• workplace transformation
    麦肯锡:AI赋能职场,企业如何跨越管理障碍,实现智能化未来?员工对 AI 的适应速度远超领导层的预期 AI 如何重塑职场? 人工智能(AI)正在以惊人的速度重塑职场生态,许多企业正试图利用 AI 提高生产力、优化决策流程并增强市场竞争力。然而,AI 技术的广泛应用远非一蹴而就,企业的 AI 部署不仅涉及技术升级,更考验管理者的战略眼光和执行力。 麦肯锡的《Superagency in the Workplace》 这份报告深入研究了 AI 在职场中的应用现状,基于对 3,613 名员工和 238 名 C 级高管 的调查,揭示了企业在 AI 落地过程中的机遇与挑战。报告认为,AI 在职场的变革潜力堪比蒸汽机之于工业革命,但当前的最大障碍并非技术问题,而是领导层的行动力不足。 尽管 92% 的企业计划在未来三年增加 AI 投资,但只有 1% 认为自己 AI 发展成熟,表明大多数企业仍停留在 AI 试点阶段,尚未实现全面部署。更值得注意的是,报告发现员工对 AI 的接受度远超管理层的预期,但企业的 AI 发展速度依然滞后。领导者的犹豫和执行力缺失,正成为 AI 规模化应用的最大瓶颈。 本文将从员工接受度、领导层挑战、组织架构变革、AI 治理、商业价值实现等多个维度,介绍报告的核心观点,并补充对 AI 发展的进一步思考。 一、员工比领导更快接受 AI,企业行动缓慢 报告的核心发现之一是:员工已经在积极使用 AI,而领导者仍然低估了 AI 的普及度。 数据显示: 员工使用 AI 的频率比领导层预期高出 3 倍,但许多企业尚未提供系统性培训; 70% 以上的员工认为 AI 在未来两年内将改变至少 30% 的工作内容; 94% 的员工和 99% 的高管都表示对 AI 工具有一定熟悉度,但只有 1% 的企业认为 AI 应用已成熟。 这一现象表明,AI 在企业中的主要障碍并非员工适应能力,而是管理层的滞后决策。许多企业高管仍然停留在探索 AI 价值的阶段,而员工已经在日常工作中广泛使用 AI 工具,如自动生成文档、数据分析、代码编写等。员工在推动 AI 发展方面的主动性,远远超出管理层的认知。 然而,企业未能为员工提供足够的 AI 培训和资源,导致 AI 的应用仍然停留在浅层次,难以转化为真正的生产力提升。例如,48% 的员工认为 AI 培训是 AI 规模化应用的关键,但许多公司仍未建立 AI 学习机制。企业如果不采取措施缩小这一认知鸿沟,可能会错失 AI 带来的长期竞争优势。 二、AI 领导力挑战:速度焦虑与执行落差 尽管 AI 的发展潜力巨大,但报告指出,47% 的企业高管认为公司 AI 发展过于缓慢,主要原因包括: AI 技术成本的不确定性:短期 ROI(投资回报率)难以量化,导致企业不敢大规模投资; AI 人才短缺:AI 相关技术人才供不应求,企业缺乏相应的招聘和培养体系; 监管与安全问题:企业在数据隐私、算法透明度等方面的担忧阻碍了 AI 落地。 这种“速度焦虑”让企业在 AI 发展过程中陷入试点—停滞—观望的循环: 试点阶段:部分企业已启动 AI 试点项目,如客服自动化、数据分析等; 停滞阶段:由于短期收益不确定,试点项目难以规模化推广; 观望阶段:企业倾向于等待行业先行者经验,而非主动探索 AI 的商业价值。 报告强调,AI 的落地不仅是技术问题,更是企业管理问题。领导者需要具备更强的战略决心,加快 AI 投资,并明确 AI 在企业中的角色,才能真正推动 AI 规模化应用。 三、如何实现 AI 规模化落地? 1. AI 人才培养 AI 的大规模应用依赖于系统性的 AI 人才培训。然而,报告发现,近一半的员工认为企业提供的 AI 支持有限。企业需要采取措施: 建立 AI 培训体系,涵盖 AI 基础知识、业务应用和 AI 伦理等内容; 推广 AI 试点项目,让员工亲身参与 AI 工具的开发和使用; 设立 AI 激励机制,鼓励员工利用 AI 提升工作效率。 2. 组织架构调整 AI 不能仅仅作为 IT 部门的创新项目,而应当成为企业整体战略的一部分。报告建议: 设立 AI 战略委员会,确保 AI 发展与企业长期战略保持一致; 推动 AI 在各业务部门落地,提升 AI 在实际业务流程中的应用深度; 强化 AI 风险管理,确保 AI 应用在数据安全和监管方面的合规性。 3. AI 治理:平衡速度与安全 虽然 AI 带来了极大的商业价值,但报告指出,企业在 AI 治理方面仍存在诸多挑战: 51% 的员工担心 AI 可能带来的网络安全风险; 43% 的员工关注 AI 可能导致的数据泄露; 企业需要建立 AI 伦理标准,确保 AI 透明、公正、合规。 四、AI 时代的商业价值:企业如何真正实现 ROI? 尽管企业对 AI 充满期待,但报告显示,目前仅 19% 的企业 AI 投资带来了 5% 以上的收入增长,表明大多数企业的 AI 应用尚未转化为可观的商业回报。为了提升 AI 价值,企业需要: 从“技术驱动”转向“业务驱动”,确保 AI 应用直接创造商业价值; 优化 AI 目标设定,明确 AI 在核心业务中的定位; 加强 AI 应用场景探索,特别是在客户服务、供应链管理等高回报领域进行深入部署。 AI 成败的关键在于管理层 AI 的成功不仅依赖技术本身,更取决于企业领导者的执行力和战略眼光。企业若要真正迈向 AI 时代,需要: 加速 AI 战略落地,推动组织变革; 加强 AI 人才培养,提高员工 AI 适应能力; 建立 AI 治理体系,确保 AI 安全合规发展。 在 AI 时代,最危险的不是迈得太快,而是思考得太小、行动得太慢。 附录:《Superagency in the Workplace》 下载
    workplace transformation
    2025年03月14日
  • workplace transformation
    Josh Bersin:Digital Twins, Digital Employees, And Agents Everywhere 2025 年,数字员工和人工智能助理的崛起将彻底改变人力资源运营,改变招聘、数据分析和员工管理等任务。 这些技术包括数字双胞胎和智能代理,它们将与人类专业人员一起工作,以提高生产力和优化工作流程。 随着人工智能工具成为日常业务不可或缺的一部分,人力资源领导者必须拥抱这些创新,同时继续关注技能培训、心理健康和包容性工作环境。 向人工智能的转变还将重塑团队动态,这对人力资源部门重新设计角色和流程以保持竞争力提出了挑战。 I recently heard Elon Musk predict that every citizen would have multiple Optimus robots in their homes within five years. And while I often ignore his predictions because they’re exaggerated, I think he’s on to something. We are about to witness an explosion of Digital Employees in our companies, and these may be the “robots” we’ve heard about for years. Let me explain. This week I talked with dozens of vendors and clients at Unleash and then visited our development partner Sana Labs in Stockholm. It’s now clear that we’re going to be working with multitudes of “digital employees” in the year ahead. (And as Dario Amodei, the founder of Anthropic explains, AI can do many more positive things in business, science, and health than we ever imagined.) By “digital employee” I mean a software powered agent that can talk with us, answer detailed questions, solve complex analytic problems, and navigate a multitude of systems. ChatGPT and its peers, which introduced the idea of an agent, has now spawned dozens of “agentic” use cases, which I’d be willing to refer to as personalities. Let me start with a “Digital Twin.” Imagine you have a superb customer service agent with years of experience helping your most demanding clients. If you load the last five years of their emails, coupled with all their internal documentation, and a log history of their last two years of service calls, you can essentially “create him or her” digitally with all the knowledge, style, and internal contacts this person has developed. This twin, which may look initially like an AI assistant, could then carry on this employee’s work when the real life worker is on vacation. One of our clients, a large insurance company, has already built “digital twins” for claims processing. If you think about the complexity and workflow of processing a claim, much of it could be learned by an agent, making the “claims robot” an expert on this important process. And as you change claims rules and limits, the agent will learn new guidelines in only seconds. Our AI assistant Galileo, a trained expert on HR (Galileo is trained on 25 years of research and thousands of conversations with clients and vendors), is essentially a “digital twin” of me and the other analysts in our firm. I’m not saying Galileo is as fun to talk with as we are, but I can assure you that he (or she) is as knowledgeable and supportive. And Galileo is even smarter than I am: he has instant knowledge of skills models, compensation benchmarks, turnover statistics, and other data bases which I can only access by looking them up on demand. And using the Sana platform we can configure Galileo to have multiple personalities. Galileo the “Recruitment Agent” might have in-depth knowledge of screening, interviewing, and candidate skills assessment and he may have direct linkage to SeekOut, Eightfold, or any other sourcing applications. In his candidate facing personality he may be able to answer candidate questions, explain shift schedules, and “sell” the company to top job candidates. (This is what Paradox has done for years and vendors like Eightfold and LinkedIn are launching now.) But there’s more. Imagine that this “digital twin” or “digital employee” has intimate knowledge of Workday, SuccessFactors, or a variety of other systems. Now the assistant can not only answer questions and help solve problems, he can also process transactions, look things up, and run reports against multiple system. The digital employee has turned into a “digital analyst,” who can find things and do work for you, saving you hours of effort in your daily life. (Vee from Visier is designed for this.) Suppose you ask your digital friend to attend meetings for you, participate in conversations on certain topics, and alert you in real time when urgent issues come up for discussion. He could help you scale your time, keep you informed about decisions you need to know about, and help you manage your action items. And the list goes on and on and on. Best of all, what if your digital twin can talk to you. Suppose he “checks in” with you about the project you asked for help with last week, so you inform him how things are going and he gets “smarter” about what you may need next. Galileo does this today, prompting you to dig into a problem and explore areas you may not have considered. And if you ask him about management or people issues, he could give you advice and coaching, based on the leadership models or even CEO interviews in your own company. (BetterUp, Valence and others are working on this.) This is not science fiction, my friends. All this is becoming reality and will certainly be common next year. Every vendor has a slightly different focus. The Microsoft Copilot specializes in MS Office-related activities, ServiceNow’s focuses on internal service and support, Galileo is focused on the needs of HR, and Joule is an expert on all the functions of SAP. Each of these “digital employees” needs training, feedback, and connections to stay current and relevant. So it’s doubtful that one digital employee will do everything. (Training a digital employee means managing his or her corpus of information, which will be a major new role in HR.) One thing is very clear: we are going to be living and working with these guys. And as we use them and see what they’re capable of, we’re going to redesign work. Little by little we’ll offload tasks, projects, and workflows. And as we do, we’ll get smarter and smarter about redesigning our teams. I liken the process to that of a carpenter who gets a new multi-function power drill. Before the drill he may have manually drilled holes, carefully selecting the drill bit and the level of pressure based on wood density. Now he drills holes faster, more accurately, and with more precision. Soon he just speeds through the process, spending more time on cabinet fit, finish, or design. The same thing will happen to our HR tasks, projects, and designs. And these new digital employees are programmable! So once we figure out what they’re capable of we can adjust them, customize them, and connect them together. Eventually we’ll have intelligent assistants that operate as entire applications. And that’s the threat to incumbent software companies – the agents hollow out many of our existing applications. How Do Our Digital Employees Impact Our Own Work? One more observation. Many a few of the clients I talked with kept asking “what about our softskills?”  What work is truly human? I think that’s the wrong question. Rather we should ask the opposite: how much can I delegate to my new friends as fast as possible! Have you been upset that your vacuum cleaner took away the rewarding human work of sweeping a floor? How much joy do you get from washing dishes? Did your dishwasher make you feel deflated when you stopped splashing around in the soapy water? Of course not – these tools eliminated tasks we considered to be “drudgery.” Well today, thanks to digital assistants, creating a pivot table to do cross-tab analysis has become drudgery. You can stop getting your hands wet with that task – ask Galileo or Copilot to analyze the data, and then ask him to chart it, add more data, and try new assumptions. The more we learn to use these new digital employees the more “drudgery” we can stop doing. And consider complex “human-centered” activities like “change management.” A client asked me “how could Galileo help me with change management for our new HCM system?” I answered her with dozens of ideas: ask Galileo for case studies of other companies and have it build a checklist to consider based on what other companies did. Then ask Galileo to build a training plan; ask it to read the user documentation and create a table of what features are new; then ask Galileo to rewrite that change plan by role. And finally ask Galileo to write a press release about success, craft some compelling communications to employees, and ask it to compute the ROI of all the steps eliminated. These are all “manual” human tasks we do today and they take time and ingenuity to figure out. If you went through this process in Galileo you could ask your digital employee to save these steps and prompts in a “template,” and you have just taught your digital employee how to do change management. The next time you need him he can step you through the process. As I started to explain this to my client I stopped and said: wait a minute. I can’t possibly show you everything Galileo can do. You have to try it for yourself. And that’s my big message. Don’t wait for a vendor to drop a finished solution in your lap. These are intelligent, trainable, digital experts. You have to get to know them so you can figure out where they fit in your job, your projects, and your company. Just like you do with any new hire. I say it’s time to get started. No more sweeping floors or washing dishes by hand. Let’s meet our digital employees, tell them about our projects, and ask for their help. Step by step, day by day, we can redesign our jobs to be more more productive, liberating us to do greater things.
    workplace transformation
    2024年10月19日