The AI Jobs Transition Framework for the EU
Author: Alex Martin Richmond Date: June 2026 Organization: OpenAI Economic Research Sign-off: Ronnie Chatterji, OpenAI Chief Economist
中文摘要
OpenAI 经济研究部发布 AI 就业转型框架的欧盟版本，将 2609 个 ESCO 职业按三条维度（技术暴露度 + 人类必要性 + 需求弹性）映射为四条转型路径：
| 路径 | 欧盟占比 | 美国占比 | 含义 |
|---|---|---|---|
| 更高自动化潜力 | 14% | 18% | AI 可直接替代大部分任务 |
| 将重组 | 27% | 24% | AI 改变工作流但不消灭岗位 |
| 可能随 AI 增长 | 12% | 12% | 成本下降扩大需求 |
| 短期变化较小 | 47% | 46% | 物理/监管/关系壁垒限制替代 |
欧盟特色：比美国更小的自动化份额、更大的重组份额，反映欧洲制造业/护理/教育/公共服务的职业结构差异。
人类必要性分类（为什么人仍然不可替代）：49% 物理性、28% 监管/问责、9% 关系性、14% 无明确障碍。
核心洞见：技术能力本身不决定劳动力市场结果——制度（许可、公共预算、法定服务义务、多语言要求、人际信任）决定了技术能力如何转化为实际就业变化。
Full Text
[18-page PDF extracted in full — content archived at /tmp/ai_jobs_eu.txt]
From the Desk of the Chief Economist
Europe is entering a new phase of the AI transition. Across workplaces, households, firms, and public institutions, AI is beginning to change how people access information, make decisions, and get work done...
Key takeaways:
- EU: 12% growth, 14% higher automation, 27% reorganize, 47% less immediate change
- Country variation: Luxembourg/Sweden/Netherlands have largest growth shares; Germany/Greece/Italy have largest automation shares
- Demand elasticity: lower costs may increase demand where work is digitally deliverable/discretionary/project-based
- Human involvement: 49% physical, 28% regulatory/accountability, 9% relational, 14% residual
Framework: Three questions
- Technical exposure: how extensively can AI perform the occupation's tasks?
- Human necessity: do regulation, accountability, physical presence, or relationships keep a worker at the center?
- Demand elasticity: do lower costs expand quantity enough to offset labor-saving gains?
Country transition shares (full table)
See extracted text for complete 27 Member State breakdown.
Policy implications
- AI Observatory: live labor-market intelligence connecting occupation, skills, vacancy, wage, training data
- National readiness plans linking AI adoption, skills, public employment services, social dialogue
- Universal AI literacy
- Portable skills, work-compatible training, career guidance
Methodology
- ESCO-O*NET crosswalk (4253 links across 2609 ESCO occupations)
- Europe-tailored human necessity and demand elasticity estimates
- Model-based estimates, not observed causal data
- Country differences reflect occupational composition, not readiness ranking