Where will GenAI work exposure land first? OECD data point to capital and highly educated regions
Across 29 OECD countries, 52% of employment in regions with the highest tertiary-education shares was classed as highly exposed to generative AI, versus 42% elsewhere. Exposure means task acceleration is plausible—not that jobs will disappear.
The short version
- What happened
- In 29 OECD countries, highly educated regions averaged 52% of employment in occupations classed as highly exposed to GenAI, compared with 42% in other regions.
- Who it affects
- Professional workers in capital and highly educated regions are more likely to have tasks language models could accelerate, but exposure does not determine job loss.
- What remains unproven
- The exposure indicator is task-based, not observed workplace deployment or productivity.
Keep in mind
The exposure indicator is task-based, not observed workplace deployment or productivity.
A 50% task threshold classifies whole occupations and hides variation.
Firm adoption covers European regions and AI broadly; exposure spans 29 OECD countries and GenAI potential.
The report is descriptive and does not establish causation.
Source analysis · 1 linked source. Primary analysed source date: 4 November 2026.
Keep following the evidence
Join The Impact Brief founding list for selected AI reports, their effect on people and the evidence limits. Read a sample first.
Read the full report · 7 min The evidence, methods, implications and limitations
The direct finding: potential exposure concentrates where advanced skills already cluster
The OECD estimates that 52% of employment in regions with the highest national shares of tertiary-educated residents is in occupations highly exposed to generative AI, compared with 42% in other regions. The comparison covers large regions across 29 OECD countries. Capital-city regions average 63.6%, with Greater London at 74.1%, Warsaw at 70.7% and Prague at 69.2%. The pattern points to metropolitan and administrative centres as places where AI-assisted task change may arrive earlier and affect a larger share of work.
The finding reverses a familiar automation story. Earlier automation often concentrated risk in routine physical or clerical work. Large language models are relatively capable at text-heavy, non-routine tasks common in professional occupations, so highly educated regions appear more exposed. Exposure is not redundancy. The indicator asks whether AI could potentially cut the time required for occupational tasks; it does not predict whether an employer will adopt a tool, reorganise work, raise output, reduce headcount or create new tasks.[1]
How tasks become a regional employment share
The report combines occupation-level assessments derived from task research with labour-force surveys and regional employment-by-occupation tables. An occupation is labelled highly exposed when more than 50% of its tasks could potentially be completed in half the time with a large language model, now or in the near future. The regional figure is the share of workers employed in those occupations. The 52% and 42% numbers are averages across region groups, not the percentage of people already using generative AI.
That denominator matters. A lawyer, analyst or developer can work in a highly exposed occupation while using no AI, and exposure can lead to augmentation rather than substitution. The method also assumes a task description can travel across employers and countries. Language, regulation, data access, bargaining arrangements, procurement and management choices shape what can be automated. A threshold that classifies a whole occupation hides variation between tasks and workers inside the same job title.[1]
The distinction that matters
Exposure, adoption and outcomes are different measurements
The OECD indicators describe different stages. None by itself predicts job loss.
Task exposure
Assess whether LLMs could halve task time
Potential capability
Regional share
Weight exposed occupations by employment
52% versus 42%
Firm adoption
Measure businesses reporting AI use
11% versus 26% in covered European regions
Work redesign
Employers change tasks and processes
Not measured by exposure
Worker outcome
Track pay, hours, productivity and jobs
Requires longitudinal evidence
Business adoption is uneven too, but it is a separate measurement
The OECD adds a second signal from European business statistics. In covered countries, an average of 11% of firms in Central and Eastern European regions reported using AI, compared with 26% in other regions covered. Unlike occupational exposure, this is an adoption measure, though it covers AI broadly rather than only generative systems. The two indicators should not be merged: one describes technically exposed work; the other records firm use under a different geography, population and definition.
The gap suggests that places with many exposed occupations may still move at different speeds depending on capital, infrastructure, management capability and specialist staff. Policy cannot be reduced to retraining after deployment. Regional agencies may need to help smaller firms evaluate tools, share expertise, improve connectivity and negotiate procurement, while worker institutions need evidence about workload, monitoring and job quality. Adoption without capability can produce poorly governed experiments; capability without access can widen the lead of successful cities.[1]
Innovation finance and patents reinforce the metropolitan advantage
The report places work exposure inside a broader geography of innovation. In 32 of 39 OECD countries, one small region receives more than half of national venture-capital funding and more than its population share would suggest. The largest functional urban area led patent activity in 27 of 31 countries between 2019 and 2024. These are not AI-only measures, but they show how finance, research and skilled labour can accumulate in places where GenAI exposure is high.
Concentration can be productive: dense labour markets and research networks make it easier to match ideas, finance and expertise. It can also intensify territorial inequality if high-exposure regions capture productivity gains while other places face weak adoption, fewer complementary jobs and out-migration. The report notes that some mid-sized places—including Cambridge, Eindhoven, Daejeon and Modena—produce unusually high patent shares relative to population. That complicates a capital-city narrative and points to specialised regional ecosystems as another route to innovation.[1]
What the figures mean for workers and local leaders
For workers, a high regional percentage should prompt questions about task redesign, not panic about a matching percentage of jobs. Employers decide whether time saved becomes higher output, shorter hours, fewer roles or different services. Outcomes vary between a public-service office, bank, design firm and hospital even when occupations share an exposure classification. Training should be tied to real workflows and rights: how to verify outputs, protect confidential information, challenge automated monitoring and retain professional judgement.
For local leaders, the useful unit is place as well as occupation. A region may need broadband, secure compute, adult education, university-industry links and procurement support, but also social infrastructure and affordable housing so skilled workers can stay. The comparison is descriptive and cross-sectional. It does not test whether high-exposure regions have already gained productivity, lost employment or widened wages. Policies should be evaluated against observed worker outcomes rather than treating exposure as destiny.[1]
What would change this assessment
Evidence would strengthen with repeated regional measurements that separate generative AI from other AI, link tool use to tasks, and follow productivity, pay, hours, hiring and displacement. Firm-size, sector and worker-demographic breakdowns would show whether averages conceal unequal gains. Comparable adoption data outside Europe are especially important. Audits should also test whether the occupation-task mapping remains accurate as models, prices and organisational practice change.
The assessment would weaken if employers adopt slowly in highly exposed occupations, capability fails under real constraints, or new tasks offset assumed time savings. The OECD report is an official international statistical synthesis with source tables, not a causal forecast. Its contribution is geographical: exposure, adoption capacity and innovation resources are distributed unevenly. That is consequential for workforce and regional policy, but it is not an urgent event and does not meet the Breaking standard.[1]
Watch the background, then check the evidence
Illustrative photo · not a video frame
UNESCO · 11 min 27 sec · Background explainer
UNESCO Talks: Interview with 2024 Nobel laureate Professor Daron Acemoglu
YouTube loads only when you choose; playback starts when you press play.
Economist Daron Acemoglu discusses how choices about AI affect workers, inequality and society. Pair the interview with our comparison of task exposure, hiring and skills evidence.
Evidence check: An expert interview from 2024, offering a perspective on policy and work rather than a measured forecast of job losses.
Published 2024-10-14. Source checked 14 October 2026. Check captions on the publisher’s video page.
Read the AI and jobs guidePhoto: Dylan Gillis / Unsplash · Licence
Continue the story
Read next
Continue exploring work: compare the finding and its evidence limits.
Do agentic tools now make up most business AI use in the EU?
Within OpenAI's Enterprise and Business customer sample, yes: Codex and ChatGPT Work produced 55% of included output tokens at the end of September. That is not a measure of all EU firms, completed work or productivity.
Read the next reportExplore the full source and evidence record
What this means for people
- Professional workers in capital and highly educated regions are more likely to have tasks language models could accelerate, but exposure does not determine job loss.
- Smaller firms and lagging regions may need shared expertise and infrastructure to adopt useful systems safely.
- Governments need to track wages, hours, job quality and training access, not only AI adoption.
Global context
AI policy is often national, but jobs, firms, universities and investment cluster locally. The same technology can meet a dense professional-services economy, a region with exposed work but weak adoption, or a place with less GenAI exposure yet substantial risk from older automation. Place-based evidence is necessary to avoid one-size-fits-all workforce policy.
What the evidence does not yet show
- The exposure indicator is task-based, not observed workplace deployment or productivity.
- A 50% task threshold classifies whole occupations and hides variation.
- Firm adoption covers European regions and AI broadly; exposure spans 29 OECD countries and GenAI potential.
- The report is descriptive and does not establish causation.
What to watch next
- Regional adoption data outside Europe.
- Longitudinal pay, hours and employment outcomes.
- Firm-size and demographic breakdowns.
- Exposure updates as models change.
Living evidence record
Impact record IAI-0NCHU8C
Evidence stage
Observed
Confidence
Supported
Reporting basis
Source analysis
Independent or research support
Not yet
Record status
Monitoring
Last checked
5 November 2026
Source trail
1 direct source across 1 source type.
People impact
Documented in this record.
Uncertainty
Limits and next checks are explicit.
Stages describe the evidence available—not whether a technology is good or bad. See the public method.
Single-source reporting disclosure
This record analyses one direct source. It can establish what OECD published or reported, but it is not independent corroboration of every performance claim or predicted outcome. The confidence label will change only when broader evidence is added.
Evidence trail
Sources used for this report
Links checked 5 November 2026
Research question, topics and regions
How occupational mix, education and business adoption create uneven regional exposure to generative AI across OECD economies, and what the indicator does and does not predict
Topics: Work · Regional inequality · Generative AI · Skills · Business adoption · Cities
Regions: OECD · Europe · North America · Asia-Pacific
This report is labelled source analysis. We summarise and analyse source material in our own words; company statements remain attributed claims until independently supported. Translated summaries preserve the meaning of the original source and link back to it. Read our editorial standards.
Reader commentary
What does this mean where you are?
Tell us what you have seen at work, in your community or in everyday life. A question is welcome too. No account is needed; we review notes before publishing them.
Explore commentary across the portal →Published reader notes
0No published reader notes yet. You can start the evidence-led discussion above.
Prefer a private correction or response? Contact the newsroom.