AI 引发的矛盾感受与工作重塑
还有其他人有这种感觉吗？无论你对 AI 持什么看法，总觉得哪里有点不对劲？用它，你担心自己在偷工减料；不用它，你又觉得自己正在错失机会。感到焦虑会被人轻视，保持乐观又会招来白眼。所有事情都显得有点别扭。在过去的几年里，我一直在为企业高管和董事会领导层提供咨询，探讨如何将 AI 融入工作之中。走过一间又一间会议室，走访一家又一家公司，我可以告诉你，有这种感觉的绝不只有你一个。这种令人困惑的感受，正是某种更大变革的表征。AI 已经在重塑工作了。
英文原文
Does anyone else feel, no matter what you think about AI, [that] it just feels a little bit wrong? You use it, you're afraid you're cutting corners. You don’t use it, and you feel you're missing out. You're anxious, you get dismissed. You're optimistic, you get side-eye. Everything is just a little bit off. I've been advising executives and board leaders for the last several years on how to integrate AI into work. Sitting in room after room, company after company, I can tell you, it's not just you. This confusing feeling -- it's a symptom of something bigger. AI is already redesigning work.领导团队要么慌乱不堪，要么陷入停滞：要么以缺乏协同的方式一下子掀起过多变革，要么任由我们所有人各自摸索、自行重塑自己的工作。这种人人各自为战的工作重塑或许能带来个人的生产力，但无法实现规模化。
英文原文
Leadership teams -- they’re either frantic or frozen: spinning up too much change all at once in an uncoordinated way, or letting all of us do individual DIY of our own job redesign. This person-by-person job redesign can create individual productivity, but it doesn't scale.让我给你们举个例子。我最近接触过一位资深管理者，她说：“我的团队产出增加了 10 倍，速度加快了 10 倍，可我的工作却感觉难了 20 倍。”这位管理者觉得她更孤单了——协作并没有变多。产出的质量有时更好，有时更差。我同许多管理者聊过，不止是这一位管理者如此，许多人都有这种体会。我想向大家传达的核心观点是：未来几年内，AI 将大幅重塑半数以上的工作。因此，这从来不是我们要不要重塑工作的问题，而始终是一个如何去做的问题：我们该如何引领这场重塑。
英文原文
Let me give you an example. One well-seasoned manager I spoke to recently said, "My team is producing 10x more output, 10x faster, yet my job feels 20x harder." This manager felt she was more lonely -- not more collaborative. The output sometimes was better, sometimes was worse. I talked to lots of managers. It wasn't just this one manager. Lots of people feel this way. My key message to you: AI will dramatically reshape more than half of jobs in the coming years. So this was never a question of whether we redesigned work. This was always a question of how: how we lead through that redesign.今天我将向大家阐述三个领域，在我看来，我们在这些领域的重塑中尤其需要领导力。我将通过一些例子说明我们如何应对这些问题：注意力（attention）、责任机制（accountability）、学徒机制（apprenticeship）。
英文原文
I'm going to take you through three areas today where I feel we particularly need leadership in the redesign. I'll walk through some examples of how we might address this: attention, accountability, apprenticeship.第一领域：注意力与“大脑过载”
首先是注意力。我和同事们一直在研究与 AI 协同工作的体验。一位资深工程经理曾说：“这就像脑子里打开了十几个浏览器标签页，全都争抢着注意力。我费力去管理工具的心思，比用来解决实际问题还要多。”
这就是“大脑过载”（brain fry）。这并非职业倦怠（burnout），而是另一种状态。它源自对 AI 工具的高度密集监督。它表现为大脑昏沉、决策放缓以及错误增多。经历大脑过载的员工报告的严重错误增加了 39%，产生离职倾向的比例也高出 39%。你最活跃的 AI 使用者正面临流失的风险。
英文原文
First, attention. My colleagues and I have been studying the experience of working with AI. One senior engineering manager said, "It's like having a dozen browser tabs open in my head, all fighting for attention. I was working harder to manage the tools than to solve the problem." This is brain fry. Not burnout -- something different. It results from the acute oversight of AI tools. It’s mental fog, slower decision-making, more errors. Workers who experience brain fry report 39 percent more major errors. They are also 39 percent more likely to intend to leave. Your most active AI users are at risk of walking out.没有人故意要去制造大脑过载。它的发生，是因为我们在现有工作流之上不断叠加工具，却没有减去任何工作，也没有从团队层面去通盘考量。我们新增的工作往往对认知有着更高的要求。
那么我们该如何应对？我们需要以结果为导向，由团队协同重塑工作和工作流。我们需要让管理者参与进来。我知道管理者手头上已经承担了很多工作，但当管理者出面为员工解答关于 AI 的疑问时，心理疲倦度评分会降低 15%。有一家公司深入思考了这一点，着手重塑岗位说明书，明确了更具战略性的工作。他们去除了不再需要的交接环节，制定了广泛的技能提升计划。这一切都是协同推进的，而不是由个人单打独斗完成。
英文原文
No one set out to create brain fry. It happened because we piled tool on top of tool on existing workflows without removing work or thinking about it as a team. The work we added was often cognitively more demanding. So what do we do about it? We need to redesign work and workflows as teams with outcomes in mind. We need to engage managers. I know managers already have a lot on their plate, but when you have managers answering questions about AI for employees, mental fatigue scores are 15 percent lower. One company thought through this and started to redesign job descriptions, naming the more strategic work. They took out handoffs that were no longer needed. They created a broad upskilling plan. All of this was done in concert -- not individually one by one.第二领域：责任归属与审查松懈
其次是责任机制。各家公司正把 AI Agent 列入组织架构图，赋予它们名字和职位头衔。我们对此感到好奇，于是做了一个实验：相同的文档，相同的错误。唯一的区别在于我们告知的起草人是谁。在第一组中是 AI 工具；在第二组中是 AI “员工”；在第三组中是一名真实的员工。
对于那些本就熟悉将 AI Agent 列入组织架构图这一概念的管理者而言，当他们审核由 AI “员工”完成的工作时，个人责任感下降了 9 个百分点。他们发现的错误减少了 18%，向上呈报（escalation）的情况增加了 44%。实际上，他们对工作的审视变少了，发现的错误变少了，并且更频繁地将自己的本职工作推给他人。
英文原文
Second, accountability. Companies are putting AI agents on org charts. They're giving them names, job titles. We were curious. We ran an experiment: same document, same errors. Only difference: who we said the drafter was. In one group, an AI tool -- in one group, an AI “employee” -- in the third group, an actual employee. For managers who were already familiar with the concept of AI agents on org charts, personal accountability fell by nine percentage points when they reviewed work done by an AI "employee." They found 18 percent fewer errors. Escalations were up by 44 percent. Effectively, they scrutinized the work less, found fewer errors and gave their job to someone else more frequently.我们研究中的一家公司甚至给他们的 AI 取了名字。我们姑且叫它鲍勃（Bob）。当差错发生时，就变成了“鲍勃犯了错”。这带来了实质性的组织成本。领导者们，我们并非有意去重构责任机制。我们原本以为将 AI Agent 放到组织架构图上会让它看起来更亲切。然而，那一个小小的决定，却影响了审核工作者所感受到的责任感程度。
为了化解这一问题，Agent 系统需要人类监督者。我们需要明确一个拥有清晰决策权、上报路径并在出现差错时承担职业后果的人。我们还需要为员工提供关于如何审查 AI 内容的培训。在给你的 AI 取名字之前，请先把这件事做好。
英文原文
One company in our study even named their AI. Let's just call it Bob. When mistakes happened, Bob made the mistake. This has a real organizational cost. Leaders, we didn't mean to rewire accountability. We thought putting AI agents on org charts would make it more familiar. That small choice impacted the amount of accountability that was felt by those reviewing work. To overcome this, agentic systems need a human overseer. We need a person with clear decision rights and [an] escalation path and a career consequence if something goes wrong. We also need training for employees on how to review AI content. Do this before you name your AI.第三领域：学徒机制与新人培养
第三是学徒机制。这是我思考最多的一个问题。起草、分析、初稿、研究这些入门级工作，如今在某种程度上越来越能由 AI 来完成。然而，这种工作正是人们过去学习的途径，是我们所有人职业生涯早期都需要经历的“击球机会”。我们的研究表明，当下属使用 AI 处理具有挑战性的拔高任务时，他们能够产生不错的产出。但是，在缺乏专家支持的情况下，他们无法在底层建立起相应的判断力。这是没有学习效果的单纯产出。
每周我都至少五次被企业最高管理层问到这个问题：在 AI 时代，我该如何培养人才？
英文原文
Third, apprenticeship. This is the one I think the most about. The entry-level work, the drafting, analysis, first cuts, research, increasingly, that can be done by AI, at least somewhat. This work was how people learned. It was the at-bats that we all needed early in our careers. Our research shows when juniors use AI for stretch assignments, they're able to produce good output. However, without expert support, they don't build the judgment underneath. It's output without learning. I get asked this question every week at least five times from C-level executives: How do I build talent in the age of AI?积极应对这一挑战的企业正在改写岗位职责，使其更具轮岗性质、更加基于项目驱动。他们将初级人员与资深专家人才进行结对。他们甚至善用初级人才对 AI 的天然熟练度，把他们推向组织前沿，作为前线部署的“AI 赋能者”。
更重要的是，他们正花时间重写管理人员与领导层的胜任力要求。他们将辅导辅导、AI 技能提升、以团队为单位重组工作等要素纳入其中。也许，这正是我们投资于那种我们一直渴望的、注重人才发展的管理者的契机。
英文原文
Companies that are tackling this are rewriting job descriptions to make them more rotational, more project-based. They're pairing juniors with senior expert talent. They're even leaning on the natural AI fluency of more junior talent and putting them forth in the organization as a forward-deployed "AI enabler." Even more importantly, they're taking the time to rewrite manager and leadership competencies. They're including coaching, upskilling on AI, rewiring work as a team. Perhaps this is our opportunity to invest in the development-focused managers we've always said we wanted.领导者在 AI 重塑工作中的核心任务
注意力、责任机制、学徒机制。如果我们不够谨慎，这三个领域的工作重塑就会演变为被动承受，而不是由我们主动引领。当前领导者的核心工作，是在兼顾良好成果的前提下，融合人类与机器智能的优势来优化这一过程。
那些从 AI 中获益最多的公司已经在践行这一点：
- 他们让高管和领导团队深度参与 AI 的概率高出 12 倍；
- 专门投入时间并系统化开展 AI 学习的概率高出 4 倍；
- 重塑工作流的概率高出 2.5 倍。
英文原文
Attention, accountability, apprenticeship. These are three areas where the work redesign is happening to us instead of being led by us -- if we're not careful. The core work of a leader right now is to optimize this work with good outcomes in mind and the best of human and machine intelligence. Companies that are getting the most from AI -- they’re already doing this. They are 12 times more likely to have executives and leadership teams engaged with AI. They’re four times more likely to dedicate time to and structure AI learning. They are two and a half times more likely to reshape workflows.我开篇提到的那些相互矛盾的感受——羞耻感、暴露在外的风险、错失恐惧——其实并不必然以这种方式存在。这只是置身于这项工作转型期之中的感受。我们需要领导者明确指出正在发生的变革。我们需要领导者以创造性思维去解决这类挑战，并保持极其积极的思考。
过去几年我们一直在进行的这场关于 AI 的讨论，此前主要集中在技术本身，而现在，它终于能够回归到我们深知它终将属于的命题：一场关于人、关于变革、关于领导力的讨论。
谢谢大家。