AI就业转型框架：绘制人工智能对就业的短期影响

The AI jobstransition frameworkThe Al jobstransition framework Mapping AI’s near-term impact on jobsMapping Al's near-term impact on jobs Alex Martin RichmondAlex Martin Richmond Note from theChief EconomistNote from theChief Economist AI is transforming our lives. It’s helping us completeimportant tasks, reshaping how we work, andexpanding what individuals and organizations can do. Al is transforming our lives. It's helping us completeimportant tasks, reshaping how we work, andexpanding what individuals and organizations can do. Those capabilities also create important questions for the broader economyand job market: what will AI’s labor market impact be? Where will the impactsbe felt most and when? And how can we ensure that the AI transition worksfor everyone? While AI capabilities are advancing very quickly, businesses,institutions, and labor markets take time to adjust. That lag means we mustavoid two kinds of error: overstating immediate disruption and understatinglong-run impact. Those capabilities also create important questions for the broader economyand job market: what will Al's labor market impact be? Where will the impactsbefelt most and when?And how can we ensurethat the Altransition worksinstitutions, and labor markets take time to adjust.That lag means we mustavoid two kinds of error: overstating immediate disruption and understatinglong-run impact. That is why we see this transition as a force we can shape, not simply observe.Our role is to identify where economic changes are already occurring anddisseminate this data widely, so workers, firms, and policymakers can act with thebest possible information. We're working to address this gap through sharing whatwe're seeing and informing a wide set of stakeholders throughreleasing data,publishing policy ideas like ourIndustrial Policy for the Intelligence Ageto jumpstart conversation and debate, and continuing to develop and deploy moreadvanced models that are easier to use so more people — not just a select few —can participate in the AI economy.That is why we see this transition as a force we can shape, not simply observe.Our role is to identify where economic changes are already occurring anddisseminate this data widely, so workers, firms, and policymakers can act with thebest possible information.We're working to address this gap through sharing whatwe're seeing and informing a wide set of stakeholders through releasing data,publishing policy ideas like our Industrial Policy for the Intelligence Age to jumpstart conversationanddebate,andcontinuingtodevelopanddeploymorecan participate in the Al economy. Since I started in this role in 2024, AI capabilities have increased atunprecedented speed, and early adopters are already seeing meaningfulproductivity gains. Several studies have compared the tasks we do at work towhat AI can do today and suggest immediate and widespread economic impact.At the same time, the aggregate labor market impact remains uneven andsometimes contradictory, reflecting the typical frictions that slow technologicaladoption. But we cannot be complacent. Given how quickly AI capabilities areadvancing, there is still potential for broader disruption.Since I started in this role in 2024, Al capabilities have increased atunprecedented speed, and early adopters are already seeing meaningfulproductivity gains. Several studies have compared the tasks we do at work towhat Al can do today and suggest immediate and widespread economic impact.At the same time,the aggregate labor market impact remains uneven andsometimes contradictory, reflecting the typical frictions that slowtechnologicaladoption.But we cannot be complacent. Given how quickly Al capabilities areadvancing,thereis still potential forbroader disruption. With increasingly capable models on the horizon, we want to help people andorganizations understand this transition and plan for it. That includes developingnew and more nuanced approaches to estimating the impact of AI on jobs. In thisreport, we propose a simple framework that moves beyond the traditional AIexposure measures and tests it against what we see in data from the world’slargest AI platform, ChatGPT. We combine the following elements: where AI hascapability, where humans remain essential, how demand responds to lower costs,and where AI is actually being used today. Together, these factors provide aclearer picture of which jobs are most likely to face pressure first, which are morelikely to be redesigned, and which may grow.With increasingly capable models on the horizon, we want to help people andorganizations understand this transition and plan for it. That includes developingnew and more nuanced approaches to estimating the impact of Al onjobs.In thisreport, we propose a simple framework that moves beyond the traditional Alexposure measures and tests it against what we see in data from the world'slargest Alplatform,ChatGPT.We combinethefollowing elements:whereAl hascapability,wherehumans remain ess