近期与人工智能相关的裁员在数据上仍然有限，但在话术上却十分醒目，惠普计划在三年内裁减约5,000个岗位，而今年10月美国裁员公告中约有五分之一将原因归于AI。许多公司将缩减人员包装为技术进步而非疫情期间招聘过度，尽管当前关于劳动力市场大规模被AI重塑的证据仍然薄弱。

人工智能的采用在计算机编程和客户服务领域最为深入：约三分之二的程序员每周都会使用AI工具，GitHub Copilot拥有2,600万用户，而Anthropic聊天机器人约三分之一的查询与编程相关。客户服务方面，85%的经理计划在今年试验AI，自从2022年ChatGPT推出以来，客户服务外包商Teleperformance的股价已下跌约四分之三，显示投资者对即将到来的颠覆抱有强烈预期。

这些岗位之所以易被自动化，是因为其任务重复、对公司背景依赖较少且结果容易验证，同时又产生大量可用于训练的代码库、通话记录和论坛“点赞”等数据。美国约有300万客户服务人员和200万软件开发者，两者作为最常见的非体力岗位之一，被AI公司视为巨大的目标市场，而随着AI使用成本下降和企业数据被整理得更干净，类似工具很可能扩展到初级投行、律师和其他白领岗位。

Recent AI-linked layoffs remain modest in data but striking in rhetoric, with HP planning around 5,000 job cuts over three years and AI cited in roughly one-fifth of American layoff announcements in October. Many firms frame reductions as technological progress rather than pandemic over-hiring, even though current evidence of large-scale labour-market displacement is still limited.

AI adoption is deepest in computer programming and customer service, where around two-thirds of coders now use AI tools weekly, GitHub Copilot has 26 million users, and about one-third of Anthropic chatbot queries involve coding. In customer service, 85% of managers aim to experiment with AI, while Teleperformance’s share price has fallen by roughly three-quarters since ChatGPT’s 2022 launch, signalling investors’ expectations of disruption.

These occupations are ripe for automation because their tasks are repetitive, context-light and easily verifiable, and because they generate abundant training data from code repositories, transcripts and forum upvotes. With about 3 million U.S. customer-service workers and 2 million software developers—together among the five most common non-manual jobs—AI firms see a huge target, and falling AI costs plus cleaner corporate data are likely to spread similar tools to junior bankers, lawyers and other white-collar roles.

Source: Lessons from the frontiers of AI adoption

Subtitle: Why coders and call-centre agents are especially vulnerable to the technology

Dateline: 12月 04, 2025 04:50 上午