传统公司层级常被比作金字塔，但若 AI 让每一层所需人数显著减少，组织形状可能从“方尖碑”甚至走向“一个点”，而近期最受担忧的是“菱形”：若入门与初级岗位被不成比例地冲击，白领组织会在顶部与底部变窄、在中层变宽。现有数据仍不确定，但这种几何变化的风险集中在早期职业岗位被压缩所带来的结构性空心化。
证据迄今并不一致：斯坦福大学 Erik Brynjolfsson 等人在去年末发表的论文显示，美国软件工程与客户服务领域的 22 至 25 岁人群就业出现大幅下降；伦敦国王学院 Bouke Klein Teeselink 的近期研究则表明，英国中更受 AI 影响的企业更可能减少入门级招聘。与此同时，匹兹堡大学 Morgan Frank 等人的新论文警告称，AI 暴露行业的毕业生前景在 2022 年 ChatGPT 发布之前就已在恶化，提示下行趋势可能早于此轮 AI 冲击，而且在不确定时期年轻人历来更容易受伤。
管理者仍有三个理由不应简单削减入门岗位：无人能确定 AI 的长期影响，没有未来人才管道的风险可能高于培训流失的成本，而仅靠自然减员也不利于打造 AI 素养劳动力。根据 OpenAI，18 至 29 岁员工在工作中使用 ChatGPT 的可能性超过 50 岁以上人群的两倍，因此更优做法是重塑初级岗位内容，把文书性杂活交给机器，让培训生把时间转向更高阶任务并通过轮岗（例如创新或法律科技团队，周期以“数月”计）为进一步变化做准备。
The classic company hierarchy is a pyramid, but if AI sharply reduces headcount per layer it may shift toward an obelisk or even a dot, and the biggest near-term fear is a diamond: if entry-level and junior roles are hit disproportionately, white-collar firms become narrower at the top and bottom and wider in the middle. The numerical risk concentrates on early-career jobs being compressed enough to reshape the whole distribution of roles.
Evidence remains mixed: a paper published late last year by Erik Brynjolfsson of Stanford University and co-authors reports big employment drops in America for 22- to 25-year-olds in software engineering and customer service, and a recent study by Bouke Klein Teeselink of King’s College London finds that more AI-exposed British firms are more likely to cut entry-level hiring. A new paper by Morgan Frank of the University of Pittsburgh and co-authors cautions that graduate prospects in AI-exposed industries were worsening before ChatGPT’s 2022 launch, implying the trend may predate the recent AI shock amid broader economic uncertainty.
Managers still have three reasons not to simply slash entry-level jobs: AI’s long-run effects are unknown, losing a talent pipeline can be riskier than training attrition, and attrition is a poor route to an AI-literate workforce. OpenAI reports that 18- to 29-year-olds are more than twice as likely to use ChatGPT at work as those over 50, so the better response is to redesign junior work—offloading document-heavy grunt tasks to machines and shifting trainees toward higher-order work and multi-month rotations (for example in innovation or legal-tech teams) to prepare for larger changes ahead.
Source: How big a threat is AI to entry-level jobs?
Subtitle: And what should companies do in response?
Dateline: 1月 29, 2026 04:18 上午