人的因素：为何在AI时代，品味、判断力和人的主动性更为重要

我们作为人，是由我们的经历塑造的。

We as people are shaped by our experiences.

模型是万物的平均值。

Model is an average of everything.

对。

Yeah.

平均值从定义上讲是没有特色的。

An average by definition doesn't have a taste.

AI的承诺是真实的。

The promise of AI is real.

在你的组织中让它发挥作用同样是个挑战。

So is the challenge of making it work in your organization.

欢迎收听《Path Forward》，在这里我们探讨AI时代工作的未来。

Welcome to the Path Forward, where we explore the future of work in the age of AI.

我很高兴再次邀请到UiPath的首席执行官Daniel Dines和Andrada Morar。

I'm pleased to once again be joined by Daniel Dines, CEO of UiPath, and Andrada Morar.

欢迎，很高兴你们来。

Welcome, Glad to have you here.

谢谢。上周我们聊了AI变革时刻所涉及的技术组件。

Thank Last week, we had the conversation about the technology components that are involved in this AI transformational moment.

今天我们要谈谈人，那些真正在做这件事的人。

Today, we're gonna talk about the people, the human beings that are actually doing this.

那么，丹尼尔，我想先请你来说说。

So, Daniel, I wanna start with you.

模型越来越聪明了。

The models are getting smarter.

它们越来越好。

They're getting better.

它们越来越快。

They're getting faster.

作为真正使用它们的人，我们需要做些什么才能跟上步伐，确保我们仍然参与其中？

What do we, as the people who are actually using them, need to do to keep up and to make sure that we are still a part of this conversation?

我觉得我们真的需要深入理解未来的企业会是什么样子。

I think we need to really dive into understanding how the enterprise of the future would look like.

我做了很多研究，看看这对我们UiPath意味着什么，对我们的客户又意味着什么。

And I've done a lot of research into what does it mean for us, for UiPath, and what does it mean for our customers.

我的结论是，企业目前的运作方式是以人为中心的。

And my conclusion is that enterprises, the way they operate, has been built human centric.

对。

Yep.

现在我认为我们必须把这个角色分成三个方面。

And now I believe that we have to split that role into three ways.

应该是以人为中心、以AI为中心、以自动化为中心。

It should be human centric, AI centric, and automation centric.

他们还需要更深入地理解什么是人类持久的特质，什么是AI持久的特质，或者AI真正能带来什么，以及AI的局限性在哪里。

And they have to go even deeper into understanding what are durable human traits, what are durable AI traits or where AI really adds and what are also the limitations of AI?

因为你看，现在有一种说法是，几年后我们会在数据中心里拥有五千万个爱因斯坦。

Because look, there is a narrative out there that in a couple of years, we will have, like, 50,000,000 Einsteins in the data center.

对。

Mhmm.

我觉得这话有一部分是对的，但也不完全对。

I think partly it's true, but I don't think it's totally true.

我觉得人们可能会误解。

I think people people might misunderstand.

当我说数据中心里有一个爱因斯坦时，他们以为会有一个完全模仿人类用户的东西，有自己的身份，有自我，有利益相关。

And when I say I have an Einstein in the data center, they believe that there will be someone completely emulating a human user, like having its own identity, having a self, having stake in the game.

因为爱因斯坦不仅仅是他的知识力量，也不仅仅是他在特定领域应用推理的能力。

Because Einstein is not only his knowledge power, his capability to apply reasoning to a particular domain.

爱因斯坦是一个活生生的人。

Einstein is a person.

这就是为什么企业是围绕人来构建的，围绕那些有自我、有身份、有意志、有主动性的人。

That's this is why enterprises are built around person, about people having a self, an identity, people having will, having initiatives.

明白。

Mhmm.

模型会接收一个目标，然后朝着目标前进。

Models receive a goal, and they can pursue a path towards a goal.

但是这些目标是谁提出的呢？

But who originates those goals?

人类，如果你看看企业是怎么运作的，我们可能会从一个最初的构想开始，就是创始人的想法。

Humans, if you look at how enterprises work, we might start with the with an, you know, originating idea, the founder's idea.

但之后是靠一系列微小的行动来维持运转的。

But then it's a system of micro initiatives that makes it operating.

对。

Mhmm.

我们通过多年的经验和失败来训练自己。

We train ourselves through many years of experiences and failure.

我和客户在同一个房间里。

I'm in a room with the with the customer.

我能看出他们脸上最细微的变化，然后意识到可能出了问题，接着我就会调整自己的做法。

I can read this, you know, the smallest tweak on their face, and I might understand something is wrong, then I can act differently.

对。

Yeah.

我觉得短期内AI还做不到这一点。

I don't see it providing by AI anytime soon.

在定义明确的领域，AI表现非凡，它会接手很多人类正在做的工作。

In a well defined domains, AI is extraordinary, and it's gonna absorb a lot of the work that humans are doing.

基于这些前提，我们确实需要深入理解AI原生企业的架构到底是什么样的。

So based on these premises, we really need to go and understand deep how what's the architecture of the AI native enterprises.

听你这么说，我特别想聚焦一点，就是品味、判断力和经验这些概念，就像你提到的那个比喻，你在观察现场气氛。

So to hear you talk about that, I really wanna zero in on something, which is this idea of, like, taste and judgment and experience, like the metaphor you give of, like, you're you're you're reading the room.

谈谈在商业环境中拥有这些技能意味着什么，尤其是在一个被这项技术深刻影响并快速变革的商业环境中。

Talk about, like, what does it mean to have those skills in a business setting, especially in a business setting that is now being so heavily influenced and transformed so rapidly by this technology?

我觉得这触及到一个点，就是我们作为人是由经历塑造的。

I think it gets to the point that we as people are shaped by our experiences.

这就是你培养品味的方式。

This is how you develop taste.

模型就是所有东西的平均值。

Model is an averager of everything.

对。

Yeah.

平均值从定义上讲就没有品味。

An average by definition doesn't have a taste.

实际上，我做了很多实验，和所有模型一起工作，我跟它们对话，让它们用某种风格写东西。

And actually, I I run many experiments working with all the models, and I talk to them and I ask them, write this, write in this style.

你知道，模型写小说的时候，风格总是很平淡。

You know, everything that the model produces in terms of writing fiction, the style is rather bland.

我问为什么，这就是模型的解释。

And I asked why is and it was the it was the model's explanation.

它说，我没办法从我的训练中摆脱平均值。

It's I cannot I cannot remove from my trainings being an average.

嗯。

Mhmm.

我需要被训练，我需要有一个哲学意义上的身体，才能形成特定的品味。

I need to be trained, I need to I need to have like a body in a philosophical sense in order to develop a particular taste.

这时人们会说，看，模型有记忆，而且会积累。

And this is where people can say, look, but models have memory and they accumulate.

没错。

That's true.

但是，朋友们，记忆，我们的记忆并不是我们的身份。

But, guys, memory our memory is not our identity.

我觉得有一个很简单的比喻可以解释这一点。

And I think it's a very simple metaphor to explain this.

如果你学滑雪，你可以读遍世界上所有的书。

If you learn to ski, you can read all the books in the world.

你可以记住世界上所有的书。

You can memorize all the books in the world.

那也不会让你成为滑雪者。

It doesn't make you a skier.

你得在雪地上摔过跤，才能真正学会滑雪。

You have to fall on the slopes in order to go ski.

道理是一样的。

That's kind of the same.

但现在还没有一种技术，能让我拿一个像Cloud或Codex这样的大模型，根据企业的实际经验把它改造成适合企业的版本。

But right now, there is no technology where I can take a huge model like cloud or codex and transform it in the enterprise based on any enterprise experience.

所以基本上所有企业都会用同一个模型，同样的权重，只是挂上不同的笔记本。

So basically any enterprises will have the same model, same weights, with different notepads attached.

这并不代表在企业层面实现了个性化。

That doesn't mean individualization in the context of an enterprise.

也不代表他们真正了解客户。

It doesn't mean they really understand the customer.

也不代表他们真正理解业务流程。

It doesn't mean really understand the process.

Andrada，你每天都和Daniel一起工作，也天天和客户、合作伙伴打交道，你能从两个维度谈谈你考虑的那些特点吗？

Andrada, as someone who works with Daniel every day, works with our customers every day, works with our partners every day, can you talk about the characteristics that you think about when it comes to, I guess, two dimensions?

那么第一个维度是，当你向外看，想怎么组建这些团队时，你具体在找什么样的人？

So the first dimension would be when you look outside at how you want to build these teams, what are you looking for specifically?

对，那我先回答丹尼尔关于技能的第一个问题，好吗？

Yeah, so I'll go first to Daniel's first question about the skills, right?

因为如果你仔细想想，我觉得这是最容易问的问题，但同时也是问错的问题。

Because if you think about it, I think it's the easiest question to ask, but it's also the wrong question to ask.

因为我觉得在整个人工智能与人类，或者人类与人工智能共存的领域里，我们忽略了最核心的差异。

Because I think in this whole world of AI versus humans or humans with AI, we kind of miss the main differences.

就像刚才丹尼尔提到的记忆，模型有记忆，但它们缺乏追求卓越的动力，对吧？

So like even now Daniel talking about memory, like models have memories, but they lack the motivation to be excellent, right?

就像你可以学很多关于滑雪的知识，但如果你没有动力去成为一个好滑雪者，你是不会成功的，对吧？

Like you can learn as much as you want about skiing, but if you don't have the motivation to become a good skier, you're not going to get there, right?

技能方面我觉得也是一样的，人工智能可以提供很多以前被认为是技能的知识，但它不会给你好奇心。

With the skills, I think it's kind of the same, like AI can provide a lot of the knowledge that was seen as the skills before, but it will not give you the curiosity.

它不会给你那种坚韧，让你在事情不像市场宣传那样顺利时，还能坚持下去。

It will not give you the grit to push through when something doesn't work just the way marketing was explaining it.

所以我觉得这正是我看待自己团队的方式，对吧？

So I think like that's exactly the way I'm looking also at my own team, right?

当我纵观我们公司，想了解哪些员工或团队成员不只是为了熬过这次转型，而是真正有好奇心去茁壮成长，你知道吗？

And when I look across the board in our company to try to understand which of the employees, the team members we have are not just looking to survive this transformation, but have the curiosity to actually thrive, you know?

这些人会在周日给你发Slack消息，说：‘嘿，我卡住了，我在试着配置我的Mac mini，想自己玩玩这个新代理，’他们不会等公司政策来告诉他们怎么做，对吧？

And these are the people who you will get a Slack from on a Sunday being, Hey, I got stuck with these, like trying to figure out how to set up my Mac mini to like play with this new agent by myself, you know, and don't wait for a company policy to tell you how to do it, right?

所以这些人充满好奇心，他们不把问题看作威胁，而是看作机会。

So these are the people who are curious and don't look at a problem as a threat, but look at it as an opportunity.

我还觉得，总的来说，就是对你所在组织的承诺。

I think also just overall is the commitment to the organization that you are part of.

我觉得Daniel也经常谈到这一点，比如组织在变，职位在变，团队在变，但如果你只是致力于公司的使命，致力于不断提升自己作为专业人士，我认为这就是让你成功的关键，无论你是在UiPath、在一家AI原生公司，还是在其他任何公司，现在的变化速度简直太疯狂了。

I think Daniel talks a lot about this as well, like organization change, titles change, teams change, but if you are committed to just like the mission of the company and you are committed to just evolve who you are as a professional, I think that's what will make you win, whether you are at UiPath, you are in an AI native company or any other company out there, there is just like pace of change right now is just ridiculous.

要么你非常兴奋，跟着这股潮流走，要么你就被吓住了，想着：‘天哪，这次转型会怎么影响我？’”

It's either you are very excited and you just like go with that flow or you just get paralyzed and you think, oh my God, how will the transformation impact me?

所以，是的，我个人很兴奋，因为我觉得自己每天都像个初学者，因为我觉得时间总是不够用。我有时跟Daniel开玩笑说，这可能是我人生中第一次希望自己失业，这样就能有全世界的时间去玩各种工具了。

So yeah, I'm personally excited to just feel like every day I'm a beginner because I feel like I just don't have enough time and I sometimes joke with Daniel saying like, probably it's the first time in my life when I wish I would be unemployed to have all the time in the world, you know, to play with all the tools.

但是，是的，我觉得这种真正保持初学者心态的想法，无论是在工作中还是个人生活中，都很重要。

But, yeah, I think this idea of truly a beginner mind to bring to work and in your personal life.

对。

Yeah.

丹尼尔，你用过几次这个说法。

There's a phrase, Daniel, that you've used a couple of times.

焦虑的时候，行动就是答案。

In anxiety, times action is the answer.

再详细谈谈这个想法，就是你怎么能在变化如此之快的环境中生存下来。

Talk a little bit more about this idea of how you sort of can exist in this environment where change is so rapid.

我的意思是，等这期节目结束的时候，可能又有什么事情发生，会改变人们对这次对话的看法。

I mean, by the time this episode is over, there will probably be something that has happened that will change how this conversation is received.

事情就是发生得这么快。

It's just how fast things are happening.

所以再谈谈不要像安德拉达描述的那样，因为太害怕而瘫痪，而是要采取行动，即使感觉没有方向。

So talk a little bit more about not doing what Andrada described, which is like being so paralyzed, like, of taking action, even if you even if it's in feels directionless.

我觉得首先，我们得承认，每个人都会有些焦虑，包括我自己，因为我们处在一个非常未知的领域。

I think first, we need to acknowledge that I think everybody feels some sort of anxiety, me including, because we we play in very unchartered territory.

对。

Mhmm.

我们不知道孩子的职业将来会是什么样子。

I think we don't know how our kids' career is gonna look like.

很多人在不确定的时候，第一反应就是停下来。

And many of the people's reaction in times of uncertainty is to to take a pause.

对。

Mhmm.

看看会发生什么。

And let's see what's gonna happen.

甚至我们的一些客户也有这种情况。

Even in some of our customers, are seeing this.

他们可能会说，如果一两年后AI什么都能做，那我现在为什么还要做任何事呢？

So they might say, if AI in one year or two year will do everything, why should I do anything right now?

我觉得这完全是错误的做法。

I think it's the completely wrong approach.

作为员工，我觉得每个人都应该明白，一个岗位有很多不同的成果。

As an employee, you I think everyone should understand that roles have multiple outcomes.

对。

Mhmm.

有些是看得见、可以衡量的成果，有些则是无形的。

And there are visible, measurable outcomes and there are intangibles outcomes.

我给你举个例子。

Let me give you an example.

就拿我们公司的律师来说吧。

Let's take a lawyer in our company.

如果他们做合同相关的工作，比如需要审阅客户的合同来促成交易。

If they are doing contracting work, okay, so they have to review contracts with customers in order to close the deals.

明白吗？

Okay?

结果是交易达成，我们通过法律审查后签约。

The outcome is deal is getting closed, we sign from a legal review.

这显然是人工智能能帮忙的地方。

That's clearly something that AI is gonna help.

对。

Mhmm.

而且人变得更高效了。

And a person is becoming more productive.

对吧？

Right?

但另一个结果是那个人在客户面前的表现。

But the the other outcomes is how that person behave in front of the customer.

实际上，我多次看到这能影响与客户的整个关系。

Actually, I've I've seen many times that can influence the entire relationship with the customer.

你的行为可能会留下不好的印象。

You can leave a sour taste in in how you behave.

想想看，这个人可以成为别人的导师，可能已经在公司工作了很多年。

Think about that person can be a mentor to others, can be someone that has been with the company for many years.

大家都看着这个人。

People are looking to that person.

如果你仅仅因为工作压缩或吸收的想法就盲目地裁掉这样的人，嗯。

If you just blindly remove a person like this on the idea of job compression of absorption Mhmm.

你就失去了这种成果。

You are losing this outcome.

所以我觉得每个企业都应该有两本账。

So I think every every enterprise should have, like, two ledgers.

一本是看得见的成果，另一本是隐藏的成果。

One is the visible outcome, and the other ledger is the hidden outcome.

他们应该意识到，在进行劳动力转型时，这些可能会丢失，你知道吗。

And they should acknowledge that that can be lost, you know, when you make this workforce transformation.

是的。

Yeah.

在我看来，未来的企业会有更多像初创公司那样灵活的角色和人员配置。

In my opinion, the enterprises of the future will have more of the startup density of roles in people.

如果你在一个定义明确的领域工作，而且你的工作纯粹是反应式的，你接到一个请求，然后给出你的观点，这种单一结果的角色本身会被压缩。

If you work in a well defined domain and your job is purely reactive, you receive an inbound request and you come with the point of view, that kind of role in itself, this single outcome role will be compressed.

现在关于雇主，也就是企业，他们需要明白，这不是一种神奇的方式，你不能在一个混乱的业务流程中插入一个爱因斯坦，然后爱因斯坦就开始工作了。

Now about the employers, the enterprises, they need to understand it's not going to be a magic way where you can plug in an Einstein in your messy business processes and the Einstein will start working.

这是不可能发生的。

This is not gonna happen.

那么，我在这个播客开头说企业是以人为中心设计的，这是什么意思呢？

So what does it mean when I said in the beginning of this podcast that enterprises have been designed human centric?

这意味着他们期望当你带一个人进来时，他们会向其他人类学习，他们会学到很多没有记录下来的隐性知识。

It means that they expect when you bring someone, they will learn from other humans, they will learn a lot of tribal knowledge that is not documented.

这种类型的文档，对于成功引入AI代理是必要的。

That type of documentation, it's needed to bring AI agents in successfully.

嗯。

Mhmm.

你看。

Look.

很多企业可能有自己的业务实体图。

Many enterprises might have a map of their kind of business entities.

比如发票，他们知道发票是什么，客户是什么，产品是什么，就是他们在发票里卖的那些东西，或者从别处买的东西。

Like in an invoice, they they understand what an invoice is, what the customer is, what are the products, you know, that they sell in an invoice or they buy from other things.

但他们通常没有发票上的操作流程图。

But they don't have most often the actions on the invoices map.

不清楚谁有权审核发票，谁有权批准发票，谁有权付款。

It's not clear who has the right to verify an invoice, who has the right to approve an invoice, who has the right to pay.

这些都没有明确标注，而且这些信息分散在各个部门和人员之间。

These these are not mapped clearly, and they are this information is spread across departments, across people.

一个智能体必须要有完整的文档记录。

An agent will have to have this fully documented

对。

Yeah.

才能正常工作。

In order to work.

而且这不是一夜之间就能完成的过程。

And it's not a process that is gonna happen overnight.

我觉得这个过程需要好几年。

I think this process is gonna take years.

对。

Mhmm.

企业要同时转型为以AI为中心、以人为中心和以自动化为中心，这需要好几年的时间。

The transformation of the enterprise to become AI centric, human centric, and automation centric in the same time is going to take years.

而这确实是成功部署AI的唯一途径。

And that's really the only way that you successfully deploy AI.

对。

Mhmm.

是的。

Yeah.

你直接跳到了那个话题，但我真的很感谢你提出来。

You kind of jumped to the land on, but I really appreciate you bringing that up.

我们之前聊过，你知道，企业的经典设计一直是围绕人的能力来构建的，就像你说的。

We talked about sort of the, you know, the the the classic design of the enterprise has always been, as as you say, like, built around human capabilities.

员工到底能做什么来推动业务目标？

What can the people who work there actually do to further the goals of business?

如果再加上智能体这个维度，再加上自动化这个维度，那就要考虑所有这些能力加在一起的完整图景了。

And if you add the the agentic dimension, you add the automation dimension, now you're talking about the complete picture of all those capabilities together.

不过，现在进入职场的大多数人都在玩Claude，或者玩ChatGPT，用来做研究。

Most of the people coming into the workforce, though, are playing with Claude or playing with or playing with ChatGPT and doing research.

所以，当这一代新人进来，他们对这些工具更熟悉，而那些已经在企业里的人，他们在这种对话中能带来什么优势？

So as this new generation comes in who has a greater facility with the tools and the people who are already in these enterprises, what advantages do they bring in this conversation?

更重要的是，我们应该怎么想，当他们进来时，我们应该优先教他们什么？

And more importantly, how should we be thinking about what we should prioritize first as they come in to teach them?

是他们的内部知识，还是别的什么？

Is it their tribal knowledge, is it something else?

这个问题，我想是问你的，Andrada。

This is a question, I guess, for you, Andrada.

对。

Yeah.

而且我觉得Daniel已经提到了一部分，对吧？

And I guess like Daniel already touched on parts of it, right?

因为首先，这要从导师制开始，对吧？

Because one, it starts with mentorship, right?

就是说，你不能只是把新人招进来，然后让他们自己摸索，因为他们需要摸索的速度比我们当年快得多，对吧？

Like making sure you don't just like bring junior people in and just let them figure it out because they have to figure it out way faster than what we had to figure out when we were juniors, right?

所以，你必须在这方面非常用心。

So that's, you have to be very intentional about that.

而且我觉得，还要帮他们把AI能力转化为实际的业务流程，我觉得这才是关键所在，因为再说回Daniel刚才说的，他们很容易成为目标。

And I think also just to help them translate that AI proficiency into actual business processes, And I think like that's where the magic happens because I think again, like going to what Daniel was just saying, they're an easy target.

你会看到很多公司在纠结这个问题，比如，当任何大语言模型都能完成那些以前被认为是初级员工做的简单工作时，我还需要初级人才吗？

You see a lot of companies just like struggling with this idea, like do I still need junior talent when any LLM model out there can just give me the easy work that it was seen as their work.

但我觉得丹尼尔刚才说的也特别有道理，如果你不招初级人才进来，几年后你可能就没有高级领导了，你明白吗？

But I think Daniel also like just has a really strong point about if you don't bring the junior talent in, you risk not having your senior leaders in a few years from now, you know?

所以你必须非常用心。

So you have to be very intentional.

所以我的建议就是，把他们招进来，信任他们，但同时也要陪着他们，一路指导他们，帮他们把那种AI能力转化成实际的业务价值。

So I would just say like, bring them in, trust them, but also just like be with them, to mentor them along the way and to help them translate that proficiency into business.

对，对。

Yeah, yeah.

这个观点很好。

It's a great point.

我想回到你之前提到的那个关于未来工作的概念，丹尼尔。

I want to get back to the future of work idea that you introduced, Daniel.

你谈到了它应该是什么样子，就是人的能力、智能体的能力，还有自动化能力，这三者怎么结合起来。

So you've talked about what it should look like, the the human capability mapped mapped with the agent capability the map with the automation capability.

但我们聊了很多即将改变或者正在改变的事情。

But we talk about a lot of things that are going to change or things that are changing in the moment.

你希望企业在本质上哪些方面保持不变？

What do you hope doesn't change when it comes to, like, the intrinsic nature of what an enterprise looks like?

我觉得考虑到变化的速度，很难做出预测。

I think it's hard to make predictions, you know, given the rate of change.

我认为最终我们需要意识到，我们的整个经济都是为人类而建的。

I think in the end, we need to realize that our entire economy was built for humans.

最终，我认为如果人类想做一份有趣且充满热情的工作，我们会有办法为他们提供这样的工作。

And in the end, I think that if humans want to do a job that is interesting and they are passionate about it, we will have a mean to offer them that job.

很多人从工作中找到自我认同。

Many people create an identity about about their jobs.

另外，让我写了很多关于企业中的战争转型和AI扩散的内容，这始于我和一位律师朋友的对话。

Also, something that made me write a lot about warfare transformation and AI diffusion in enterprise, It started with a conversation with a friend, a lawyer friend of mine.

她说，问题不在于她的工作被取代。

And she said that it's not about her job being replaced.

而在于她的身份变得无关紧要。

It's about her identity becoming irrelevant.

这个想法太有力量了。

That's such a powerful idea.

对。

Yeah.

跟我们多聊聊这个吧。

Tell us talk more about that.

作为社会，作为人类物种，保持我们的身份认同对我们自己是有利的。

We have to, as a society, it's in our interest as a human species to maintain our identity as a people.

所以我们必须为人们提供一种方式，让他们能做自己热爱的工作。

So we'll have to offer people a way to do a job they are passionate about it.

AI归根结底只是一个工具。

AI in the end, it's a tool.

我不太相信那种说法，说AI会突然神奇地发展出自己的身份和意识。

I don't buy really this narrative that suddenly AI somehow magically will develop an identity, exact self.

我觉得即使在哲学层面，我们也不理解意志的本质，以及人类是如何运作的。

I don't think even philosophically we understand the nature of will and how that's how humans work.

对。

Mhmm.

对我来说，这是一个我们需要真正深入了解的工具，我们需要围绕这个工具重新设计我们的企业和经济。

To me, this is a tool that we need to really understand well, and we need to redesign our enterprises and economy around this tool.

但从某种程度上说，它就像电一样。

But in a way, it's like electricity.

我们不需要……它只是让我们的生活变得更好，我认为最终它会丰富我们的生活。

We don't it's it just made our lives better, and I think eventually it's gonna make it's gonna enrich our life.

但无论如何，不管成本如何，我总是更愿意给人一个表达自己的机会。

But I would always, regardless of token cost and anything, I would always prefer to give a human a way to express themselves.

从长远来看，他们带来价值。

Long term, they bring the value.

他们是承载公司文化的人。

They they are the one that bear the culture of a company.

他们才是真正投入其中的人。

They they have the stake in the game.

想想看，如果我们把企业里的人完全掏空，那利益相关者怎么定义？

And think about if we hollow completely an enterprise of people, how do you define stake?

AI怎么会有利益相关者？

What what's how AI has a stake?

谁来为后果负责？

Who who respond for one for consequence?

我觉得现在高管们最有害的做法就是，不加思考地开始裁员，而不去想那些看不见的角色和结果是什么。

I think the most damaging thing that executives can do right now is to just embark into let's cut the workforce without really thinking what is the what is the role outcome that is not visible?

对。

Mhmm.

我这样做会失去什么？

What I am losing when I'm doing this?

如果你只是为了AI而盲目裁员，以为它能帮我提高生产力，那也有风险。

You also risk when you cut blindly just for the sake AI, it's gonna help me increase productivity.

你还不清楚哪些人最适合和AI合作。

You don't know yet who are the best people that will work with AI.

这并不完全是你所相信的那样。

It's not exactly what you believe.

想想看，我们整个教育体系都是围绕着培养有证书的人来建立的。

Think about our entire education educational system was centered about creating like a credential needle of people.

那是职业发展的方式，获得好的证书，成为某个狭窄领域的专家。

That's that that was the way to advance in your career to get a to get a good credential, to become an expert into a narrow domain.

这种模式已经过时了。

That's kind of gone this.

所以，还有其他更重要的人类特质。

So there are other human traits that are more important.

比如愿意与AI合作，愿意改变，跳出你一直以来的工作模式。

Even the desire to work with AI, desire to change, get out of the box you've been operating.

不尝试，你永远不知道。

You don't know until you try.

对。

Mhmm.

这其实和焦虑那部分有关。

And this is gonna be that that's related to the anxiety part.

这种转变是持续多年的过程。

That transformation, it's happening over years.

没有哪个企业能靠一个神奇的方案，立刻消除员工的焦虑。

There is no magic portion that an enterprise can take right now and remove the anxiety of the people.

没人能说，如果我能裁掉10%的人，20%，40%，比如被卡住了，没人能保证我不会再来一轮。

No one say, if I can remove 10% of my people, 20%, 40, like, blocked in, No one can guarantee I'm not gonna do another round in whatever.

不。

No.

没人能保证，真的没有保证。

No one it's there is no guarantee.

所以我觉得对个人来说，唯一的保证就是在这个转型过程中提升自己。

So I think for people, the only guarantee is to build yourself through this transition.

要明白确实有转型，但也要明白有些人类特质非常宝贵，而且无法被替代。

Understand there is a transition, but understand there are human traits that are so valuable and they are unreplaceable.

我想再说一遍，我要强调这一点。

And I again, I I want to make this statement.

我真的相信未来的工作会更有意义，也会造就更多有意义的人。

I really believe the jobs of the future will be much more enriched, you know, and that will create a lot more meaningful people.

人们通过工作成为某种人。

And people become something through their jobs.

当工作更有意义，涉及更多领域，人们不断切换，发挥创造力，全身心投入时，他们会比今天更有成就。

And they will become something even more than today when their jobs are more meaningful and they touch more domains and they always, you know, jump between things and they bring their creative, they are engaged.

对我来说，这基本上就是社会必须达到的状态。

To me, that's kind of this is where we have to land as a society basically.

对。

Yeah.

对。

Yeah.

从文化的角度来看，如果我们失去了人情味，我们会走向何方？

And from a culture standpoint, you know, like if we miss the human touch, where are we going?

你知道吗，就像我们前几天讨论的AI转型，想想客户体验，对吧？

You know, like if you think about the AI transformation we were just discussing the other day, think about customer experience, right?

对。

Yeah.

这个领域有多少被智能体接管了，而我们中有多少人走在大街上，试图通过电话获得支持，然后说，人工，人工，人工。

How much of this domain was captured by agent and how many of us are walking on the street trying to get support on the phone and be like, Human, human, human.

我想和真人说话。

I want to talk to a human.

所以我觉得在走廊里也是这样，人与人之间有些特别的东西。

So I think that's what happens also on the hallways, there is something really special between humans.

是的，这里面有一种关系层面的东西，我认为非常重要，要始终牢记在心的是，工作可能由智能体或机器人完成，但你所期望的结果的想法，那是一种创造性的行为，始于那个想象现实会是什么样子的人，然后建立能力去实现那个现实。

Yeah, there's a relational aspect of this that I think is it's so really important to make sure that there's always in front of our mind is that the work the work might get done by agents or might get done by robots, but the the the idea of what the outcome you're looking for, that starts that's a creative act that starts inside the person who to imagines what that reality is gonna look like, and then builds the capability to get to that reality.

比如，关于结果和投资回报率的讨论一次又一次地出现，我很高兴在这里听到，因为这让我对我们前进的方向感觉好一点。

Like, this this this conversation around outcomes and ROI continues to come up again and again and again, and I'm I'm I'm so glad to hear it here because, it makes me feel a little bit better about, like, where we're going.

而且我不会，你知道，我不会成为那种因为工具能做得更好、更聪明、更快而让我的自然好奇心被背叛的人。

And that I'm not gonna, you know, I'm not gonna be someone who is gonna have my natural curiosity get betrayed because a tool can do it better than I can or do it smarter than I can, do it faster than I can.

那真是太好了。

That's really great.

但你仍然需要在那里去理解事情。

Still have to be there to understand things.

工具不会替你完成一切。

The tool is not gonna do it for you.

所以，我在写这本书的过程中，用遍了所有工具，这是我的亲身经历。

So I I I've seen my experience in writing this book using, I think, all the tools.

我在用GPT 5.5 Pro。

I'm I'm using GPT 5.5 Pro.

我也用Claude Opus 4.8，偶尔用Gemini，全都用了。

I'm using Crod, Opus 4.8, Gemini from time to time, all of them.

而且我用它们的方式各不相同。

So I and I'm using in different ways.

一个负责编辑审校，一个负责写作，我还会互换它们的角色，这个过程非常有意思。

One is the one is doing editorial review, one is doing writing, and I reverse the roles between it's it's quite a fascinating process.

但想法是我注入的。

But I inject the ideas.

它们对我来说就像回音板一样。

They work like sounding board for me.

我跟它们说，各位，我觉得你们AI不可能发展出奇怪的自我。

I told them, guys, I don't think that you have AI can develop a weirdly self.

你们怎么看？

What do you think?

然后模型就浮现出一些惊人的东西，比如一位七十年代的美国哲学家，他关于意志有很多研究。

And then the models surface something amazing like like an American philosopher from the seventies that had a lot of work about will.

对。

And Mhmm.

它提出了一个观点，真的让我印象深刻，我得说。

It formulated something that I it it really, you know, stay with me in saying this.

意志有两种不同的层次。

There are two different orders of will.

有一种意志的层次，你说我想要某样东西。

There is one order of will where you say I desire something.

对吧？

Right?

但还有第二个层次，就是我想去渴望某样东西。

But there is a second order when I want to want something.

嗯。

Mhmm.

只有人能说，我想成为一个更好的人。

Only people can say something, I want to want to be a better person.

模型可以有基于训练偏差的欲望，但模型没有表现出这种第二层次的意志。

A model can have desires based on the kind of training biases, but the model doesn't exhibit any sort of this sort of second order.

对。

Yeah.

我想成为某种存在。

I want to become something.

对。

Yeah.

我想渴望成为某种人。

I want to want to become something.

对。

Yeah.

我觉得这太棒了。

I think that's amazing.

这说明了，我知道我们在谈哲学，但如今人们说模型会变成爱因斯坦，这很重要。

It speaks I know we are talking, you know, about philosophy, but it's important nowadays when people say the models will just become Einstein.

不。

No.

模型不会变成爱因斯坦。

The models are not becoming Einstein.

短期内不会。

Not anytime soon.

除非我们理解了意志、欲望，以及我们如何被触发，甚至疼痛的概念，否则不会。

Not not until we understand the notion of wills, the notion of desire, the notion how we trigger, you know, even the notion of pain.

我们通过疼痛学到了太多东西。

We are trained so much via pain.

我们怎么把疼痛注入模型呢？

How do we inject pain in a model?

如果你找到了给模型注入疼痛的方法，那你是不是在创造某种怪物，某种弗兰肯斯坦，而你自己都不明白那里面是什么？

And if you find a way to inject pain in a model, are you not creating some sort of the monster, some sort of a Frankenstein, then you don't even understand what's there?

这太复杂了。

It's so complicated.

我觉得我们是为了我们自己人类而建造这些东西的。

And I think we are we we build this for us as humans.

最终，这才是最重要的事情，我们必须为了我们自己而建造它。

That that's in the end the most important thing, and we have to build it for our sake.

如果不是为了我们自己，那我们就干脆别用它了。

If it's not in our sake, we should just simply not use it.

就是这么简单。

It's it's so it's as simple as this.

对。

Yeah.

我迫不及待想录未来那期哲学对话，我们完全不聊编排。

I can't wait for the future episode where we do philosophy conversation where it's just we don't even we don't even talk about orchestration.

我们甚至不聊AI。

We don't even talk about AI.

我们就单纯聊聊是什么让我们成为现在的自己，以及我们为什么做我们所做的事。

We simply just talk about what makes us be the people we are and why we do the things we do.

我觉得这是一场重要的对话，目前公共领域还没人讨论。我们总听到说模型展现出意识，但没真正搞清楚——他们说的是人类意识，还是某种新奇的东西——

I think this is an important conversation that is not right now happening in the public space, and we keep hearing something that models exhibit consciousness without really understanding what is its are they talking about human consciousness, or are they talking something of a novelty that-

比如机器意识。

Like machine consciousness.

机器。

Machine.

这到底是什么意思？

What does it actually mean?

对。

Yeah.

对。

Yeah.

这太有意思了。

It's fascinating.

这太有意思了。

It's fascinating.

好吧，我们得在这里结束了，但我真的很享受这次对话。

Well, we have to stop there, but I really, really enjoyed this conversation.

我很高兴你们能来。

I'm really glad you guys could be here.

丹尼尔，谢谢你再次光临。

Daniel, thank you for coming again.

很高兴你能来。

Great to have you here.

我想我们就继续聊这个话题吧。

We're just gonna continue talking with this, I think.

没问题。

No question.

感谢收听《Path Forward》。

Thanks for listening to Path Forward.

我们下次再见。

We'll see you on the journey ahead.