AI正在改变人们从事的工作。通过对美国ChatGPT用户超过80万条信息的分析，我们的新研究表明，16.8%与工作相关的信息以及43.5%与特定职业相关的信息，涉及的是与其他职业相关联的任务。
小企业主可以独立起草文案、审阅合同或进行基础财务分析。销售人员可以利用AI探索客户数据集，而这些数据过去可能需交由分析师处理。营销人员无需等待开发人员即可自行排查网站问题。在每种情况下，AI不仅改变了工作的完成方式，也改变了谁负责做什么。
我们的新报告《前沿工作：AI如何拓展人们的工作内容》研究了这一转变。我们将由此产生的模式称为任务交叉：历史上与某一职业相关的工作，出现在另一职业人群的AI使用中。在我们的_AI职业转型框架_(在新窗口中打开)中，我们认为许多工作可能会重组：这些工作的日常任务可能发生重大变化。本报告是我们新系列《前沿工作》的第一篇，该系列旨在探索AI如何实时改变工作方式。
借助我们观察工作世界如何变化的独特视角，我们将基于证据定期提供数据驱动的见解，以指导政策与实践。
近半数特定职业的AI使用跨越了职业边界
许多关于AI与工作的研究，首先会列出与特定职业相关的固定任务清单，然后询问模型能否完成这些任务。我们的证据表明，AI也在改变谁承担哪些任务。
为了衡量任务交叉，我们首先将与某一职业相关的工作与跨多个职业普遍存在的工作区分开来。某些活动，如写作、总结和日程安排，在各职业中广泛共享，不能作为交叉的证据。我们将这些归类为通用任务。对于其余信息，我们判断任务是否属于用户自身职业范畴。在非通用信息中，43.5%超出了用户自身职业，这为我们提供了一个早期窗口，观察AI如何在任务内容变化反映到职位描述或头衔之前，就已重塑工作的任务构成。
这表明，相当一部分与工作相关的ChatGPT使用，源于用户正在拓展自身角色。这种模式在几个群体中尤为明显。排除通用工作后，外部职业任务占比为：
- 客户体验工作者中，77%的特定职业信息涉及外部任务
- 设计师中为75%
- 人力资源工作者中为69%
- 法律工作者中为56%
- 营销人员中为53%
这些职业正在从其他角色“借用”任务。证据指向了分工的变化：一些过去需要交接的活动，现在可以由最初遇到需求的人自行完成。
某些任务比其他任务传播得更远
下方的热力图通过询问每条非通用信息最接近哪个职业的工作，展示了任务交叉情况。营销和工程任务传播得最远，频繁出现在这些领域之外的工作者信息中。
营销和工程任务跨越多个职业传播
营销和工程任务跨越多个职业传播
| 传统任务来源 |
|---|
| 工作者职业 |
| 客户体验 |
| 设计 |
| 工程 |
| 财务 |
| 人力资源 |
| 法律 |
| 营销 |
| 销售 |
职业内任务占比
当我们从职业层面聚焦到具体任务时，任务交叉并非均匀分布。财务计算和技术故障排除，在分析中所有其他七个职业群体中，均位列最常见的三大外部任务之中。营销工作也广泛传播：创建营销材料出现在其他五个群体中，在设计用户中尤为突出。
观察更广泛的任务组合，可以发现两种不同的交叉方向。一些工作从其他职业引入了大量任务。另一些工作则提供了出现在许多不同职业中的任务。
设计体现了第一种模式。设计师约35.2%的信息涉及通常与其他职业相关的工作，而设计任务仅占其他领域工作者信息的1.7%。设计师大量依赖外部任务，但设计工作本身很少出现在其他地方。
工程则更接近相反情况。仅18.5%的工程信息涉及其他领域的任务，但工程任务占其他职业工作者信息的7.4%。从软件故障排除到处理技术系统，工程是其他领域人员承担工作的重要来源。
营销在两方面都表现突出。营销人员24.3%的信息涉及与其他职业相关的任务，而营销任务占其他领域工作者信息的8.9%——这是样本中向外输出的最高比例。营销工作者结合了多个领域的任务，同时营销工作也在组织内广泛传播。
小企业中任务交叉更多
企业的规模和结构影响着AI如何改变工作。在大公司中，员工可能拥有专业团队、既定工作流程和内部服务。在较小的组织中，最接近问题的工作者更可能亲自处理问题，而非委派他人。
在普通用户中，外部职业任务占比从2-5个席位工作空间用户的18.9%，下降到超过100个席位工作空间用户的16.3%。在重度用户中，我们并未观察到同样的单调递减模式。
一种可能的解释是，较小组织中的中等用户，在遇到原本需要其他职能才能完成的工作时，会求助于AI。而重度用户可能已经形成了稳定的AI支持工作流程，这些流程在不同组织间看起来更相似，或者他们在核心职业内更密集地使用AI。
在专业资源稀缺的地方，AI作为通用工具可能尤其有用。
任务清单本身正在变化
像这样的AI使用数据，是工作正在发生转变的指标。它使我们能够看到，在企业重写职位描述或创建新职位头衔之前，AI如何让工作者尝试新的活动组合。从这个意义上说，使用模式可能提供了职业变化的早期信号，而传统的劳动力市场统计数据将在之后才能捕捉到这些变化。
AI changes the work that people do. In an analysis of more than 800,000 messages from U.S. ChatGPT users, our new research suggests that 16.8% of work-related messages and 43.5% of occupation-specific messages are about tasks associated with another occupation.
A small-business owner can independently draft copy, review a contract, or perform basic financial analysis. A salesperson can use AI to explore a customer dataset that might once have gone to an analyst. A marketer can troubleshoot a website without waiting for a developer. In each case, AI changes not just how work gets done, but who does what.
Our new report, Work at the Frontier: How AI is Expanding What People Do at Work, studies this shift. We call the resulting pattern task crossover: work historically associated with one occupation appearing in the AI use of people in another. In our _AI Jobs Transition Framework_(opens in a new window), we argue that many jobs are likely to reorganize: these are jobs whose day-to-day tasks could change substantially. This report is the first in our new Work at the Frontier series, which explores how AI is changing work in real time.
Using our unique window into how the world of work is changing, we will offer regular data-driven insights based on evidence to guide policy and practice.
Nearly half of occupation-specific AI use crosses job boundaries
Many studies of AI and work begin with a fixed list of tasks associated with a given occupation and ask whether models can perform them. Our evidence suggests that AI is also changing who takes on which tasks.
To measure task crossover, we first separate work that is associated with an occupation from work that appears across many jobs. Some activities, such as writing, summarizing, and scheduling, are shared too broadly across occupations to be evidence of crossover. We classify these as generic. For the remaining messages, we ask whether the task falls inside or outside the user’s own occupation. Among non-generic messages, 43.5% fall outside of the user’s occupation, offering an early window into how AI may be reshaping the task content of jobs before those changes appear in job descriptions or titles.
This suggests that a substantial part of work-related ChatGPT use is from users expanding their role. The pattern is especially pronounced in several groups. Once generic work is excluded, outside-occupation tasks account for:
- 77% of occupation-specific messages from customer experience workers
- 75% from designers
- 69% from human resources workers
- 56% from legal workers
- 53% from marketers
These occupations are “borrowing” these tasks from other roles. The evidence points to a changing division of work: some activities that once required a handoff can now be done by the person who first encounters the need.
Some tasks travel farther than others
The heatmap below shows task crossover by asking which occupation’s job each non-generic message most closely resembles. Marketing and engineering tasks travel farthest, frequently showing up in messages from workers outside those fields.
Marketing and engineering tasks travel across many occupations
Marketing and engineering tasks travel across many occupations
| Traditional task source |
|---|
| Worker occupation |
| Customer experience |
| Design |
| Engineering |
| Finance |
| Human resources |
| Legal |
| Marketing |
| Sales |
Share of tasks within occupation
Task crossover is not evenly distributed when we zoom in from occupations to specific tasks. Financial calculation and technology troubleshooting each appear among the three most common outside tasks in all seven other occupation groups in the analysis. Marketing work also travels broadly: creating marketing materials appears across five other groups and is especially prominent among design users.
Looking at broader task bundles reveals two distinct directions of crossover. Some jobs bring in many tasks from other occupations. Others supply tasks that appear across many different jobs.
Design illustrates the first pattern. About 35.2% of messages from designers involve work usually associated with another occupation, while design tasks account for only 1.7% of messages from workers in other fields. Designers draw heavily on outside tasks, but design work itself rarely appears elsewhere.
Engineering is closer to the reverse. Only 18.5% of engineering messages involve tasks from other fields, but engineering tasks account for 7.4% of messages among workers in other occupations. Engineering is an important source of work that people elsewhere take on, from troubleshooting software to working with technical systems.
Marketing stands out in both directions. Marketers devote 24.3% of their messages to tasks associated with other occupations, while marketing tasks account for 8.9% of messages among workers in other fields—the highest outward share in the sample. Marketing workers combine tasks from multiple domains, and marketing work also spreads widely across the organization.
More task crossover in small businesses
The size and structure of a business shapes how AI changes work. In a large company, employees may have access to specialized teams, established workflows, and internal services. In a smaller organization, the worker closest to the problem is more likely to take on the problem rather than delegate.
Among average users, the outside-occupation task share falls from 18.9% for users in workspaces with 2–5 seats to 16.3% for users in workspaces with over 100 seats. Among the heaviest users, we do not see the same monotonic pattern.
One possible explanation is that moderate users in smaller organizations turn to AI when they encounter work that would otherwise require another function. Heavy users may instead have developed stable AI-supported workflows that look more similar across organizations, or may use AI more intensively within their core occupation.
AI may be especially useful as a generalist tool where specialist resources are scarce.
The task list itself is changing
AI usage data like this is an indicator of where work is shifting. It allows us to see how AI lets workers experiment with new combinations of activities before firms rewrite job descriptions or create new job titles. In that sense, usage patterns may provide an early signal of occupational change that conventional labor-market statistics will capture only later.
本文内容采集自官方网站，排版和翻译可能与原页面存在差异。
阅读官方全文