How OpenAI’s Chief Economist Quantifies AI’s Impact on European Jobs

OpenAI’s economics team, led by chief economist Ronnie Chatterji, presented an EU AI‑employment transformation framework that evaluates 2,609 occupations across three practical dimensions: technical exposure (how much of a job AI can perform), human irreplaceability (legal, accountability, on‑site, trust factors), and demand elasticity (whether lower costs will expand demand). The intersection of these dimensions places each occupation into one of four buckets: high automation potential, likely to be reorganized, likely to grow with AI, or little short‑term change.

The report finds that 47% of EU jobs fall into the “little short‑term change” bucket, 27% are “likely to be reorganized,” 14% face “high automation potential,” and 12% may “grow with AI.” Compared with the United States (18%, 24%, 12%, 46% respectively), Europe shows a lower share of high‑automation jobs but a higher share of reorganized roles, reflecting Europe’s larger share of manufacturing, skilled trades, transport, care, education, and public‑service jobs that are physically rooted or heavily regulated.

Four categories explain why certain jobs remain human‑centric: physical presence (49% of such jobs, e.g., bedside nursing), regulatory/accountability (28%, e.g., lawyers, customs officers), relational trust (9%, e.g., teachers, crisis social workers), and residual roles (14%). Examples illustrate that a middle‑school language teacher can use AI for lesson planning but still needs a real person for classroom interaction, while a general‑practice nurse must be on‑site for patient safety.

The analysis also incorporates demand‑price elasticity, with a median elasticity of 0.7 (a 10% price drop yields ~7% demand increase). Some occupations, such as travel‑agency consultants and renewable‑energy advisors, have elasticity above 1, meaning cost reductions could expand demand, whereas roles tied to fixed schedules or public budgets (e.g., midwives) show low elasticity.

Country‑level results vary widely. Luxembourg, Sweden, and the Netherlands have the highest “grow with AI” shares (up to 22% in Luxembourg), while Germany, Greece, and Italy lead in “high automation potential” (up to 17% in Germany). The share of “little short‑term change” ranges from 59% in Romania to 25% in Luxembourg, underscoring that occupational structure, not readiness, drives differences.

Policy recommendations include centering workers in AI strategies, building AI observation stations to monitor skill, vacancy, wage, and training data, allowing each member state to craft its own employment‑readiness plan, and promoting universal AI literacy.

In summary, the report does not predict massive job loss; instead, it shows that technology is only the starting point, with outcomes shaped by on‑site requirements, regulatory responsibilities, trust‑based interactions, and cost‑driven demand shifts.