OECD: AI and the Labour Market — Automation Risk Across Member Countries

The OECD has been building the most detailed cross-country framework for understanding AI's labor market impact since 2018. Their ongoing research program — spanning employment surveys, occupation-level AI exposure scoring, and policy analysis across 38 member countries — provides the empirical backbone for much of the global AI-and-jobs debate.

Key Findings

• 27% of jobs in OECD countries are in occupations at high risk of automation. These are roles where the majority of tasks could technically be performed or significantly altered by AI and automation technologies. The share ranges from under 20% in some Nordic countries to over 35% in Eastern European member states.
• AI exposure is highest in finance, manufacturing, and agriculture. White-collar roles in financial services and certain manufacturing processes face the steepest exposure, while healthcare and education show more mixed profiles — high augmentation potential but low full-automation risk.
• Workers are worried — and their concerns correlate with exposure. OECD surveys show that 3 in 5 workers in high-exposure occupations express concern about job security related to AI, compared to 1 in 5 in low-exposure roles. Despite this, fewer than 30% of worried workers have taken concrete upskilling steps.
• Younger and less-educated workers face compounding risks. Automation risk is not evenly distributed by demographics. Workers under 30 in routine-task occupations face both higher exposure and fewer resources for transition, especially in countries without robust reskilling infrastructure.
• Policy response varies dramatically. Only 12 of 38 OECD members have dedicated AI workforce transition programs. The rest rely on general employment or education policy, which the OECD finds insufficient for the speed of AI-driven change.

What This Means for Your Career

The OECD data gives you a benchmark. If you work in one of the 38 member countries, your occupation has been scored for AI exposure — and the framework accounts for your country's specific economic structure, not just a generic global average. That matters because a financial analyst in Germany faces a different risk landscape than one in Mexico, even though the core tasks overlap.

The most actionable finding is the gap between concern and action. Most workers sense that AI is changing their field, but fewer than a third are doing something about it. If you are actively building AI-adjacent skills — learning to work with AI tools, understanding how they apply to your specific tasks — you are already in the minority. That gap is your competitive advantage.

The policy variation also matters. If your country has strong reskilling programs (Nordics, Canada, parts of Western Europe), you have a safety net. If not, the burden of adaptation falls more heavily on you. Know what your government is — and isn't — providing.

Data Highlights

• 27% of OECD jobs are in occupations at high automation risk
• 60% of workers in high-exposure roles worry about AI and job security
• Less than 30% of concerned workers have taken concrete upskilling steps
• 12 of 38 OECD countries have dedicated AI workforce transition programs
• 3x variation in automation risk rates across OECD member states

Methodology

The OECD's AI automation risk estimates combine multiple data sources: the OECD Survey of Adult Skills (PIAAC), the OECD AI Surveys of Employers and Workers, and occupation-level task analysis using the ISCO framework. Each occupation is scored based on the share of tasks that are technically automatable with current or near-term AI, weighted by the prevalence of those occupations in each country's labor market. Country-specific estimates account for differences in industrial composition, workforce demographics, and existing automation adoption levels.