.
这是因为Adobe Photoshop开放了插件架构，允许开发者写代码，做成插件形式加进去。
Because Adobe Photoshop opened up this plug in architecture so people could write code, could put things in that plug in format.
所以我看到Paul的作品时，就想起了Kai's Power Tools，不过是用在设计领域，没那么怪，但特别专注、特别实用。
And so when I saw Paul's work, it reminded me of Kai's power tools, but for design and less weird, but very focused, very functional.
我第一次看到Impeccable时，忍不住想起自己那个年代——PostScript刚出来的时候。
When I saw Impeccable, it reminded me of, just giving away my era, of when Postscript came out.
很多人不知道，PostScript其实是个奇迹，Adobe那帮人，特别是Warnock，写的PostScript手册简直完美。
Most people don't realize that Postscript was a miracle because the Adobe guys, the Warnock, the manual for Postscript is beautiful.
它的设计实在太出色了。
It's incredibly well designed.
那些工程师对平面设计的理解，比很多设计师还要深。
Those engineers understood graphic design better than most graphic designers.
所以 PostScript 准确地定义了一套完美的基础元素，来实现当时主流的视觉设计。
So Postscript properly encoded the perfect set of primitives to implement visual graphic design of the day.
这就需要既懂技术、又有设计眼光的人。
And so it required people who were technical and also design minded.
就像更早之前的 Donald Knuth，他在技术领域也是这样。
Same with Donald Knuth, before that with tech.
他是一位了不起的排版专家，也是一位数学家。
He was an incredible typographer, mathematician.
所以 Tech 在功能与形式的结合上，简直是个奇迹。
So tech was a miracle in the combination of function and form.
当我看到 Impeccable 时，虽然不想让 Paul 尴尬，但我真的觉得，这简直就是 PostScript 诞生那一刻的重现。
So when I saw Impeccable, not to embarrass Paul, I was like, woah, this is like the postscript moment.
终于有人明白了，这根本不是什么滤镜那么简单。
Someone actually gets that it's not just filters.
这还关乎做减法。
It's also about subtraction.
设计师可能会考虑的各种维度全都在里面。
And it was all the different dimensions a designer might take.
后来我看到一位特别出色的工程师兼设计师，这让我特别兴奋。
So and then I saw a really skilled engineer designer, and that got me excited.
这太了不起了。
That's tremendous.
太酷了。
That's so cool.
约翰，你能不能聊聊你们接下来打算怎么合作？
Maybe, John, talk a bit about how you guys are planning to work together.
哦，我特别兴奋，因为我最近接手了整个 GitHub Design 团队。
Oh, well, I'm super excited because I recently picked up all of GitHub Design.
时机正好，因为我们刚推出了 GitHub Copilot 这款已经正式上线的应用。
Great timing because we were launching this GitHub Copilot app that's already in GA.
这和我们过去做的东西不一样，因为现在人们对应用的品质要求已经非常高了。
And it is different than things we've built in the past in that, as you know, the craft bar is very high now for applications.
所以我们想让这款应用树立一个新的标杆。
So we wanted to set the bar with this app.
当产品负责人马里奥跟我说，‘你听说过Impeccable这个东西吗？’
And when the head of product, Mario said, Hey, do you know this Impeccable thing?
我当时就说，天哪，这可是我最欣赏的技能之一。
I was like, Oh my gosh, it's one of my favorite skills.
于是我联系了保罗，保罗也说，‘哇，我们是不是早就认识？’
So I reached out to Paul, and Paul said, Hey, oh my gosh, maybe we know each other.
之后我们通过短信聊得特别投缘。
So we had this wonderful conversation by text.
几周后，保罗将成为我们首个整合这项顶尖设计能力的人，因为我们相信，在这个数字时代，设计至关重要。
And since then, a few weeks later, Paul is going to be the first one to be in our integrations of this powerful design skill, because we believe that design is important in the digital, what do you call it, harness age.
我觉得保罗对它的理解最到位。
I think Paul has the best implementation of it.
还有，保罗，你大概是怎么
And, Paul, how do you sort of
看待这个产品在新 Git 里，也就是在新 Git 产品里呈现出来的呢？
see the product showing up within the new Git Impeccable showing up within the new Git product?
首先，我真的非常兴奋，能把 Impeccable 交到更多的开发者、创业者和独立企业用户手里。
First of all, I'm super excited to get Impeccable into the hands of into the hands of more builders and entrepreneurs and individual enterprises.
我觉得 GitHub 显然是实现这一目标的最佳平台。
And I think GitHub is obviously a great place to to do that.
现在大家都在用 GitHub，这是把 Impeccable 推广给更多人的一个好办法。
Everybody's using GitHub today, and it's a great way to roll out Impeccable to more.
我觉得把这个技能做成内置工具来开发，是一回事。
I think the building in the skill to as a built in tool is one thing.
我觉得从这儿还能拓展出很多不同的方向，这也是我特别期待的地方。
I think there's a lot more different directions that we can take from there, and that's what I'm really excited about.
GitHub Copilot 应用里有个挑选和打磨模式，让你可以选中某部分内容进行修改，快速编辑。
The GitHub Copilot app has a pick and polish mode that allows you to pick and change something and quickly edit something.
当然，Impeccable 本身也有自己的实现方式。
And, of course, Impeccable has its own implementation of that.
目前你需要手动开启，操作起来还有点麻烦。
And right now, you need to opt in, and it's a bit clunky.
我觉得通过像 GitHub 这样的官方集成合作，我们能大大改进这一点，让体验顺畅很多。
And I think through first party harness integrations, like the collaboration that I'm doing with GitHub, yeah, we can actually improve that significantly and make it so that it's a lot smoother.
现在还没有什么正式消息，但我非常期待和约翰、马里奥一起优化 Impeccable 和 GitHub 的整合。
There's nothing there to be announced yet just now, but I'm super eager eager to work with John and and Mario on actually improving the Impeccable and GitHub integration.
明白了。
So okay.
接下来有个问题，保罗。
So follow-up question, Paul.
Git 现在无疑是全球最重要的平台之一，尤其是在当下这个阶段。
So, obviously, Git is one of the most important platforms in the world, especially at this moment in time.
我们不妨往前看，想象一个场景：很多人已经开始用新版 GitHub 和 Impeccable，生成的成果远比模型默认输出的要好得多。
So let's wind the clock forward and imagine a world in which many people are using the new GitHub product and Impeccable to have, you know, these these sort of designs that are much improved from what you get default from the models.
当这些设计开始应用于下一代模型时，Impeccable的下一个版本是否需要彻底转向不同的方向，还是说这些其实是一些始终有效的最佳实践，不管模型怎么变都适用？
As those designs become in distribution for the next gen of models, do we then have to sort of does the next version of Impeccable have to steer in a different direction altogether, or are these sort of evergreen best practices that live regardless of what's in or out of distribution?
是的。
Yeah.
这是个很好的问题。
That's a great question.
其实这种事情已经在发生了。
You know, that's already happening.
这才是最有趣的地方。
That's the really interesting thing.
在2022年、2023年的时候，大家都觉得AI设计就是紫色渐变。
So in 2022, 2023, everybody thought AI slob was purple gradients.
说实话，现在的模型已经不再这么做了。
And and to be honest, the models aren't really doing that anymore.
模型现在根本不会再生成紫色渐变了。
The models aren't producing purple gradients anymore.
我的意思是，也许一些低端模型还会，但总体上已经不会了。
I mean, maybe like, you know, cheap models, but overall, they don't.
我觉得整个设计潮流已经转向了米色背景、淡色调背景、工具图标和眉形文字。
I think the world has moved on to beige backgrounds, tinted backgrounds, instrument, eyebrow text.
现在如果你做界面设计，一眼就能看出那些典型的AI痕迹。
You know, there are the typical AI tells that you look for now if you if you use cloth design.
你经常会遇到这样的设计，比如这种，但这种风格并不只限于界面设计。
Oftentimes, you'll you run into a design like that, for example, but it's not unique to cloth design.
那这到底是怎么发生的？
So what we how did this happen?
对吧？
Right?
我们当时就定了个规矩：不准用标准渐变。
We basically said, okay.
结果模型就转向了潜在空间的其他区域，挑了下一个最接近的方案。
No proper gradients, and then the model steers into a different part of the latent space and just picks the next next best thing.
没错。
Right.
所以最终，这种风格总在变，每当你换到潜在空间的另一个区域，大家一哄而上，结果就变成了算法生成的优衣库或者宜家风。
And so ultimately, SLOP is a moving target, and every time you move to another part of the latent space and everybody adopts it, you get kind of like algorithmic, Uniqlo, or IKEA or something.
对吧？
Right?
但这并不一定就是坏事。
And now that's not necessarily bad.
不代表设计本身差。
It doesn't mean the design is bad.
只是说明这设计缺乏独特性。
It just means the design is not unique.
如果你要做一个能吸引注意力的落地页，那独特性就很重要。
And you want uniqueness if you're building something like a landing page that attracts people's attention.
所以我做 Impeccable 时，就一直在调整这些规则，目的是打造这类反跟风的机制。
And so what I've tried to do with Impeccable and I'm constantly updating those rules, but what I tried to do with Impeccable is to build these types of anti attractors.
比如说，与其只是换用那些被用烂的字体，它会根据你的一段简短用户访谈，甚至运行一些脚本生成随机值，来把设计引导到潜在空间里一个完全不同的方向。
For for example, instead of just going to a different font, to the overused fonts, it actually tries to create a random seed that picks something based on a small user interview with you and maybe even some scripts that run that create randomness to actually steer it into a completely different space of the latent space.
颜色我也这么处理。
I do it with colors.
字体我也这么处理。
I do it with fonts.
我还会在很多其他方面都这么做，就是为了制造一些独特性。
I do it with a bunch of different things to create some sort of uniqueness.
后面我还会加更多，但我真的不希望所有东西最后都长得一模一样。
And there'll be more there that I'm gonna do, but I really do want to make sure that not everything ends up looking the same.
说起来，我以前就干过这种事。
I ended up ended up doing that before, by the way.
比如为什么紫色渐变会变成现在这样？你知道原因吗？
Like, the reason why purple gradients do you know the reason why purple gradients are purple gradients?
说来听听。
Tell me.
Tailwind
Tailwind.
Tailwind 的默认主题里用了紫色。
So Tailwind's default theme is has purple.
明白了。
I see.
因为默认主题的关系，紫色渐变就到处都是了。
And because of the default theme, the purple gradients were introduced everywhere.
现在大家普遍这么认为，这也是主流说法。
That's at least the common saying right now, the common theory.
说来讽刺，我以前也把网页搞成了橙色。
And ironically, I was at fault of coloring the web orange before.
是的。
Yeah.
原因是我刚开始做 jQuery UI 的时候，我们搞了个主题生成器，默认主题就是橙色。
And the reason why is because when I started jQuery UI we've created a theme framework, a theme roller kind of thing, and the default theme was orange.
我不明白为什么是橙色。
I don't know why orange.
我当时也不清楚为什么选了橙色，但一夜之间我把整个网页都染成了橙色，从此长了教训。
I'm not sure why I picked orange at the time, but I colored the whole web orange overnight, and I learned my lesson.
默认设置的力量，确实如此。
The power of defaults, Yes.
毫无疑问。
For sure.
约翰，你觉得独特性在设计中重要吗？
John, do you think uniqueness is an important attribute in design?
顺便聊聊你对独特性和定位的看法吧。
And maybe broadly talk about your view on uniqueness and slot.
保罗刚才提到，我觉得我们现在正处在一个时代，你看那些新成立的公司，都在开发设计模型或者视觉设计模型之类的。
Well, Paul was describing, I think we're in this era because you can see all the new model companies that are building designer models or visual design models, whatever.
我们正处在一个设计越来越商品化和自动化的阶段，新的设计形态因此诞生。
We're at this era where design will be further commoditized and automated so that a new kind of design emerges.
我觉得这正是接下来会发生的趋势。
I think that's what is the next wave that'll happen because of this.
每年我为西南公司做的设计与科技报告里，我一直主张从用户体验转向代理体验。
In the design and tech report I give every year for Southwest, I've been pushing moving from UX to AX, agentic experience.
代理体验是不依赖视觉的。
An agentic experience is non visual.
它属于机器人加文本的世界。
It is the world of robots dot text.
它是大语言模型加文本的世界。
It's LLMs dot text.
它是命令行加帮助参数的世界。
It's command line dash dash help.
我们正在为代理功能设计，而不仅仅是视觉功能。
It's a world where we're designing for agentic affordances instead of just visual affordances.
我相信，一旦我们能逐步淘汰更多传统用户体验设计，这种倍增效应就会出现。
That force multiplication will occur once we can retire more of UX, I believe.
但这需要一大批自动化设计工具，能覆盖其中80%的情况。
But it'll require this surge of automated design solutions that cover 80% of those cases.
剩下的20%留给真人来做，但我相信，这些人最终会转向AX领域。
20% can be spent by real humans, but the humans will move on to AX, I believe.
有意思。
Interesting.
你能再详细说说吗？
Can you say a bit more about it?
我的意思是，在AX里，人的品味和判断力该扮演什么角色？
I mean, what would be the role of human taste and judgment in AX?
哦，其实每个设计师迟早都会明白这一点。有个著名的字体设计师叫埃里克·施皮克曼，他设计了一款非常流行的无衬线字体，叫Meta。
Oh, well, the thing that every designer knows eventually is that and actually, there was a famous typographer, Erik Spiekerman, who designed a very popular sans serif typeface called Meta.
那时候Meta还没成公司名字。
This is before Meta became a company name.
他在旧金山的公司叫Meta Design。
His company in San Francisco was called Meta Design.
那时候，那是全球顶尖的设计公司之一。
At the time, one of the top design firms in the world.
Meta Design 这个名字恰恰说明，设计要么是手艺的直接应用，要么是手艺的抽象化。
Meta Design speaks to the fact that design can either be the application of craft or the abstraction of craft.
我觉得，目前来说，手艺的抽象化才是设计事物的最佳方式。
And so I think of the abstraction of craft is, at the moment, what is the best way to design things?
以前，一个设计师要是能拍下 Helvetica 字体，用美工刀裁剪，再用浆糊贴上去，那就已经很了不起了。
There was a time where being a design craftsman meant you could take a picture of Helvetica, cut it with X Acto knife, put rubber cement on the back, and you were amazing.
后来我们进入了桌面排版时代，开始使用更高层次的页面元素。
We moved to different higher level primitives of the page, etcetera, with desktop publishing.
现在，我觉得我们会进一步走向更抽象的层面。
Now I think we're going move even more meta.
这意味着设计师会更接近程序员，但不是普通程序员，而是交互式的程序员。
And so it means that designers are closer to programmers, but programmers of interaction.
我觉得，保罗在讨论评估系统的时候，很可能潜意识里就是这样在做的，保罗。
And I think Paul, probably he subconsciously is working in this way when we talk about evals, Paul.
评估系统本质上就是一种真正的元设计，存在于潜在空间的构建中。
Evals live in latent space construction as literal meta design.
我觉得这话没错。
I think that's true.
我觉得这里头确实有一整片设计领域。
I think there is a whole area of design here.
我也觉得你在评估系统上的看法大概是对的。
I also think you're probably right with the evals.
关于AX，有意思的是，很多时候你设计的对象不是人，而是智能体。
And the interesting thing with ax is that oftentimes, here, you're designing for a different audience, and that's the agent, right, as opposed to the human.
我觉得我的观点是，这片设计领域会变得越来越重要。
I think there's I guess my viewpoint is that that area of design will be increasingly important.
同时，以用户为中心的设计依然有它的空间。
I also think there's still the area of of human centric design as well.
而且就像我们一开始说的，设计正越来越转向高阶工作——也就是让东西脱颖而出的最后那20%。
And I think, I mean, to the to the point that we made at the very beginning, that is shifting towards high level work, right, sort of the last 20% of what makes something unique.
这其实挺有意思的。
And it's it's interesting.
我们已经看到这么多AI初创公司都在试图赋能创意工作，不管是音乐、视频还是别的什么。
I mean, we we've seen so many AI startups that are trying to enable creative work, right, whether it's music or video or whatever.
但目前为止，我们还没看到来自Runway或其他视频初创公司的下一个宫崎骏。
But we haven't really seen the next Hayao Miyazaki from Runway or some other video startup.
对吧？
Right?
我觉得这么说应该算公允吧，我们确实还没看到。
We haven't really seen I think it's fully fair to say.
音乐领域的AI初创公司里，我们也还没看到下一个比莉·艾利什或者Copilot那样的人物。
I I don't think we have seen our music AI startups next, I don't know, Billie Eilish or Copilot.
这是为什么呢？
And why is that?
我觉得现在大家的重点大多放在提升底线上了。
I think I think a lot of focus right now is is raising the floor.
回到这一点，我觉得现在大家太忽视提升上限了，我对此特别感兴趣。
And and going back to that, I think that that raising the ceiling, I think there's too too little time spent on that specifically, and I'm really excited about it.
所以我希望打造一些工具，既能照顾到基础需求，也能满足高端需求，兼顾两头。
So I want to build tools that all look also cater to that, so kinda like both ends of the spectrum.
但我确实觉得，世界正在向更高端的体验转变。
But I do think that that the world is shifting towards higher end experiences.
用户也在要求更高的品质。
The audience is demanding more too.
这挺有意思的。
That's interesting.
对吧？
Right?
如果所有东西都看起来精致整洁，你知道，当我们提升了基础水平，就算一个设计粗糙的网站，它也照样是个正经网站。
Like, if everything looks relatively polished and clean, you know, if we're raising the floor and, you have a claw designed website, looks claw designed, but it's a legit website.
对吧？
Right?
这个网站足够好，能完成任务就行。
It's a good enough website to get the job done.
真正能拉开差距的，就是我们在这基础上加的东西。
Then the only differentiator would be whatever we craft on top of that.
我一直觉得这一点特别有意思。
And I always thought that was very interesting.
我和很多设计工程师朋友一样，花了很多时间去打磨一些特别独特的东西，比如一个动画特别流畅的滑块。
I spent a lot of time, like many of my design engineer friends, we spent a lot of time crafting something really unique and particular and sort of like, I don't know, a slider that is beautifully animated.
对吧？
Right?
但很多时候，这些努力都白费了。
But oftentimes, that that work would drown.
这些工作根本看不到高层，他们有完全不一样的业务目标。
That work would not be seen by higher ups who, you know, have completely different business goals.
现在正是这些人发光的时候。
I think it's the moment for these people right now to shine.
这是我其中一个观点。
That's one of my theories.
我觉得那些能把握住最后那百分之十、二十，给东西注入独特风格的工匠，也很有意思。
I think the craftsmen that can kinda grasp the last 10%, 20%, and really put a unique spin on something, I think that's also really interesting.
我也用AX，但我更期待那种更进一步的人性化体验。
So I I too I do subscribe to AX, but I think there's also that next level human experience that I'm very excited about.
对。
Yeah.
这两种都很有意思。
They're both interesting.
说到智能代理这个话题，我在想，代理会重视美感吗？
On the agentic topic, where my mind goes is, will agents value aesthetics?
它们会有自己的审美标准吗？
Will they have their own definition?
毕竟，看起来美观或高品质的东西，其实传递了我们人类看重的信息。
Because you could argue that something that is aesthetic or appears to be high quality, there's information in that that we as humans value.
我们做决定时，至少部分是基于这一点的。
We make decisions at least in part accordingly.
对智能体来说，会有什么类似的机制吗？如果有的话。
What will be the analog for agents, if anything at all?
我不确定。
I don't know.
约翰，你怎么看？
John, what do you think?
速度、清晰度，还有啰嗦程度，以及各种边缘情况。
Speed, clarity, things that and and verbosity, and all the edge cases.
我最喜欢的一个例子就是，你得有个特别棒的帮助文档。
So the best example I like is the one of you have to have a really good dash dash help.
你得有出色的错误提示。
You have to have great error messages.
保罗说得对，一旦我们能把所有通用的优质设计都自动化掉，那提升空间就大了，这会非常令人兴奋。
But I and Paul is correct in that the ceiling raising can occur once we can wick away all the generic good quality design, and that's gonna be extremely exciting.
这让人想起19世纪末的手工艺运动时期。
It echoes the era of the arts and crafts movement of the late eighteen hundreds.
当机器接管之后，人类发展出了更高层次的工艺，这种情况也会再次发生。
When machines took over, humans developed this next level of craft, And so that will occur.
我觉得保罗所设想的那种工具，恰恰能推动这种转变。
And I think tools like what Paul is thinking of can enable that.
同时，从功能角度来看，AX这边还有很大的发展空间，我认为擅长视觉设计的人会更快突破，因为这里面有太多棘手的问题，普通工程师的思维未必能应对得当。
At the same time, from a functional perspective, there's so much runway on the AX side that I believe that primarily visual design thinkers will be able to crack that faster because there's so many gnarly problems in it that an average engineering mind will not be the right shaped mind, I believe.
我觉得你说得很有道理，我能不能赶紧回应一下。
I think that's actually a good point, if I can respond to that real quick.
我觉得一个很好的例子就是API设计，还有你提到的命令行界面设计。
I think a great example would be API design and also, like, CLI design, to your point.
对吧？
Right?
你刚才提到帮助文档，是这个意思吗？
That that do you mention help?
展开剩余字幕（还有 277 条）
但我一再听到这种说法，我也同意。
But I'm hearing this over and over, and I agree with it.
目前，智能代理本身还不是很擅长设计这些界面。
Agents themselves are not really good at designing those interfaces yet.
它们在设计 API 接口、命令行界面方面都不太行，这有点讽刺，因为现在它们恰恰是这些工具的使用者。
So they're not really good at designing API interfaces, designing CLIs, and that's kind of ironic because they're the user for them now.
但我确实觉得这话有道理。
But I do think that's right.
我觉得信息架构也是个很好的例子。
I do think information architecture is another great example.
对吧？
Right?
只有特定类型的设计师才特别擅长打造清晰的导航结构。
It's a certain type of designer that's very good at crafting a good navigation.
是的。
Yeah.
他们得换种思维方式。
And they have to think differently.
是的。
Yeah.
没错。
Absolutely.
保罗，Impeccable 的设计对我来说就是一个绝佳范例，展现了那些懂得如何操控智能体和代码的人所具备的出色计算能力。
Paul, the design of Impeccable is example to me of great computational craft for people who understand how to use manipulations of agents and code.
所以这也是我特别欣赏你工作的地方——它非常具有计算思维。
So that's why another thing I admired about your work, it's very computational.
非常有意思。
Really interesting.
但你们觉得，提升上限是源于人的直觉问题，还是工具不足的问题，或者两者都有？
But do you guys is it an instinct problem then in terms of raising the ceiling, or is it a tooling problem, or is it both?
因为你说过，约翰，大脑的结构可能不太适合，但我们也确实存在工具上的缺口。
Because if you said you know, John, you said that the brain isn't quite shaped the right way, you know, but we also have a tooling gap.
那么，这其中有多少是靠直觉呢？
So how much of it is instinct?
又有多少是靠工具呢？
How much of it is tools?
我觉得这两方面都有影响。
I would say it's probably a combination of the two.
我认为，目前还是有很多工作只能靠人工完成，AI或者现有的工具还做不到，这一点我很肯定。
I think there there's definitely a lot you can do with manual labor and manual craft that you still can't do with AI today or with current tooling, I would say.
有些CSS代码，我到现在还是手动写，因为现在的智能代理还达不到要求。
I think there's there's some CSS that I still manually write because the agents aren't good enough at it yet.
但通常都是像动画这类东西。
But that's usually things like motion.
目前来说，动画还没有好的反馈机制。
So, you know, right now, there's no good feedback loop for motion.
我以前做过一个着色器库，有个着色器是经过和模型反复调整后，让雨水向上飘动的。
I've built the the shader library in the past, and I had this one shader that was animating rain upwards after working with the model for a bit.
这可不是模型的错。
And it's not the model's fault.
模型只是截个图，然后心想：嗯，这看着像下雨吧。
The model just takes screenshots, and it's like, well, I guess that looks like rain.
但它根本不懂时间分辨率。
But it doesn't understand temporal resolution.
现在Gemini倒是多做了一点。
Now Gemini does a little bit more than that.
它会从视频里每隔一秒抽一帧，但就算这样也还是看不全真相。
It samples frames on a video with sort of like one frame per second, but even that doesn't really give you the full picture.
所以目前还是非常受限。
So it's still really, really limited.
像运动设计这类领域，今天能做的依然非常有限，但我认为更大的问题其实不在于我们没法造工具来加速这些事。
So there are areas of design like motion that are still super limited to what we can do today, But I also think the bigger problem is that it's not that we can't build tools to accelerate those things.
我觉得我们完全做得到。
I think we totally can.
只是我们还没做到而已。
It's just that we haven't yet.
我觉得这个领域还有很多工具可以开发，专门帮助那些想突破极限、提升创作高度的手艺人。
I think there's a lot of tools that can be built, can be explored here in this space that really target craftsmen that that wanna go beyond, that really wanna raise the ceiling.
目前来看，能真正推动边界的专业工具还非常稀缺。
So I think right now, it's it's a pretty barren landscape for tools that really push the envelope.
我觉得Suno是个很好的例子，它最初是为初学者和喜欢即兴创作氛围音乐的人设计的，就是想一键生成一首小曲子。
I would say Suno is actually a great example of something that started for, you know, beginners and sort of vibe sound designers, right, that wanted to just one shot a quick song.
但现在，它已经大大扩展成了真正的音乐制作工具。
But now it expanded very much into real music production.
现在有很多制作人用它来做各种各样的创作。
There there are are producers using it for all sorts of things now.
他们围绕它建了一个完整的制作系统，现在正努力攻克最后那20%的难题，这挺有意思的。
They've built a whole studio around it, and they are really trying to push sort of the last 20% now, which is interesting.
所以他们确实经历了一次重大的转变。
So they they went through a big shift there.
所以我觉得这种情况会越来越普遍。
So I think that will be happening more and more.
不过目前来看，还是属于比较小众的。
I think right now, it's still quite minor.
真的很有趣。
Really interesting.
约翰，你怎么看直觉和工具之间的关系？
John, do you view instinct versus tooling?
正如保罗刚才说的，谢谢大家提这么多问题。
Well, as Paul was saying that and thanks for all these questions.
我在工作中其实很少去想这类宏观的问题。
I don't think about these meta questions a lot in my job.
首先，视觉模型的感知方式跟我们人类完全不同。
Well, first off, the vision models don't really see the way we see.
而且现在可用的训练数据还不够多，没法让视觉模型像设计师那样去‘看’。
And there's not enough training data to help a vision model think like a designer can see.
因为视觉思维的潜在空间维度太多了，尤其是创意型视觉思维，跟找车牌号这类功能性任务完全不一样。
Because there's so many dimensions in the latent space of a visual mind still, a creative visual mind versus looking for license plates and more functional things.
我觉得，能提升这个领域上限的人数不会太多，而且在大众市场上也很难有太大价值。
I think that also the number of people who will raise the ceiling will not be a high number, and it won't be that valuable in the mass market.
所以我认为，这种工作永远不可能被完全自动化。
So I think it'll never be automated.
这对那些想继续从事这类手工创作的人来说，反而是个好消息。
And that's good news for people who want to stay in the craft of this kind of work.
现在的情况是，我们只是让普通人更容易做出普通甚至不错的成果，这正是任务型自动化的本质。
Feel like right now we're just making it easier for more people to produce average and above average work, which is the lesson of task based automation.
但那些真正渴望创意永存的人，创意本身永远都会存在。
But those people who are dying for the creativity to last, it'll always remain there.
唯一的问题是，做这种高阶创作的人，他们的客户群体会变得更小。
The only problem is the customer base will be smaller for people who are in that.
这就像是，我们还懂胶印印刷吗？
It's like, do we know letterpress printing?
天啊，我太喜欢了。
Oh my gosh, I love it.
但没多少人愿意为此买单。
But not many people will pay for it.
这就是市场的运作机制。
So it's that market mechanics.
是的，这真的非常有意思。
Yeah, it's really, really interesting.
而且我和保罗聊过，约翰，我认为未来会有更多产品面向规模虽小但依然有吸引力的市场。
And one of the things I believe, Paul and I have spoken about this, John, is that we're going to have a lot more products with small but still compelling markets.
那些百万美元级市场规模的产品，十万美金级市场规模的产品，以前因为经济上不可行，现在不一样了。
The million dollar TAM products, the $100,000 TAM products, because they were economically infeasible before.
我们可以想象一个数字主街或数字拓荒者的世界，能支撑起一些规模不大但立场鲜明、做到最后那百分之十的产品。
So we can sort of imagine a world where it's more of a digital main street that can or a digital homesteading that can support a product that may have smaller scale but be more opinionated and be that last 10%.
如果让我补充一点，我反复感受到的是，人们愿意为这种定制化、更优质的产品支付更高价格，是因为他们在为人的信任和责任买单。
If I may add to that, the thing that I feel over and over is that the reason why someone's willing to pay that much more for that bespoke better thing is they're paying for human trust and accountability.
所以对于那些用户特别在意‘我能信任这个吗？’的品牌和产品来说。
So for brands and products where people demand that level of like, can I trust this?
这种人性化的触感，反而会比以往任何时候都更有价值。
That human ability, that human smell can become even more valuable than ever before.
是的。
Yeah.
太棒了。
Love that.
实际上，唐，现在正好是个好时机。
Actually, Don, this is a good moment too.
我想念一段话，来自《福布斯》2010年关于《This Is You》的报道：软件行业正准备重新拥抱它的工艺传统。
I want to read a quote for This Is You, Forbes 2010, which is, the software industry is poised to embrace its craft heritage.
到2020年，软件将回归手工作坊模式，由众多开发者打造定制化应用。
By 2020, software will return to a cottage industry with bespoke apps made by many.
我们会重新发现创作者的价值。
We will discover the value of authorship.
我的意思是，你几年前的预测可能有点早，但现在看来，我们正慢慢走向那个方向。
I mean, seems like we're you might have been a few years early in your prediction, but it seems like we're getting there now.
你是觉得世界会朝着这个方向发展吗？
I mean, is that how you see the world playing out?
会有越来越多的小型应用程序，
Many more small apps that are
更加独特、更有个性？
all sort of more unique and opinionated?
没错。
Absolutely.
我在这一点上，整整早了五年判断错了。
I was five years incorrect on that one.
我是在领导一所知名设计学院时，通过和非计算机背景的设计师、艺术家合作才明白的。
That's something I learned by working with non computer based designers and artists when I led renowned school design.
那是在我从麻省理工学院出来之后，那里全是工程思维，后来我又回到了手工创作的路子上。
That was after a career at MIT, where it was all engineering and then back to the handmade.
我深深感受到，那个社群里充满了人的情感和自豪感。
And I was struck by how much human emotion, human pride lives in that community.
我觉得这种情感迟早也会在软件行业里显现出来。
And I felt it would have to surface through in the software industry.
作为长期支持程序员为艺术家和设计师服务的人，无论是通过Processing还是Scratch，我都觉得这些工具对他们的门槛太高了。
But as a longtime proponent of computer programming for artists and designers, whether through processing or scratch, it was too hard for those people to engage.
但现在，进入这个领域的门槛已经大大降低了。
But now it's that that floor of getting in has been lowered.
现在人人都能参与进来了。
Everyone's coming in.
现在真是个特别好的时候。
It's a really great time right now.
但同时，这也让人特别困惑。
It's also very confusing.
所以保罗的做法很独特，他注重品质和深度思考，这给更多人提供了指引，让他们不至于选错方向。
So that's why Paul's Paul's approach is unique in that it's high quality and high thought based, which gives hope for more people to not choose the wrong solution.
保罗，你觉得像 Impeccable 这样的工具，有机会去塑造设计思维吗？
Paul, do you think there's an opportunity to start to shape the design mind through something like Impeccable?
因为它似乎触及了其他设计工具都没能覆盖的那部分解决方案。
Because it feels like it touches on a part of the solution that no other design tool does.
这是个很好的问题。
That's a great question.
我觉得肯定可以。
I think for sure it can.
而且它出人意料地，能同时影响设计思维和工程思维。
And I think it can shape ironically or interestingly both the design and the engineering mind.
有人曾主动来找我，说他们现在把 Impeccable 当成默认工具，安装在所有工程师和设计师的电脑上。
So I've had people walk up to me and say, well, we're now using Impeccable as, like, a a default thing that we install on all of our engineers and designers' computers.
我就问他们，为什么呢？
And I asked, so so why?
真有意思。
Interesting.
因为他们用了之后沟通顺畅多了，我觉得这一点特别有意思。
And it's because they now communicate much better after using it, and I thought that was really interesting.
工程师现在更多地使用设计术语，设计师也更常使用工程术语，因为这套工具通过严谨的词汇和方法，教会你哪些行得通、哪些不行，帮你找到准确表达需求和不需求的方式。
Like, engineers are using design language more, and designers use engineering language more because the the tools, by using this careful vocabulary and approach, actually teach you what works, what doesn't, give you the language to express what you want and what you don't want.
所以我觉得这里确实有个教学的契机，我长期和模型打交道，也学到了不少东西。
And so I do think there's there's a there's a teaching moment here, and I learned something by working with models all the time.
说实话，我一向都是这么提问的。
I've always prompted that way, to be honest.
我一直想让模型多给我点反馈，逼我思考，挑战我，甚至教会我新东西。
I always wanted to get more out of models by having it test me and challenge me and sort of, like, teach me new concepts.
比如马特那个文档技能，就挺厉害的。
And there's some cool skills out there like like Matt's grill with doc skill, for instance.
但Impeccable在某些方面也有类似的效果。
But Impeccable does that in some ways too.
所以我觉得这里面机会很大，真的让我很兴奋。
So I I do think there's a there's a lot of opportunity in that there, and that excites me.
这其实触及了一个我一直很在意的话题，最近我也参与了不少关于它的讨论，那就是认知上的放手与认知上的分工。
I think that really touches on a topic that is dear to my heart and was part of a lot of discussions that I was in recently, and that's cognitive surrender versus cognitive delegation.
我不知道你有没有听过这两个词，但举个例子，认知分工就是我用谷歌地图。
I don't know if you if you heard those terms, but really cognitive delegation is, you know, I'm using Google Maps.
对吧？
Right?
谷歌地图告诉我怎么走最快。
And Google Maps tells me where I can go as quickly as possible.
对吧？
Right?
我的意思是，我会自己想办法到达目的地。
I mean, I'm gonna root my way to something.
但如果我让谷歌地图替我决定该去哪儿呢？
But what if I let Google Maps decide where I want to go to?
这时候，我就相当于把决策权交给了这个应用，主动放手了。
Now I kinda like cognitively surrender to the application.
如果你不够小心，这一点在大语言模型上特别明显。
And I think that's very true for LLMs if you're not careful.
用大语言模型的时候，经常会出现这种情况，比如你启用计划模式，对吧？
Like with LLMs, oftentimes, LLMs, I mean, you, for instance, plan mode, right?
听起来挺不错的。
Sounds like a good idea.
你给它一个提示，它就能生成一份特别完美的计划，结果长达八页。
You prompt something, and then it creates a beautiful plan, and then it's eight pages long.
你根本不会通篇读完这份计划。
You're not gonna read through that plan.
对吧？
Right?
你最多就是翻一翻、扫几眼，心里想：嗯，模型应该知道自己在干嘛。
I mean, you're gonna scroll through that plan and skim it, and it's like, I guess the model knows what it's doing.
没错。
Right.
然后你就点一下，好了。
And then you just click, okay.
于是你就慢慢把主动权交给了这个流程。最近我参加的一个会议上，有位哈佛教授和其他几位专家也聊到这个问题，说现在学生的学习方式变了，这已经成了一个隐患。
And so now you kind of like surrendered yourself to the process, and there was a great conversation with a professor from Harvard and a bunch of others in a in a recent conference that I was at, where they touched on, you know, how students are now working differently and how that is becoming a problem.
所以总的来说，我觉得认知外包挺好的。
So anyway, I think I think cognitive delegation is great.
把任务交给模型来处理，确实很有帮助。
I think delegating to the model is great.
但关键是要保持自己的判断力，确保你始终是主导的一方，这一点同样非常重要。
But preserving that point of view, making sure that you're still the one driving is also really important.
我希望你和Impeccable的合作，能变成一种来回互动的默契配合，你们俩都是积极的合作者。
And I want you working with Impeccable to be the sort of game, that back and forth, where you're both active collaborators.
所以，保罗，我有个关于品味的问题想问你。
So so, Paul, question for you on taste.
现在大家讨论的都是自动化，比如目标设定、循环、多智能体协调这些事。
Right now, automating things away, goals and loops, multi agent orchestration, that's what everybody is talking about.
但没人问的是，这一切最终会走向哪里？
But the question that nobody seems to be asking is, where does it all lead?
我们是不是，如你所说，正在放弃人类的意图？
Are we sort of, you know, as you said, surrendering human intent?
在我们打造的这些产品中，人类的意图和品味究竟该扮演什么角色？
And what is the role of human intent and taste more broadly in these products we're creating?
确实如此。
Yeah.
我觉得这是个有意思的问题，因为现在大家都在讨论，像大语言模型和扩散模型这些技术，到底有没有品味。
I think it's an interesting question because right now, everybody's trying to solve or figure out whether models, whether LLMs have taste, right, as well as diffusion models.
但要说清楚是或不是，其实没那么简单。
And I think the answer is a little bit more complicated in saying yes or no.
我觉得这些模型其实拥有数百万种不同的品味定义。
I think these models have millions of definitions of taste.
它们接受的训练数据里，包含了太多对品味的不同理解。
They are trained on so many different interpretations of taste.
但有一件事非常值得了解和思考，那就是大语言模型训练用的是人类的产出，而不是输入。
But one thing that is very useful to know and think about is that LLMs have been trained on the output of humanity, not on the input.
所以模型并不知道某个设计决策背后的原因。
So what led to a design decision is not something that the LLMs know.
因此，与其说它们有品位，不如说它们只是模仿了某个时代特定受众的偏好。
So there's it's an approximation of taste, I would say, a certain type of audience at a certain time.
只要你清楚自己想要什么，就能得到不错的结果。
So you can get pretty far if you know what you're asking for.
但问题来了，这到底是谁的品位？
But then the question is whose taste is it?
很多时候，真正的品位是稀有的、属于人的。
Oftentimes, taste is rare and human.
我们跟人聊品位的时候，都知道那不是随便就能复制、批量复制的东西。
And, you know, I think the way we talk about taste when we talk to humans, it's not something that can be easily recreated or spread to lots of different instances.
所以我想问，模型真的能有品位吗？
And so I think, can the models be tasteful?
是的，它们确实能设计出有品位的东西。
Yeah, they can design tasteful things.
但我还是觉得，带有人类视角的作品能走得更远。
But I also still think that something that has a human viewpoint goes far beyond.
所以我认为，放大人类的品位是个更有意思的课题，比在模型层面复制品位要深刻得多。
And and and so I think amplifying human taste is an interesting problem, a more interesting problem than creating a replication of taste at the model layer.
对。
Yeah.
这确实很有趣。
It's really interesting.
如果你把‘品位’换成‘人性’这个词，意思反而更贴切，甚至更有力量。
It's if you sort of substitute the word taste for humanity, it sort of carries the same meaning perhaps even a powerful more powerful way.
约翰，你觉得品位是天生的，还是后天培养出来的？
John, do you think taste is innate, or is it something that's cultivated?
我认为品位始终是文化的产物，不同文化之间的品位高低，往往是因为它们历史更悠久，积累更深厚，因而更成熟。
I think taste is always cultural, and different cultures have higher taste, specifically because they've usually been around longer, they've had the material around longer, and therefore they've had maturity.
比如你去丹麦，随便坐一把椅子、一张沙发或者一张桌子，都会觉得这东西真讲究。
For instance, if you go to Denmark and you sit in a furniture, any type of furniture, like a chair or sofa or table, you're like, Wow, this is nice stuff.
但你要是逛美国，说实话，有时候真没那么出色。
When you walk around The US, it's like, I mean, I'm sorry, but not that great sometimes.
毕竟我们可是发明了泡沫塑料餐盘的人。
Because we invented the Styrofoam plate, for instance.
你去日本，那里几百年来都在不断打磨对原材料的运用方式。
You go to Japan, centuries of evolved approaches to raw materials.
不仅如此，原材料还很稀缺。
And not only that, scarcity of material.
东西越少，人们就越珍惜，也更用心去设计。
When raw material is scarce, we tend to make it more precious and design things well.
品味，是在稀缺中生长出来的。
Taste emerges through scarcity.
最后一点，我们得记住，欧洲那种设计观念，原本是来自王室，为了彰显独特性。
The last point on this is we have to remember that design in the European sense came from royalty and the desire to be distinctive.
我有皮草，有黄金，有翡翠，我很重要。
I've got fur, I've got gold, I've got emerald, I'm important.
你没那么重要，因为那时候材料稀缺。
You're less important because they were working with scarce materials.
我觉得这个时代的有趣之处在于，当所有人都能轻易获得各种材料时，品味这个概念就站不住脚了。
I think what's interesting about this era is that this idea of taste doesn't fit when all the materials are available to everyone.
这个观点说得真好。
That's a really great point.
确实如此。
Yeah.
这简直太引人深思了。
That's absolutely fascinating.
我来读一段你最近博客里的评论，乔纳。
Here's a comment from your recent blog, Jonah, that I think I'll read.
现实很残酷：在真正做决定的场合里，品味只是耳语，而速度却是扩音器。
So the hard truth is that in rooms where decisions actually happen, taste is a whisper and velocity is a megaphone.
面对紧迫的截止日期，很难为设计上的细微之处或直觉辩护。
It's almost impossible to defend a design nuance or instinct against a looming deadline.
问题不在工程团队，而在激励机制。
Engineering isn't the villain here, incentives are.
我们建立的组织，总是奖励那些可量化的成果，而不是真正有意义的东西。
We build organizations that reward the measurable over the meaningful.
你同意吗？
Agree?
你不同意吗？
Disagree?
在你所在的场合，这种情况是怎么体现的？
How do you see that show up in the rooms that you're in?
说到底，关键还是在领导。
I mean, it's all about the leader.
对吧？
Right?
你提到史蒂夫·乔布斯，他把设计、商业和技术感完美地结合在了一起。
You mentioned Steve Jobs, a unique combination of design, business, technical sense.
这种人真的非常少见。
That's very rare.
我想再回头聊聊保罗的作品，这次可别再让他不好意思了。
And I want to go back to Paul's stuff and not to embarrass him again.
Impeccable 这个产品有意思的地方在于，它的 API 设计特别出色。
So what's interesting about Impeccable is the design is very interesting from the API perspective.
已故的比尔·阿特金森，那位参与过 Macintosh 开发的工程师，可能就熟悉一个叫 QuickDraw 的东西，那是苹果的图形库。
And Bill Atkinson, the engineer who recently passed away, who worked on the Macintosh, Specifically, may know this thing called QuickDraw, which was Apple's graphics library.
Photoshop 的出现，正是因为 QuickDraw 的 API 设计得够好。
Photoshop happened because of the design of QuickDraw's API.
Windows 版本的 Photoshop 之所以不可能出现，是因为 DirectX 没有类似的 API 结构。
It was impossible for there to be a Windows version of Photoshop because it didn't have the same API shape in DirectX.
所以我想说，保罗，那些懂设计的工程师，特别擅长设计 API 的形态，让它们既能和人顺畅协作，也能和智能体无缝配合。
So I would argue that, Paul, design engineers are able to design the shape of the APIs really well so they can work with agents as well as they can with humans.
这简直太让人着迷了。
It's completely fascinating.
我完全不知道还有这回事。
I had no idea.
API 里是什么关键设计让 Photoshop 成为了可能？
What was the unlock in the API surface that made Photoshop possible?
天哪。
Oh my gosh.
Macintosh 的图片系统用了一种叫‘区域’的概念。
So the Macintosh Picture had this routine built upon this idea of regions.
其他系统都用多边形。
Everyone had polygons.
但 Mac 的区域是基于像素形状的，能做逻辑上的与或运算。
The Mac had regions based on pixel shapes, which you could do logical set or and.
所以像油漆桶填充这样的功能，速度优化得特别极致。
So things like flood fill were super duper optimized for speed.
所以再说一遍，好的 API 不只是结构好看，更要性能出色。
So again, great APIs are not just shape, but performance.
太有意思了。
Super interesting.
这真的特别令人着迷。
That's absolutely fascinating.
保罗，你觉得对于在场的设计工程师们，你过去是怎么有效传达直觉的价值的？比如，怎么跟截止日期竞争，或者跟其他不太有根据的直觉去说服人？
Paul, how do you think, you know, maybe advice for our design engineers in the room, how have you in the past effectively kind of communicated the value of an instinct versus, you know, a deadline or even, you know, another instinct which may be less informed?
是的。
Yeah.
这确实很难，因为我在这上面失败过无数次。
That's a tough one because, certainly, I've failed many, many times on this one.
我慢慢总结出一些方法，但很多时候，关键是要让决策者进入一种能做出我所期待决定的心态。
And I've developed on strategies, but oftentimes, it's about getting a leader into the right mindset to make a decision that I would make.
对吧？
Right?
我的意思是，约翰说得对。
What I mean by that is and John is right.
如果你的领导主要关注经济成果，比如那一周他特别看重业绩，或者他特别关注性能，那我觉得你的责任就是用一种方式让他看到你所憧憬的未来。
I mean, if you if you have a lead who's mostly work at who's mostly focused on economic outcomes, for example, during that week, right, or mostly focused on performance, ultimately, I think it's your job to present to them in a way that makes them dream the way you're dreaming about the future.
这一点我是反复碰壁、一次次才学会的。
And and that's something that I learned the hard way over and over and over.
你得带着他一起走，让他也能踏上你的情感旅程，或许吧。
You have to bring them along to where you want to go and kind of get them on the same emotional journey, maybe.
这样表达才对。
That's the right way to frame it.
只要他们看到了你看到的东西，看到了你想要创造的未来，你们就能达成共识。
And I think once they see what you see, once they see the type of future that you want to create, then you're on the same page.
说起来容易，做起来难多了。
Now that's way easier said than done.
我觉得这是一门艺术：先从他们关心的事切入，然后不知不觉把对话引向你们共同期待的未来。
It's, I think, an art form to start with something that they need and then suddenly shift in that conversation towards something that you both might want in the future.
我知道这听起来可能有点抽象。
I know that's maybe a bit abstract.
但是
But
说到这个，我上周吃饭的时候遇到一件事。
So on that topic, I had a I was having dinner, like, last week.
你认识杰·帕里克吧？我以前在Meta的老板。
You know Jay Parikh, my boss, formerly of the Meta.
他用了一个我好久没听过的词，叫‘信念’。
He's used this word that I hadn't heard in a while, which is conviction.
我觉得这个词特别贴切，因为一个有品位的领导者，一定是有信念的。
And I think it's such a great word because someone, a leader with quote unquote taste, has conviction.
但这不只是设计上的信念。
But it's not just designee conviction.
也不只是商业上的信念。
It's not just business y conviction.
也不是单靠工程上的坚持。
It's not just engineering conviction.
真正厉害的领导者，是那种既有信念又懂得坚持的人。
It's that combination in a leader when they have conviction.
这是一场押注，但押的是全局最优解，而不是局部最优解。
It is a bet, but it's a bet aimed at the global maximum, not the local maximum.
所以，那些坚持追求全局最优解的领导者，往往能带来巨大的改变。
And so I would say leaders that are doing that global maximum conviction make a huge difference.
正因如此，说到产品负责人——虽然这么说有点不好意思，但我觉得保罗在设计自动化上追求的就是全局最优解，而且他到现在都没放弃。
And that's why product leaders, again, to embarrass Paul, I think Paul shot for a global maximum for design automation, and he hasn't given up yet.
这真的很了不起。
And that's a beautiful thing.
我特别认同。
I love it.
本总说，CEO和领导者的责任不是自己想出点子，而是当好点子出现时，能认出来，并且带领团队全力投入。
Ben always says that the job of the CEO, the job of the leader is not to have the idea, but to recognize the idea when they see it and bet the team on it.
还有一件关于史蒂夫·乔布斯的轶事，你知道的，他特别喜欢打断别人，就是为了考验他们。
There's there's also, you know, another Steve Jobs sort of anecdote, right, where he's very well known or he was very well known for shutting down people just to test them.
对吧？
Right?
有意思的是，这种坚定不仅来自领导者，也来自员工。
So the conviction is both on the leader and on the employee, interestingly.
很多时候，有人走进房间找史蒂夫，他就会直接说：‘我觉得这主意太蠢了。’
Because very often, somebody would walk into the room with Steve, and Steve would just test him and say, like, you know, I think that's a stupid idea.
其实他心里可能觉得这主意特别棒。
Even though he might think that's a really great idea.
是的。
Yeah.
但他就是想看看，对方会不会被吓退，转身就走？
But just to test him, right, and to see like, no, dude, will they just cave and walk out of the room?
还是会坚持自己的观点，加倍努力？
Or will they actually double down?
他们会说，不。
Will they like, no.
不。
No.
不对，史蒂夫。
No, Steve.
我觉得你错了。
I think you're wrong.
对吧？
Right?
一旦他感受到这种坚定的态度，他往往会说，好。
And once he got that signal, once he got that signal of conviction, he often would say, yes.
就这么干。
Let's do it.
我觉得这特别有意思。
And I think that's fascinating.
我觉得约翰说的特别有意思。
So I think what John said is really interesting.
我觉得这两方面是相互的。
I think it goes both ways.
天哪。
Oh my gosh.
这简直是大语言模型当裁判的完美例子。
That's a perfect example of LLM as judge.
不管是生成信念还是反馈信念，两边我们都得有。
We have to, on both sides, on a generation of conviction and the feedback of conviction, need both.
就像反物质一样。
The anti matter matter.
太棒了。
Love that.
我也觉得太棒了。
I love that.
咱们一年后再录一期，看看用新Git产品配合Impeccable开发出来的软件，是不是真能更用心，达到前10%的水平，那可就太有意思了。
Well, we'll record this pod again in a year, and we'll see if all of the software that's been created through the new Git product with assistance by Impeccable turns out to be more thoughtful and in that upper 10%, and that would be a fun outcome to see.
这挑战我接了。
Challenge accepted.
太厉害了。
Incredible.
约翰，你有什么想最后说的吗？
John, any closing thoughts for us?
接手GitHub Design之后，我才意识到，原来有这么多人都是在GitHub Design成长起来的，他们都是计算设计师或者工程师。
After picking up GitHub Design, I realized how so many people grew up at GitHub Design who were computational designersengineers.
比如Notion的Max，他以前就是GitHub的设计负责人。
There's Max at Notion who used to lead design at GitHub.
还有Vercel那帮人。
There's all the Vercel people.
他们到处都是。
They're all over the place.
有人让我想办法把这个系列重新带回来。
And I was asked to somehow bring back the franchise.
所以我特别兴奋，因为保罗的工作体现了一种根植于设计历史的深刻方法，能让更多工程师更可靠地做设计，也让设计师有底气去追求更高的目标。
And so I'm excited because Paul's work represents that kind of seminal design history infused approach that can really make it easier for more engineers to design reliably and also more designers to be able to reach for the skies.
太棒了。
I love it.
保罗，有什么想说的结尾话吗？
Paul, any closing words?
我特别期待，等我们提升了基础之后，会创造出怎样疯狂又惊艳的体验和设计。
Well, I'm just super, super excited to see what will happen once we've raised the floor and what insane, crazy experiences we'll build afterwards and design afterwards.
我想要的未来是这样的，而不是那种千篇一律、让人觉得已经放弃一切的黯淡前景。
That's the kind of future that I want as opposed to, you know, the bleak future where everything looks same, and and people feel like they've surrendered themselves.
所以我对这个全新的工艺时代充满期待，这也是我正在努力的方向。
And so I really am excited about that new era of craft, and that's something I'm working towards.
非同凡响。
Extraordinary.
为这个工艺的新时代干杯。
Well, cheers to a new era of craft.
感谢你们两位的出色工作。
Thanks to the good work from both of you.
谢谢你们两位今天到场。
Thank you guys both for being here.
很期待看到产品上线，亲自体验一下。
Excited to see the product out in the wild and play with it ourselves.
非常感谢你们邀请我。
Thanks so much for having
感谢您收听这期的 a16z 播客。
for listening to this episode of the a 16 z podcast.
如果喜欢这期节目，别忘了点赞、评论、订阅，给我们打分或写评价，也分享给你的朋友和家人。
If you like this episode, be sure to like, comment, subscribe, leave us a rating or a review, and share it with your friends and family.
想收听更多节目，请前往 YouTube、Apple Podcast 和 Spotify。
For more episodes, go to YouTube, Apple Podcast, and Spotify.
关注我们在 X 平台的账号 a sixteen z，并订阅我们的 Substack 频道 a16z.substack.com。
Follow us on x at a sixteen z, and subscribe to our Substack at a16z.substack.com.
再次感谢收听，我们下一期再见。
Thanks again for listening, and I'll see you in the next episode.
提醒一下，本内容仅作信息参考，不构成法律、商业、税务或投资建议，也不应用于评估任何投资或证券，且并非针对任何 a sixteen z 基金的投资者或潜在投资者。
As a reminder, the content here is for informational purposes only, should not be taken as legal business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any a sixteen z fund.
请注意，a sixteen z 及其关联方可能仍持有本节目所讨论公司的投资。
Please note that a sixteen z and its affiliates may also maintain investments in the companies discussed in this podcast.
如需更多详情，包括我们的投资组合链接，请访问 a16z.com/disclosures。
For more details, including a link to our investments, please see a 16z.com forward /disclosures.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。