阻碍人工智能采用的最大障碍并非技术，而是这一点 | 安永的 Errol Gardner

大多数组织都已经接受了云计算，并把能迁移到云上的东西都搬过去了。

Most organizations have embraced the cloud and have moved most of what's movable to the cloud.

看起来这正在慢慢演变成一个全新的行业。

It seems like it's sort of morphing into a a new industry.

任何组织在变革时最大的障碍，通常都跟人有关。

The single biggest impediment for any organization changing is usually something related to the human beings.

可能是领导层，可能是赞助方，可能是中层管理者，也可能是基层员工，或者全都有，也可能是其中几种的组合。

It could be the leaders, it could be the sponsors, it could be middle managers, it could be the workers, it could be whole, it could be a combination of.

员工们对人工智能是否会真正取代他们的工作感到非常焦虑。

There's huge anxiety in the workforce about whether AI effectively is going to displace their job.

明白了。

Okay.

我其实很愿意聊这个，因为我经常跟另一家大型咨询公司交流。

Well, I'm interested in talking because I speak quite a bit to one of the other big consulting firms.

我偶尔为他们写点东西，所以对他们的看法挺了解的，尤其是他们如何看待人工智能，特别是智能代理这个新领域。

I actually do some writing for them and so I have a good sense of what how they view this world, the new AI and particularly agentic world.

我觉得这是个好机会，想听听你们怎么看企业现状，AI在大型企业中的实际应用情况，尤其是智能代理系统。因为我的播客里经常请初创公司，他们都说自己的系统被广泛使用，但说实话，除了帮人管理邮箱这种简单任务，再往上就复杂多了，这正是我感兴趣的地方。

And I thought this would be an opportunity to hear from you guys how you see the corporate landscape, the penetration of AI in production in large corporations and in particular agentic systems because the, you know, I, I end up having a lot of startups on the podcast and they're all selling systems and they all say, Oh, it's, you know, everyone's using our system, but, you know, when you get beyond an agent, you know, watching your email inbox, it gets more complicated and I'm, I just, that's what I'm interested in.

通常我会让嘉宾先做个自我介绍，讲讲跟咨询或AI相关的背景。

So, usually I have the guests start by introducing themselves, giving a little bit of their background in so far as it's relevant, for, either consulting or AI.

然后你可以先谈谈，你对AI在企业中究竟带来了什么变化的看法或观点。

And then maybe you can start by talking about view or thesis on what's happening with AI in the workforce or in the enterprise, I should say.

谢谢你邀请我。

Well, thank you for having me.

能和你聊这个话题真的很棒，这无疑是当前企业、政府机构以及其他许多组织里最热门的话题。

It's great to be speaking to you about this, arguably the hottest topic in most corporate environments, government environments, and many other organizations.

我叫埃罗尔·加德纳。

So my name is Errol Gardner.

我负责我们全球咨询业务。

I lead our consulting business here at globally.

我已经干了大约六年了。

I've been doing that for about six years.

我在这个行业已经待了三十五年多，见过不少技术潮流，所以我想把AI放在历史上那些曾被说成会迅速改变世界的趋势里，来给大家一些背景。

I've been in this industry for over thirty five years, so I've seen a number of different technology trends during that time and probably can look at AI in the context of many things that have been historically spoken about as changing the world very rapidly and then maybe give some context on that.

我很期待和你聊聊这些不同的话题。

But look forward to talking to you about all these different topics.

我就先说说目前我在企业环境中看到的AI是什么样子，企业们在做什么，又是怎么应对的。

Maybe just in terms of how I see AI currently within the corporate environment and what businesses are doing and how they're responding to that.

我觉得首先得指出一个现实：在大公司里推动变革真的非常困难。

I think the first thing maybe just to highlight is actually the reality that change within large organizations is very difficult to execute.

所以，任何声称变革很容易的人，要么是从来没干过，要么就是对实际能实现的目标有点夸大其词。

So anybody who references or or pretends that to be otherwise, I think, is either has never done it before or potentially maybe being slightly economical with what is achievable.

这其中当然有很多合理且积极的原因。

And there are lots of very good and positive reasons for that.

大公司之所以存在，尤其是那些规模庞大、历史悠久的，往往有一套套成熟的机制、障碍，或者说推动因素——这些让它们成功，但也常常拖慢了变革的速度。

Organizations exist, and those that are big and at scale generally have been around for quite some time, and they've got, you know, various mechanisms, barriers, arguably enablers that have made them successful, but also sometimes can act as barriers to them achieving change quickly.

比如各种组织架构、管理体系，都是这么建立起来的。

That can be lots of structures organizationally that get put in place, management structures, management systems.

也可能是组织里的员工本身。

It can be people that work in the organizations.

也可能是某些世代的领导者，他们的思维模式和工作方式，未必能跟上新技术，也难以推动创新和创意。

It can be leaders of a certain generation, mindset, work style which doesn't necessarily align itself to new technologies and driving innovation and creativity.

也可能是监管环境的影响。

It could be a regulatory environment.

还有许多其他因素，都会对这件事造成阻碍。

There's a whole bunch and a number of other factors that can very much come in the way of that.

所以我觉得，目前市场上普遍在大力宣传人工智能能给企业带来的好处——比如提升生产力、提高效率、改善客户体验，以及在不同模式下创造价值。

So I think fundamentally what we're seeing in the market at the moment, I think, is a lot of talking up about the benefits that AI will bring to the corporate sector, what that looks like and what that could be both productivity wise, efficiency wise, customer experience, driving value in different paradigms.

但要实现这些目标，必须先跨越我刚才提到的那些障碍。

The route to achieve that, though, has to go through all of those other elements that I've just spoken to and about.

就我们接触的客户来看，当前的进展确实不错：员工们正在积极接纳人工智能，公司也在开展大量培训项目，帮助既有技术背景的人，也帮助没有技术背景的人，理解AI能做什么，以及如何把这种可能性转化为真正可扩展的能力——无论是覆盖整个价值链，还是贯穿完整的业务流程，更别提跨地域、跨服务和跨职能的全面推广了，但离真正实现这些，我们还有不小的距离。

So I think you'll find the progress that's happening currently, as we're seeing with our clients, is very good in terms of people adopting, employees embracing AI, lots of development programs in terms of those with both technology skills as well as those that don't have technology skills understanding the art of the possible, how you then translate that into building capabilities that genuinely scale, be it across a whole value chain, even across a full business process, let alone then across geos, across different services and functions, we're still quite a way away from that.

我说的‘还有不小的距离’，并不是指要花好几年时间。

And when I say quite a while, I don't mean years and years.

我的意思是，实际情况并没有像很多人说的那样，一夜之间就全面爆发了。

I just mean it's not happening like yesterday in the way that a lot of people are talking about in my whole opinion.

对。

Yeah.

你这里说的AI，是指广义的AI，还是特指智能体AI？

And you're talking about AI broadly or about agentic

我现在说的是智能体AI。

relation now to agentic AI.

这个区分其实非常好。

That's a very good qualification actually.

AI本身其实已经存在很长时间了，只是形式不同而已。从你的经验，还有我跟很多人的交流来看，你对这些不同形态都很清楚。

So, I mean, obviously AI actually has been around for quite some time, so in different variants, and I know from your own experience and talking with many others, you know about those different elements.

如果要说比较新一点的版本，比如机器学习，现在已经在很多组织的流程中大规模应用了，显著提升了效率，也深深融入了各个行业和流程。

So I think if you kind of talk about the slightly more modern versions of it, I think machine learning is used at scale in a number of different organisations across their processes is driving efficiency and is well embedded in many different industries and processes across the board.

再说到生成式AI，大概三年前开始，我们陆续部署了像大语言模型这类技术，本质上就是快速捕捉信息、处理信息，然后生成类似人类会写的文字和回复。

If you think about Gen AI, if you like, the kind of, I guess we were probably starting three years ago with the deployment of models, LLMs, and things of that nature, so effectively just capturing information and processing it as fast as possible and converting it into texts, responses that effectively a human would would would, put into place.

我认为很多组织都在广泛使用它来实现这个目的。

I think that is being used quite extensively in a number of organizations for that purpose.

比如，有些情况是利用组织外部的信息进行分析，这取决于具体流程——如果是做研究，互联网上显然有大量公开资料，可以借助各种工具来处理；同时，也有企业构建私有的大语言模型，让数据留在组织防火墙内，充分利用内部的知识和洞察，以安全可控的方式保护企业信息。

So as a way of, be it using information that sits outside the organization and doing analytics around that depending on the nature of the process, if you're doing research that's clearly widely available on the Internet and using all the tools that can enable that, but then similarly creating private LLMs that would stay within the firewall of the organization, leveraging the data, knowledge, and insight within the organization itself to deliver a secure and protected way of protecting the corporate information within the four walls of the organization and using that as well.

所以，我认为这种应用在商业领域正越来越普遍，但可能没有些人想象的那么广泛。

So I think that is used, again, increasingly across business, but that's probably, again, not as pervasive as some may may think that to be.

而我刚才提到的另一个重点是，过去一年左右，大家开始普遍谈论智能体。

And then the the element I was speaking to, obviously, is everybody is now talking about agentic and probably have been for the last twelve months or so.

我认为，如果真能实现，智能体才是真正具有变革性和深远影响的部分。

I think that's the piece that arguably is genuinely transformational and impactful were it to be able to deliver.

但当你开始思考智能体如何取代那些串联成工作流、融入业务流程、构成价值链的环节时，这才是真正需要下大功夫的地方，因为它从根本上改变了组织的运作方式。

But I think when you start looking then at agents replacing activity and activities that join up into a workflow, that join up into a process, that are part of a value chain, that's the piece, I think, that is there's a lot of work, because that is genuinely transforming an organization.

这不仅仅是优化现有的做事方式。

That is genuinely not even just tweaking how they do things.

你实际上是在说，必须重新构建整个组织的运作机制。

You're essentially saying you have to rebuild how an organization works and operates.

在规模上实现这一点，挑战还在前面，而不是已经过去了。

The challenge of doing that at scale is one that's ahead of us, not one that's behind us.

是的。

Yeah.

我有几个问题，关于JIN AI和智能体方面。

And I have a few questions, both on the JIN AI and agentic side.

智能体的采用或实验，是自下而上推动的吗？

Is the adoption being driven or experimentation with regards to agentic being driven from the bottom up?

比如员工自己用商业AI智能体来管理邮件、填表之类的？

You know, employees who are using an AI, commercial AI agent to track their emails or, know, fill out forms or something like that?

还是因为安全顾虑，主要由高层推动？我觉得后者会慢得多。

Or is it being because of security concerns and that sort of thing, is it really being driven from the top down, which I imagine would be a much slower process?

我觉得这两种模式，还有其他方式，其实都在并存，大家都在这个过程中不断学习。

Well, I think you could argue that both of those and other models are out there, and I think everybody's learning on this journey.

正如我之前提到的，大型企业通常都有各种管控机制，来保护自身数据、客户数据、员工数据，以及其他各类信息。

So, I mean, obviously, as I mentioned earlier, with large corporates, they have checks and balances in place that take things like data privacy of obviously their own corporate data, but also their customer data, their employee data, and various other things.

所以他们会建立机制，确保信息不会泄露到外部的大型语言模型中。

So they they would have put mechanisms in place to make sure that information doesn't get out into LLMs in the broader world.

但问题在于，怎么控制员工使用他们私下更习惯的工具，把公司数据导入这些工具，就为了能用它做点分析，或者快速解决某个问题。

So the challenge with that, of course, then is how do you control the employees that sit within an organization from leveraging tools that they may be more comfortable using in their private lives and moving corporate data into those tools in order to oh, I if only I could use it to to do this bit of analysis or understand, what the quick answer to is to this particular problem.

这确实是个现实而棘手的问题，很多组织都在纠结：自上而下和自下而上的力量该怎么平衡。

So that is a real and genuine issue I think that a number of organizations have to grapple with in terms of where the top down, bottom up meets.

如果员工普遍熟悉生成式AI工具，你该怎么防止他们把公司数据无意中上传到公共平台？

So if you have a workforce that is well versed in the usage of Gen AI tools, if you like, how do you control them not misusing data within an organization and putting it a little bit out there in the public domain?

因为你知道，互联网上那些大型语言模型都在用这些数据训练，而且一旦泄露，就完全无法控制了。

Because as you know, all these LLMs that sit more broadly on the Internet are training models and using that data then and propagating that in a way that is not controllable from outside of the organisation.

所以现在，几乎所有大型企业都会设置防护措施，监控员工是否把工作电脑上的数据传到那些不受控的平台或环境。

So I think I think increasingly you will now find most major, or I would argue all major organizations now will put guardrails to prevent that from happening and will monitor usage by employees of any movement of data from their work laptops to platforms, environments that that that are not controlled in that respect.

对一些富有创新精神的员工来说，这可能会觉得限制了他们的创造力。

And some therefore may see if you're more of the innovative creative side as an employee, you may see that as constraining your ability to innovate.

如果你只能用公司内部的工具和专属的私有大模型，可能就不是你最顺手的那个工具了。

So if you're using then the house tools and the house LLM to or the and and the curated private corporate LLM that is used by the organization, that you may not be it may not be your favorite tool of choice.

可能不是你习惯用的工具，而且你也不太喜欢这种限制。

It might not be the one that you're used to using, and you might not like the constraint.

这就容易成为一个问题。

And and so that can become an issue.

我觉得一方面，员工层面自下而上推动创新、尝试自己的创意和边界，是毫无疑问的，但可能得在一些他们不太适应的限制下进行。

So I I think you have no doubt both that adoption at, if you like, bottom up at the employee level driving innovation, testing their own creativity and the boundaries of what they'd like to do, but probably with guardrails they're not so comfortable with.

而从上往下看，我认为大多数企业都在鼓励员工接受人工智能，提升AI素养，多做相关培训，了解如何使用它，因为大家都知道，员工如果能熟练使用AI，会显著提升个人工作效率。

And then as I say from a top down perspective, I think most organizations are encouraging their employees to embrace AI and become AI literate and do more l and d around this and understand how to use it because I think people understand the benefits of employees being comfortable with it and enhancing their own personal productivity.

但正如我所说，这些做法都会配上各种限制，这就让员工感到矛盾：你们一边让我这么做，一边又给我设这么多条条框框。

But as I say, they'll do it with guardrails, which creates that that some at some levels, some of that disconnect in the eyes of the employees in terms of, well, what you're asking me to do one thing, and then you're you're putting constraints on my ability to do it.

是的。

Yeah.

这在咨询行业有变化吗？

Is this changing, consulting?

我的意思是，你们除了提供建议，也会自己开发工具吗？

I mean, do you do you guys build, for example, as well as advise?

是的，完全没错。

Yes, absolutely.

不过回答你的第一个问题，咨询行业确实在变化，而且我相信这种变化还会继续下去。

So but to answer your I mean, your first question, it it it is changing consulting, and I'm sure it will continue to change consulting.

你知道，咨询行业的核心本质就是推动变革。

And, I mean, the whole essence of the consulting industry, as you know, is essentially change.

我们的主要职责和目标，就是帮助组织实现转型。

That is our, you know, primary function and objective is to help organizations change.

所以，从定义上讲，我们这个行业必须随着工作世界中的每一次宏观乃至微观变化不断调整和进化。

So, by definition, we're a business, we're an industry that, also has had to evolve and change with every, you know, macro as well as, in some cases, micro change that happens in the working world.

因此，我们自身也在变化，这一点其实并不奇怪。

And so us changing is not particularly unusual in that respect.

我们曾经经历过很多周期，比如大量工作被转移到海外。

We've been through cycles with, you know, a lot of workloads move significantly offshore.

现在印度和其他许多地方都有庞大的业务团队，而三十年前根本不可能有这样的情况。

So, know, you have huge practices now that sit in India and various other parts of the world, which thirty years ago didn't happen.

企业内部咨询团队的兴起，让我们不得不调整和应对。

You've got the rise of in house consulting operations within corporates that we've had to change and react to.

这些年来，我们得应对的变动不少，这不过是最新的一波，我可以从两三个角度来总结。

So there's there's a number of things that that over time we've had to react to, and this is just the latest, which says probably I crystallize it in two or three different ways.

我们不能一边跟客户讲AI能怎么帮他们提升业务，一边自己却不用AI。

One is we can't show up to clients and tell them about the benefits of how they could, if they deployed AI, change their business unless we're doing it ourselves.

所以首先，我们必须在自己的公司里率先用上AI，展示我们是怎么把它融入工作和运营的。

So in the first instance, we have to show in our own business that we're at the front edge of deploying AI into what we do, how we operate.

这一点很重要。

And, you know, so that's one.

其次，我们向客户交付服务的方式，过去五六年也经历了巨大转变。

Secondly, the way we deliver services to clients, which obviously, interestingly, has also been through quite a significant transformation in the last five, six years.

在2020年之前，这个行业有个普遍认知：如果客户看不到你坐在他们的办公室、出现在他们的办公场所，就会觉得你根本没在认真干活。

So pre 2020, generally, this was an industry that if the client didn't see you sitting in their offices or their building or premises in some way, shape, or form, they didn't believe you were working for them.

他们会觉得你在别处偷懒。

They thought you were doing something else.

新冠疫情彻底改变了以往必须亲自到场服务客户的模式。

And obviously with COVID, we've kind of completely changed that paradigm in terms of people physically being present with client.

但现在有了人工智能，我们开始从以投入为导向转向以成果为导向——过去我们看的是你花了多少小时、多少天，做了什么才产出这个成果；现在我们可以说，这明明是用AI完成的，怎么可能花了这么久？

But now with AI, you get to a position of, okay, we're moving from a model where we measure consulting primarily by inputs as opposed to outputs, so how many hours have you worked, how many days have you worked, what have you done in order to produce this deliverable, this outcome, you can now say, well, surely used AI for that, so it can't possibly have taken you as long.

那这会是什么样子呢？

So what does that look like?

这对我们来说既是挑战也是机遇：我们不仅能比过去更快地完成工作，还能做更多事，最终取得更大的成果。

So that presents both a challenge and an opportunity for us in terms of how we operate because we can do things faster than we would have done historically, but we can also do more than we would have done historically as well and therefore achieve more as a consequence.

它正在彻底改变我们的服务交付方式。

So it's completely changing us in terms of how we do service delivery.

而我认为另一个关键点，也是我们最根本的职责，是如何帮助其他组织尽快适应这种变革。

And then the other critical piece of what I think are and probably our most fundamental role then is how do we help other organisations move across that change as fast as possible.

我敢说，任何组织在变革中最大的障碍，通常都跟人有关。

And what I'd say is the single biggest impediment from any organisation changing is usually something related to the human beings.

可能是领导者，可能是项目发起人，可能是中层管理者，也可能是基层员工，或者这些因素的组合。

So, it could be the leaders, it could be the sponsors, it could be middle managers, it could be the workers, it could be whole it could be a combination of.

那我们该怎么帮助组织部署并接受这项技术呢？

So how do we help organizations deploy and embrace the technology?

你觉得该怎么思考和重塑这个新世界，充分利用人工智能、生成式AI或者智能代理团队？

How do you think about and reimagine what the new world could be leveraging the AI, Gen AI, and or an agentic workforce?

但你怎么让那些必须推动变革的部门真正认同并全力配合呢？

But then how do you get the parts of the organization that need to make that happen bought in and aligned to making it happen?

这话题我们后面可以再深入聊，不过你应该也知道，我们之前做过一项关于‘重塑工作’的研究。

And it may be one for us to talk more broadly about later, but the as you can imagine, we talk a little bit about the Workforce Reimagine study that we did.

员工普遍对AI是否会取代他们的岗位感到强烈焦虑，这种担忧会导致他们在工作中抵制企业级解决方案的落地。

There's huge anxiety in the workforce about whether AI effectively is going to displace their role, their job, which is going to lead to resistance in the workplace to the deployment of enterprise wide and enterprise grade solutions.

确实如此。

Yeah.

你提到这一点挺有意思，作为顾问，要想有公信力，就得自己先亲身经历整个过程。

And that's interesting when you say that, you know, in order to be a credible consultant, have to, you know, go through the process yourself.

你能具体说说这个吗？

Can you talk about that?

你们是怎么开始采用生成式AI和智能代理技术的？我们先说这个，因为我还有一个关于咨询行业变化的问题。

How has adopted both Gen AI and agentic AI and well, let's do that first because I have another question about how consulting has changed.

我们显然是一家雇用了大量员工的公司。

Well, we obviously are an organization that employ a lot of people.

从全球范围来看，我们总共有大约40万人。

So at the broadest sense for globally, we're about 400,000 people overall.

这就是整个组织的规模。

That's the whole organization.

我们很早就决定，尽可能让所有员工都能使用这些工具，以便在公司内部充分利用生成式AI。

And and we very early made the decision to give as many of our employees as possible access to the tools in order to be able to leverage and use Gen AI within our environment.

我们同样很早就搭建了一个私有大语言模型，把大量内部知识和信息都迁移到里面，并配上我之前提到的管控机制，让员工可以自由探索和使用公司内部的数据与信息。

We, again, very early built a private LLM and moved a lot of our knowledge and information into that and gave people access to that with the controls that I've described so that they could experiment and do things with the information and data that we have within our organization.

我们还大力推动了学习与发展计划，特别是因为虽然很多员工本来就具备技术背景，但也有相当一部分人并不具备这些能力。

We also embarked on a huge learning and development exercise around especially though because we have a lot of people who've got technology skills natively, of course, but we have a number of people that don't have those skill sets.

我们鼓励尽可能多的人去接触AI，尤其是从生成式AI的角度来看，它让技术变得前所未有的易用。

And just encouraging as many people as possible to see AI and and especially if you think about it in the Gen AI context, it makes technology as easily accessible as you could possibly think.

所以我们要让他们放下心理负担，别觉得这是技术门槛，或者自己不是技术人员就做不了。

So to get it to stop thinking about the fact that this is a barrier or I'm not a technologist.

其实这件事没那么复杂，非常直接。

I can't do this to, actually, this is straightforward.

就像用搜索引擎一样简单，想查点什么信息，动动手指就能拿到。

It's as easy as using a search engine or something else in that in that context to get some of the information that you need.

我们花了大量精力，给员工提供工具和平台，配套培训，然后鼓励大家动手尝试。

So we've we've done a lot of activity around giving people the tools and the platform, giving them the learning and development, and then asking them to do experimentation.

同时，我们也在思考，哪些业务领域能最快见效？比如研究方式、提案时与客户互动的方式、合同流程的整合与提速——把这些环节优化后，就能更快地构建解决方案，把全球各地的知识资源，整合成一个任何人都能轻松访问的系统，让任何地方的顾问都能迅速调用，及时交付给客户。

And then also, I guess, down is to say which parts of our business can we move fastest in terms of driving, you know, be it the way we do research, the way how we think about engaging with clients from a proposal perspective, for instance, or how we contract and think about consolidating that and making that a much faster process to then, as I say, building solutions that are how we access knowledge that sits obviously anywhere globally, that we can now bring that into a much more accessible system to any consultant who's sitting anywhere and can bring that to the client as fast as possible.

我们做了很多方面的努力。

So a number of different elements of what we've done.

当然，我们也推出了不少宣传推广活动，推动员工主动拥抱这种变化。

And then, of course, there's big campaigns in terms of pushing that and and encouraging the workforce to to embrace that change.

同时，我们也明确告诉团队：只有展现出全新的面貌，我们才能在市场中保持可信度。

And as I say, to make it clear to our people that we need to show up differently in order to be credible in the marketplace.

这不仅仅局限于咨询业务。

And that's not just in consulting.

我们的税务业务也是如此。

It's true of our tax business.

审计业务也同样如此，从市场角度看，我们正在推动这些改变。

It's true of our audit business as well in terms of what we're driving from that market perspective.

这些就是我们重点关注的一些方面。

So those are some of the things that we focused on.

正如我所说，我们公司大约有九万名技术人员，他们参与构建和开发更复杂的解决方案。

As I say, we've probably got about 90,000 technologists within our business who are part of helping to build and engineer the more sophisticated solutions around that.

同时，我们也会与第三方、我们的生态合作伙伴合作，共同推动和实现这一目标。

And also, work with third parties, some of our ecosystem partners, to drive and enable that as well.

但这就是我们推动组织内这一变革的核心做法。

But that's the essence of how we've approached moving this forward within the organization.

对。

Yeah.

那关于智能代理呢？

And what about agentic?

你们有没有在试验或者已经采用了任何智能代理的工作流程？

Are you experimenting or have you adopted any agentic workflows?

是的。

Yeah.

这么说吧，过去大概一年左右，我们在这方面的投入明显增多了。

So I mean, again, that's probably for the last twelve months or so, we've had a lot more focus on that.

我们内部的代理数量一直在不断增加，你可以这么理解。

And so we we've we've got ever increasing number of agents, if you like, that sit within the business.

一开始大家还热衷于数一数有多少个，但现在不行了，毕竟我们是个遍布多国的大机构。

It's it it initially became a bit of a sport to try and count them, but now it's you know, we're a big organization across many countries.

这已经变成一种常态了。

It becomes a state.

现在根本没法再精确统计了，我们更多是通过定性的方式来看它们带来的影响。

You can't really count them in that sense anymore, and we look more qualitatively about what we're seeing.

所以，你之前提到过关于处理邮件的事，怎么分析信息、实时更新自己，无论是每小时还是每天，同时从服务交付的角度，把智能体部署到那些重复性很高的任务里。

So be it I think you mentioned earlier about in terms of looking at emails, about how you think about analyzing information, about updating yourself on new and information as it comes in on an hourly basis, could be a daily basis, but also deploying agents into tasks from a service delivery perspective which are quite consistent.

我们在税务业务中就看到很多这样的例子，比如自动生成报税表，现在在审计服务的交付方式上，也越来越普遍了。

So we've seen that a lot in our tax business that we've been able to do that in terms of producing returns and increasingly in the way that we deliver audits, that's a key element.

同样，在软件开发生命周期里，我们也从智能体的角度挖掘了不少效率提升的空间。

And similarly, the software development life cycle, there are a number of efficiencies that we're driving from an agentic perspective in that as well.

这方面已经越来越成熟了。

So that's maturing.

就我们自身而言，已经在企业内部把这套体系落地并实现了规模化运行。

As I say, we're our own corporate organization in that respect in terms of getting that and operating it at scale.

这些就是我们在全公司范围内应用智能体的主要场景。

But those are some of the areas that we're using agents across the business.

是的。

Yeah.

我能看出来这种融合，也许这种趋势已经持续很久了。

I can see this, convergence and maybe it's been going on for a long time.

我不确定。

I don't know.

但现在的顾问公司其实都在自己开发系统。

But, between, I mean, the the consultants are now building systems.

这些公司本来就是做系统建设的，但他们又像顾问一样，会去客户那里做业务审计，然后说：这个地方我们可以自动化，那个流程也可以优化。

They're companies that build systems that in effect are consultants because they'll go to prospective client, an audit of their business and say, well, this is where we can automate, you know, this aspect or this workflow.

再就是大语言模型本身，它们已经吸收了大量报告，包括麦肯锡的所有报告。

And then you've got, the LLMs themselves, which increasingly have ingested all of the reports, all of the McKinsey reports.

咨询行业的各种做法和成果，其实都有大量公开信息。

You know, there's a lot of, public facing information about, the consulting industry's practices and outcomes.

你可以直接去问大语言模型。

And can turn to an LLM.

我跟一家公司聊过，他们管这叫‘决策智能’，正在打造一个CEO的智能助手，让CEO能和这个智能系统一起梳理公司情况、头脑风暴、生成想法、获取信息，这样再去跟管理层或董事会提方案时就有备而来。

I mean, spoke to one company about, they're calling it decision intelligence and they're building in effect a CEO co pilot where the CEO would work alongside with this agentic system to explore the organization, brainstorm, come up with ideas and get information so that then they could go prepared to their management or their board to propose changes.

所以，你有没有注意到这种融合趋势？有些系统本身就在扮演顾问角色，有些初创公司卖产品，但又靠咨询手段把产品推入企业；而传统咨询公司，现在也开始自己开发产品了。

I mean, so do you see that kind of convergence that there's, you know, systems that are acting as consultants there, there are startups that are products, but then have a consulting side to get their products into the enterprise and then consulting companies that now build products.

这看起来正慢慢演变成一个全新的行业。

It just it seems like it's sort of morphing into a new industry.

是的。

Yeah.

你说得很有道理，这方面确实有很多讨论。

I think I think, there's a lot of in what you say, and there's a lot of both discussion about this.

围绕这个话题，可能还存在一些过度推断的情况。

There's maybe some extrapolation that takes place around this topic.

但我觉得这跟你之前提到的咨询行业变革有关。

But I think that's part of to your earlier point about the consulting industry changing.

可以说，整个生态都在发生变化——企业做什么、组织边界在哪里，科技公司做什么、他们的边界如何，初创公司做什么，还有你提到的那些传统服务商和咨询机构在做什么，都在变。

Arguably the whole ecosystem is changing, both what an organization you know, what a corporation does and what the boundaries of that organization are, what the technology companies are doing, what the boundaries of their organizations are, and what startups are doing, as you say, and what established service providers or consulting organizations are doing and what they're doing.

所以，在边缘地带，这些角色的界限正变得越来越模糊。

So I think at the margins, those things may be becoming a little bit more blurred.

我还想说，这正把我们推向一个最佳解决方案：各种角色之间形成协作模式，才能打造出最具吸引力的方案。

I'd also say that it's driving us to a point that the solution, if you like, the sweet spot is probably a collaboration model between various actors that I've outlined and others in order to make the most compelling proposition.

但我个人觉得，现在大家谈得很多的是，产品公司正在变成服务公司，服务公司又深度融入软件，诸如此类的现象。

But I do think, and my kind of personal view around this, is there's a lot of talk around product companies becoming service companies and service companies being embedded in software and and all of these types of things.

但事实上，这种变革的核心恰恰发生在服务型组织内部。

And I think in reality, the the process of change is fundamentally one that is embedded in a service organization.

如何将技术融入业务，从而实现商业成果和价值，咨询公司在这方面积累了多年的经验。

And how you how you implement technology and apply technology into a business in order to generate a business outcome and business value is something that the consulting organizations have many, many years of experience of doing.

我认为，我们还没到能把这套能力完全产品化的阶段，除非这个产品是被某个代理或非人类实体购买的。

I I don't think we're yet at the stage that you can codify that into a product unless the product is something that's been arguably bought by an agent and or a nonhuman.

只要我们这个行业本质上还是在帮助和支撑企业里的人实现业务目标，人们就依然希望与真人互动来完成这些事。

So I think and as long as we're an organization, an industry that is fundamentally helping and supporting human beings, people in corporations to deliver outcomes for their wider business, they will want to interact with people to do that.

同样，如果他们的团队是人和智能代理的混合体，那也必须确保人不会成为实现整体业务目标的瓶颈。

And similarly, if their workforce is made up of a blend of agents and people, you still have to make sure that the people are not acting as a limiter on the achievement of that overall business.

所以，确保这一点至关重要。

So having and making sure that's the case is critically important.

至于那种反乌托邦的设想——完全无人介入、彻底自动化，我觉得这倒是个有趣的理论模型，但你得告诉我，现实中到底怎么落地。

And then if you get to the dystopian world that says, yes, but maybe we'll operate touchless without any people, no humans in the loop, I think people need to deal with, that's a very interesting theoretical model, but tell me how that's gonna work in practice.

我觉得技术存在的意义，应该是让人类的生活和企业运营变得更好、更充实，不是靠金钱，而是看我们能达成什么成就，而不是变得多余。

Because I think the reason technology exists is to make, hopefully, the human condition and human life better and corporate life better and and enriching us, not in a monetary sense, but in in terms of what we can achieve as opposed to in a redundant sense.

是的。

Yeah.

对。

Yeah.

说得对。

Good point.

关于智能代理，你觉得在企业整体层面，我们现在处于什么阶段？

The On agentic, where do you think we are in the enterprise writ large?

打个比方，从零到十，零代表职场里完全没有智能代理。

Say on a scale of zero to ten, zero is no agentic AI in the workforce.

十代表智能代理的应用达到极致。

10 is sort of maximal agentic use.

我不确定那会是什么样子，但估计我们最终会走到那一步。

I don't know what that looks like, but presumably we'll get there.

我的意思是，你觉得我们现在大概就处在一到二的水平吗？

I mean, you think we're just, you know, at one or two on that scale?

我们是不是更接近五了？

Are we closer to five?

你觉得这个进展会有多快？

And how quickly do you see that moving?

这是个很好的问题，不过我得说，我怎么回答都有可能让自己陷入被动。

It's a great question that I I I can realize that the way I answer this could, could lead me as a hostage to fortune in many different ways.

但我要是看技术过去是怎么被应用的，也许我反过来问你一个问题，可以吗？

But I mean, what I would say if I look at how technology, as I say, has been deployed historically in the and maybe I'll ask you a question, actually, if I'm allowed to do that.

你觉得云计算在企业界现在处于什么水平？一到十分的话？

Where would you say cloud is today in the enterprise world on a one to 10 scale?

虽然我不做这行，但听来的印象是，大概在七分左右吧。很多公司因为合规限制，没法把所有东西都上云，但绝大多数已经把能迁的都迁过去了。

Well, not being in the business from but from what I hear, I would say it sets seventy and seven on a scale of one to 10 that most organizations in so far as they can, their regulations have prevent some stuff from moving to the cloud, but most organizations have embraced the cloud and have moved most of what's movable to the cloud.

我不太确定。

I don't know.

这跟你的观察差不多吗？

Is that close from what you see?

我觉得这个问题很难回答，但我觉得应该不到70%。

Well, don't I don't I think it's a very difficult question to answer, but I would say it's less than 70.

也就是不到十分之七。

So so less than seven out of 10.

我这么说不是因为有七成的公司已经采用了云技术，而是他们实际用到了什么程度？

And the reason I say that is not that it may be seven out of 10 organizations have embraced it, but how wide is their usage of it in reality?

你说大多数能上云的都上了，但到底什么才算‘能上云’呢？

So you said that most that could be moved to the cloud, whereas what's determining what could be moved?

我觉得这个标准在不同行业里差别挺大的。

Because I think you'll probably that find that varies by the same the same industry.

不同公司对哪些能上云、哪些不该上云，判断都不一样。

Different companies will make a different determination of what could and shouldn't be moved.

我举这个例子是因为云技术已经存在十五年了，到现在才达到你所说的十分之七，已经花了这么长时间。

And the reason I use that as an example is that cloud's been around for fifteen years, and it's taken that long to get to let's even use your number of seven out of 10.

我的意思是，正如你刚才说的，我们的实际水平连百分之一都不到。

So, I mean, what I was going to say, which I think is validated by what you say is we're less than one.

现实是，我们要想达到十分之七的水平，还有很长的路要走。

And and the reality is that we there is so far for us to go to get even to seven out of 10.

我并不是说需要十五年，但肯定远不止十五天，甚至十五个月都不够。

I'm not saying it'll take fifteen years, but it's gonna certainly take longer than fifteen days or even fifteen months.

我觉得，基于我前面提到的那些限制因素，这件事要做得更好，难度要大得多。

And I think that for all the constraint points that I made earlier, it's it's much more difficult to do this.

实际上，你甚至可以说，上云几乎是实现这一切的前提条件。

You could argue, actually, that cloud is almost a prerequisite to being able to do it as well.

所以，这两者之间存在很强的相互依赖关系，必须同时推进。

So there's there's a a codependency there in terms of making sure that you can do that.

当然，也有人会说，情况未必如此。

I mean, some would argue that's not necessarily the case.

但我认为，我们在 adoption 曲线上还处于非常非常早期的阶段。

But but I think that we are very, very early in terms of its adoption curve.

再说一遍，可能有不少机构在用，但我估计这比例最多也就20%，而且只是浅尝辄止。

Again, maybe you have a relatively large number of organizations, although I'd say that's still probably only 20%, who use it a little bit.

我是说，真正用在生产环境里，不是玩玩而已。

I mean, in production, not playing with it.

确实是用在实际业务中了，但只占他们整体业务的极小一部分。

I mean, it's genuinely being used in a production scenario, but it will be at a very, very small element of their overall business.

是的。

Yeah.

所以我认为，现在的情况就是我刚才说的这样。

So I think we're we now that will be my view.

正如我所说，我这样精确地描述，无疑是在冒险，把话说得太死。

And as I say, I'm a hostage to fortune, no doubt, in terms of committing to something so mathematical in in describing it.

但我想要强调的是，这背后其实藏着巨大的机会——企业如果能安全、负责任地应用这项技术，就能真正获得它的价值。毕竟，这项技术可能对员工产生深远影响，我们应该着眼于为员工创造新的职业机会，让他们以完全不同的方式工作，而不是仅仅盯着效率，最终变相削减人手。

But what I'd say to you is that that reflects the huge opportunity that's still ahead for organizations to get the real benefits of deploying this technology and doing it in a safe way, doing it in a responsible way given the impact it could have potentially on employees, doing it in a way that will generate and create new opportunities for employees in the future to work in a very different way as opposed to purely looking at it from a an efficiency perspective that may end up ultimately be a proxy for saying that we want to reduce our headcount.

所以我觉得，我们还处在非常早期的阶段。

So I I think I think we're early.

我真心觉得，在这种情况下，实际比例还不到百分之一，我也明白市场上有些说法过于夸张，好像我们已经走得远多了。

I genuinely think we're at less than one in that scenario, and I appreciate some of the hyperbole in the market would suggest that we're much further down.

但就我目前接触的企业领域来说，确实如此。

But in the enterprise space I'm speaking to now Yeah.

我觉得，我们离真正起步都还很近。

I think, we're still very much close to the starting gate.

是的。

Yeah.

根据你在安永看到的代理工具，无论是实验阶段还是已经投入生产，你认为这些炒作最终会成真吗？还是说，研究人员永远解决不了信任和可靠性的问题，导致代理工具最终只能帮你看邮件，而无法真正支撑关键业务流程？

And are you from what you've seen of agents either in experimentation or production at EY, do you think that the hype is going to be born out or do you think, I mean, could be that they can never, the researchers can never resolve the trust issue, never resolve the reliability issue and agents just fall to these, you know, watching your inbox rather than, you know, critical business processes.

我认为，从实验和创新阶段跨越到大规模应用，真正把代理工具融入整个工作流程，而不仅仅是单个任务，这个门槛相当高。

Well, I think the, if you think about the barrier of getting from experimentation innovation to adoption at scale and effectively deploying across workflows, not just single tasks or activity, that is quite difficult hurdle to to overcome.

但我相信，肯定会有越来越多的企业在某些业务环节和价值链部分实现这一点。

But I think it will you know, you'll start to see many more organizations achieve that in terms of certain elements of their business and parts of their value chain for sure.

我觉得，归根结底，有一些宏观因素可能会阻碍我们常说的这趟人工智能列车前进。

I think, you know, I go back to there are macro factors that are going to challenge, if you like, the train that the AI train that we all speak to and speak about.

我们刚才聊了点云的问题，回头再想想，也许我们可以把视角拉得更广一点，不只盯着人工智能本身。

And so I we've just talked about cloud a little bit, if you like, and and just going back to that, maybe just to zoom out a little bit from AI as as, a topic in its own right.

你知道，当今世界正发生着许多变化，这些因素可能比2011到2012年那种认为全球化会永续推进的假设，更阻碍了人工智能在全球的快速普及。

As you know, there are a number of dynamics happening in the world today that are probably acting more as impediments to the fast adoption of AI globally in compared to, if you think about in 2012 2011, 2012, the globalization train that was assumed to be acting in perpetuity.

想想过去五年世界的变化，现在各国对数据和信息的敏感度高了很多，尤其是对关键基础设施和产业是否能依赖国外供应，顾虑重重。

So if you think about how the world's changed in the last five years and how we are very much more in a space where there's sensitivity around data and information from a nation state perspective and how much you can rely on critical infrastructure and industries being deployed and delivered from countries outside of the nation that you're sitting in.

所以，国家层面推动的数据主权和相关政策，这在欧洲尤其明显。

So sovereignty and directions around that being driven by governments, that's especially the case in Europe increasingly.

但我觉得，其实在美国，这种情况一直存在，只是现在更突出了。

But I think you're seeing it well, it's arguably the case it always has been the case in The US.

现在全球越来越多国家都在担心，不能对这么一项变革性技术完全依赖，万一出问题，连基本的控制权都没有。

But you're seeing it in in many other countries around the world now who are worried about we cannot end up being dependent on such a transformative technology without having some level of control in case something goes wrong.

这算第一个因素。

So that that would be one.

接下来，还有关于工作被取代的担忧——说到底，谁来当守护者，保护员工不被大规模裁员？

You then have the challenge of the narrative around the displacement of work that ultimately you say, well, who's going to act as custodian to protect employees from mass redundancy?

公司会做这件事吗？

Is a company gonna do that?

科技公司会做吗？

Are the technology firms gonna do that?

还是得靠政府来承担这个责任？

Or a government's gonna have to do that?

我认为最终政府必须主动介入，确保这种情况不会发生——无论是通过监管、社会压力、征税，还是其他手段，来阻止企业大规模用机器人或智能代理取代人类员工。

And I think you'll find, ultimately, that governments are gonna have to lean in to make sure that doesn't happen, which is going to inevitably lead be it through regulation or peer pressure taxes or whatever it may be that will disincentivize organizations from displacing people, and replacing them with robots or agents or any other at at scale, in that respect.

而另一方面，从更长远看，世界正朝着一个生产力和劳动力供给持续下降的方向发展。

I think then you've got, the other dimension of this ultimately is is the, in many ways, the balance the other way that says the world's moving to a state where our productive workforce and capability is declining.

尤其是在西方国家，大多数主要国家的劳动年龄人口都在减少，而这个群体还要承担越来越重的负担——既要照顾婴幼儿，又要赡养老人，这些人从某种意义上说都不直接创造生产力，我想你明白我的意思。

So especially in the Western world, we've got a declining working age population in most major countries and increasingly, a burden falling on that working age population to manage the two ends of the spectrum, the babies and the old people, who are nonproductive, if you like, in a a way that you you understand what I'm saying.

是的。

Yeah.

那么，AI 能多快地填补这种生产力缺口，同时又不至于大规模裁员，从而避免陷入另一个极端？

So how quickly can AI deliver the solution to solve for that productive capacity gap without displacing people en masse that you have the problem the other way.

我觉得这是一个宏观层面的大问题，某种程度上，根本不存在现成的模式。

And I think that is such a macro issue that in many ways, you there is no model.

也没有什么现成的平台。

There's no platform.

更没有实际的证据能告诉我们该怎么解决这个问题。

There's no empirical evidence of how you can solve for that.

但我认为，无论是跨国政府还是国家政府，都得认真思考：该怎么应对这种情况？这势必要给一些技术和创新踩下刹车，限制企业快速采纳它们的能力。

But I think a number of governments, be they transnational or national governments, are gonna have to think about in terms of how do you manage this, which inevitably is gonna put the brakes on some of the things and some of the technologies that are out there and the ability of organizations to adopt them quickly.

另外我想说的是，眼下科技行业从人工智能中获取了巨大的价值提升，这种价值提升更多体现在人们的认知上。

The other one I'd just say to you as well is what's also interesting, of course, is that the the value exchange that's happening at the moment between the technology industry, which is getting a huge amount of value uplift from the AI, if you like, and the AI, lift that is is the perception of that.

我说的不是股市表现，也不是在讨论泡沫之类的问题。

And I don't mean just in stock market terms, but and I don't it's not about talking about a bubble or in things in that nature.

而是说，大量工作和任务正从传统行业转移到科技公司手上，由它们代为运作。从长远来看，这种模式能持续吗？那些传统企业——石油、天然气、金融服务业等等——它们的商业模式还能撑得住吗？

But the transition in many ways of workload and effort moving from traditional industries to technology firms doing things on their behalf, how sustainable is that in the longer term when essentially those companies, those traditional consumer companies, oil and gas companies, financial services and the like, how sustainable is that for those business models going forward?

这又是另一个值得关注的趋势：未来这种关系中的公平性会怎样发展？我们怎么确保科技行业不会变得过于强大，以至于控制甚至取代其他行业，接管原本由各大机构提供的服务？

And that's another interesting dynamic that I think we'll need to watch going forward into the future as to, well, how where's the equity in that relationship going forward, and how how do you kinda make sure that the technology sector doesn't become so powerful that it's controlling so many other sectors and or even replacing them in terms of the services that you see currently offered by other major organizations.

是的。

Yeah.

而且说实话，即便说到具备自主性的系统，从相对角度来看，可能也还远得很。

And and I mean, as much as agentic may be within, you know, speaking relatively in relative terms, maybe far in the future.

不过，这也许只需要几年，而不是几十年。

I mean, that could simply be years as opposed to decades.

但我认为，整个劳动力转型的问题，政府必须得想办法应对。

But I think that this whole workforce transition is going to have to be addressed by governments.

我不是说靠规范招聘和解雇这些做法，而是可能通过税收来给被裁员工提供全民基本收入。

I don't think through regulation of hiring and firing practices itself, but maybe through taxation to provide universal basic income to people who've been laid off.

但说真的，这已经挺玄乎了。

But you know, that gets very speculative.

今天你提了两个问题。

Today, when you're Two questions.

当客户来找你的时候，你能大概说说都是哪些类型吗？

When customers coming to you, can you break it down just very roughly?

有多少人是来寻求传统咨询服务的？

How many are coming for traditional consulting help?

我们正在转向这条新产品线。

You know, we're moving into this product line.

我们正在逐步退出那条产品线。

We're transitioning away from that product line.

我们该怎么操作呢？

How do we do that?

其中有多少是跟人工智能相关的，又有多少是属于自主型人工智能呢？

How much of it is AI related generally and how much of it is, you know, what is agentic AI?

我们该怎么利用它？

How can we use it?

我觉得，不谈具体数据的话，绝大多数还是在解决传统的商业问题。

Well, I think, again, without getting mathematical, I think the vast majority is still solving traditional business problems.

所以我们打算推出一款新产品。

So we wanna launch a new product.

我们想收购一家公司，来提升我们的价值。

We want to buy an organization to lift our value.

我们希望更好地渗透客户市场，提升客户体验。

We want to get a better penetration to our customers, better customer experience.

我们希望提高生产效率。

We want to drive productivity.

我们希望供应链能更高效一些。

We want, you know, more efficiency in our supply chain.

不管这些目标是什么，它们都在推动我们的业务发展。

Whatever all of those things are still driving our business.

不同的是，现在的挑战越来越在于，我们必须更高效、更聪明地利用AI来实现这些目标。

Probably what's different is the challenge is coming increasingly as to and we need to leverage AI to do that in a much more effective, efficient way.

这确实没错，而且我们看到越来越多的企业在投入长期的非AI技术项目。

That for sure is the case, and and we're seeing increasingly as organizations commit to be it longer term technology projects that are non AI related.

比如部署所谓的核心业务系统，或者升级他们的ERP平台。

So it could be putting in, you know, what are you called the systems of record, changing their ERP platforms.

他们越来越常说，好吧。

They're increasingly saying, okay.

这可能是个三年的项目。

That might be a three year project.

你能加快进度吗？同时用AI以一种完全不同的方式来实现？

Can you do it a, faster, but also then deploy AI to do it in a very different way, than you would have done it in the past?

他们会预先算好未来AI发展带来的成本节约，以此来达成目标。

And they'll kind of build in, almost savings expected from future developments in where AI is going in order to achieve that.

我觉得现在更普遍的是，企业依然想通过传统方式提升业务价值，但会把这当作一个跳板，尽快从中获得AI应用的好处。

So I think there's more of a push to achieve through the traditional things that companies want to drive in terms of delivering business value in the traditional sense of the way you describe it, but to use that as a mechanism to to lift, to drive, to get the benefits as a of AI and and its usage as quickly as possible.

所以现在的情况更可能是，客户不会直接说‘给我做个AI功能’或者‘把我们的业务全代理化’，而是会先问：你代理化到底是为了什么？

So that that's probably more the way it's happening as opposed to people coming clients coming, saying, can you give us an AI thing, or can you agentify our business without a context of, well, what are you agentifying it for?

是的。

Yeah.

对。

Yeah.

你觉得现在大多数行业里，有没有一些以智能体为核心的初创公司，想打造能跟行业巨头竞争的业务？

Do you think that there are in in most industries now, agentic first startups that are hoping to build businesses that compete with the majors in whatever industry they're in.

你认为未来会不会出现一大批新公司，它们不仅天生就是生成式AI驱动的，更是智能体原生的，因此比传统巨头更快速、更便宜、效率更高？

Do you think that there will be the growth of new companies in many industries that are not only Gen AI native, but agent native, and therefore will be faster, cheaper, more productive than the traditional leaders?

简单来说，答案肯定是肯定的，这点毫无疑问。

I think you well, the very simple answer to that is yes, no doubt.

不过我要补充的是，得把这个问题拆开来看，具体到某些行业，或者企业里哪些环节更容易被颠覆。举个例子，过去你可能会把这部分业务外包出去，但未来可能有家你提到的那种初创公司，直接把这块业务抢走。

The the however bit that I would say is you kinda need to break that down beyond the generic into, let's say, particular sectors where that's more likely to happen and and, you know, maybe even parts of a business that are likely to parts of the value chain of the business that you could actually you know, you a yesterday's a traditional model, maybe you'd outsource that, but tomorrow's is actually somebody comes along and takes it away from you, one of the startups that you're describing.

所以我觉得就是这样。

So I think that's it.

显然，智能体模式更适合那些以信息处理为核心、产品不一定是实体的行业。

And, obviously, agentic lends itself more to industries which are more focused around processing information and and where the product is not necessarily a physical product.

它的产品可能是数据。

It's it's actually it could be data.

也可能是数字化的服务。

It could be digital.

也可以是服务。

It could be a service.

比如，金融服务就是一个很典型的例子。

So for instance, financial services would be an obvious one, in that regard.

不过，像这样的行业肯定还有其他挑战，因为它们受到严格监管，你不能随便跳进去就开始接管其中的一些环节，还得确保自己能符合监管框架的要求。

But then you've obviously got other challenges in industries like that because they're heavily regulated, and you can't just kinda jump in and and start taking over bits of that without, kind of making sure you can adhere to a regulatory framework.

所以我觉得，这类情况肯定会发生，但要想在多个地区、多个行业内大规模落地，对很多初创公司来说，执行起来会非常困难。

So I think you for sure will see it, but the ability to do that at scale, again, both across multiple geographies but also across multiple companies within a sector, is going to be hard for a lot of the startups to to execute on.

因为就像我刚才说的，创业公司通常集中在科技领域，而且主要面向消费者市场。

Because as I say again, if you're a startup, your startup startups are generally in the technology space that are in a b to c world.

所以如果它们做的是面向消费者的业务，就能比较快地让用户接受和使用。

So if they're operating in the consumer end of the spectrum, you can find a way to get to consumer adoption relatively quickly.

但要想让企业客户买单，就没那么容易了。

Getting through to an enterprise buyer is not straightforward.

对。

Yeah.

是的。

Yeah.

没错。

That's right.

我刚才并不是想忽略《工作重塑》这份报告。

And and I I didn't mean to to ignore the work reimagined, report.

你总体的发现是什么？

What what were your general findings?

我知道时间快到了，但想聊聊AI对劳动力的影响。

I mean, I know we're running out of time, but, with with regard to AI's impact on the workforce.

正如你所想的，我之前提到过，AI的普及和成功，关键在于人这一层面。

Well, as you can imagine, I I touched on it earlier that the I mentioned earlier that the, you know, the people dimension of of the proliferation and success of AI will be significant.

比如现有员工的焦虑情绪。

And so themes around anxiety of the existing workforce.

因为很多人只是被网上和社交媒体上被放大的那些炒作吓到了，担心这会对自己的职位造成影响。

So because a lot of it is just the hype that they read, you know, online, in social media, and how things get amplified in a particular way, there's a concern as to what that will mean for their roles.

另一个问题是员工在学习和掌握新工具方面所获得的培训不足，以及公司如何鼓励他们把这些工具应用到实际工作中。

Another was around the gap of the learning and development that employees are getting to embrace some of the tools and be able to use that in their work environment and how they're being encouraged around those things.

但我认为领导者尤其需要注意的是，整个组织对人工智能的愿景——也就是AI对员工意味着什么。

But I think there was also a little bit, which I think is for leaders to be very mindful of, is the vision for an organization as to how how they want to embrace AI and what that means for the people within the organization.

这不一定是针对具体岗位，比如客服中心的员工，而是更宏观地问：我们的核心理念是什么？

So not even necessarily then at a micro level saying, you know, an individual that works in a call center, but actually just more broadly, what's the philosophy?

我们的愿景是什么？

What's the vision?

这件事有多重要？

How important is it?

当员工被要求日常使用AI时，他们开始担心，自己正被推向一个模糊不清的未来——他们根本不知道这个愿景或理念最终会导向哪里。

And by by getting people to, you know, adopt and use and leverage AI in their day to day lives, the the the staff, the employees are getting a little bit worried that they're being moved into a scenario that, you know, a vision, a philosophy, an endgame that they're not clear on in terms of what that looks like.

所以，我认为管理层、高管团队必须加强沟通，明确表达他们对AI的态度，说明在使用和部署AI时会设定哪些边界，这至关重要。

So I think just increasing the communication from a leadership and sponsorship level and and the c suite and executives into the workforce to be clear about how they feel about that, what what are what are some of the guardrails they would think to in terms of the use and deployment of this will be really important.

是的。

Yeah.

而且你似乎更关注对员工的影响，而不是管理层。

And and it seems you're you're very focused on the impact on on the worker, not so much on management.

其实在这项研究里，员工这个群体是包括管理层的。

Well, in that in that study as well, you well, I think the worker includes the management in that sense.

但如果从高层管理的角度来看，这里还涉及代际差异——通常年轻员工更愿意接受新工具，乐于尝试和创新。

But I think the if you think about it in terms of senior management, I think you also have a generational issue in terms of what you typically find, of course, is that younger workers are quite happy to embrace using the tooling and and experiment and innovate.

但他们往往在为不认同这些创新、也不愿奖励或鼓励这类行为的上一代人工作。

But oftentimes, they're working for people who are of a different generation who do not necessarily embrace that innovation themselves personally and will not necessarily reward those that and encourage those that are driving that activity.

所以这里确实存在一个代际之间的职场问题。

So there's certainly a kind of workforce intergenerational issue, at play there.

但我认为，这种影响会波及各个层级，可能关系到员工在组织里的长期发展。

But I think it affects all levels in terms of potentially the impact, on on their, you know, longevity within an organization.

是的。

Yeah.

明白了。

Okay.

我们已经聊了一个小时了。

We're up to an hour.

我的意思是，你有时间限制吗？还是我们就到这里？

I mean, you have a hard stop or should we end it here?

我没有硬性安排，不过……

I don't have a hard stop, but I

但你挺忙的。

But you're busy.

是的。

Yeah.

没关系。

No, that's okay.

我就是这种人，能一直聊下去。

It's just it's the sort of thing I can talk forever.

咱们

Let's