Why Every Country's Clerical Workers Score 8.5/10 on AI Exposure - Everywhere, Without Exception

The ISCO-08 System: Why the Score Is Universal

The International Labour Organization's ISCO-08 (International Standard Classification of Occupations, 2008 revision) classifies all jobs into 10 major groups based on what the work involves - its skill level and task profile. Group 4 is Clerical Support Workers, defined as: "tasks which require the knowledge and experience necessary to organise, store, compute and retrieve information."

Specifically, ISCO-08 Group 4 includes: general and keyboard clerks, customer services clerks, numerical and material recording clerks, and other clerical support workers. In every country that uses ISCO-08 reporting - which is most of the world, through ILO ILOSTAT (CC BY 4.0) - a clerical worker is classified by what they do, not where they do it.

The AI exposure score of 8.5/10 assigned to Group 4 by WorldJobsData reflects the assessment by Claude (Anthropic) of how much of the work in this group - data entry, classification, scheduling, correspondence, records management - can be handled by current generative AI systems. The answer is: a large share, because these are exactly the tasks where AI has demonstrated the most consistent, deployable capability. That assessment does not change based on which country the worker is in. The tasks are the same. The AI capability is the same. The 8.5/10 applies universally.

The Full ISCO-08 Risk Spectrum

To understand why 8.5/10 is significant, it helps to see where every group sits.

Source: ISCO-08 AI exposure scores assessed by Claude (Anthropic) based on task analysis of each major occupation group. WorldJobsData 2026.

The gap between Group 4 (8.5/10) and the next highest group - Professionals at 6.5/10 - is 2 full points. No other pair of adjacent groups has a wider gap. Clerical work is not just the most AI-exposed group. It is distinctly more exposed than everything else.

Where the 146 Million Are: Country by Country

Source: ILO ILOSTAT (CC BY 4.0), ISCO-08 Group 4 employment by country. Data years 2019-2024 per country. AI exposure score is uniform across all countries at 8.5/10.

Japan: The Country Most Structurally Dependent on a High-Risk Group

Japan's 20.5% clerical share is a direct product of its economic history. Post-war Japan built a large bureaucratic corporate culture - the famously hierarchical kaisha - that created millions of administrative roles. The "office lady" (OL) tradition placed large numbers of women in scheduling, correspondence and records roles. Japan's public sector is paper-intensive by global standards. The country has among the lowest rates of AI adoption in enterprise software among OECD nations, partly because its administrative processes are built around paper-based verification.

This combination - the highest clerical concentration globally, the slowest enterprise AI adoption, and 14.5 million workers in Group 4 - means Japan faces a particularly acute version of the clerical AI disruption. When Japanese enterprises do adopt AI for clerical tasks, the scale of displacement will be larger than in any other country.

Macao: 31.5% Clerical - The Extreme Case

Macao is an outlier that clarifies the mechanism. With 113,000 total workers and 31.5% in clerical roles, Macao has the highest clerical share globally. The reason is Macao's casino and gaming economy. Casino operations generate enormous volumes of administrative work: cage operations, compliance records, regulatory filings, customer account management, VIP program administration. These are all clerical roles. They are also almost all 8.5/10 AI exposure. Macao's overall AI exposure score of 5.74/10 - second globally, behind only Luxembourg - is largely explained by this concentration.

Why This Matters More Than Any Single Country's Score

The universality of the 8.5/10 score for clerical work carries a specific implication: the AI tools being deployed right now are already capable of handling the core tasks of this group. This is not a future risk. Microsoft Copilot processing documents, Salesforce automating CRM entries, chatbots handling customer service correspondence, AI scheduling assistants - these are deployed products, in use today, in exactly the tasks that define ISCO-08 Group 4.

The question is not whether AI will automate clerical work. For a substantial portion of clerical tasks, the automation is already underway. The question is how fast organisations will deploy these tools at scale, and what happens to the workers when they do. Different countries have very different answers to the second question - in terms of social safety nets, reskilling infrastructure, and labour market flexibility. But the first question - will it happen - is settled.

146 million workers. 8.5/10. The same number, everywhere.