Generative AI Occupational Exposure: ILO Global Index Reveals 1 in 4 Workers Affected

Understanding the ILO Global Index of Generative AI Occupational Exposure

Generative AI occupational exposure has emerged as one of the defining workforce challenges of the decade. As large language models, image generators, and multimodal AI systems advance at unprecedented speed, policymakers, employers, and workers urgently need data-driven frameworks to understand which jobs face the greatest impact. The International Labour Organization (ILO) — the United Nations agency responsible for setting international labor standards — has answered this call with its refined 2025 Global Index of Occupational Exposure to Generative AI.

This landmark ILO Working Paper 140 updates the organization’s original 2023 index, incorporating two critical developments: significant advances in generative AI technology and the growing familiarity of workers and employers with these tools. The result is a comprehensive, evidence-based assessment that maps generative AI occupational exposure across the entire global workforce, covering every major occupation group in the International Standard Classification of Occupations (ISCO-08).

The headline finding is striking: one in four workers worldwide — roughly 25% of total global employment — now works in an occupation with some measurable degree of generative AI exposure. This figure represents a significant recalibration from earlier estimates and reflects GenAI’s rapidly expanding capabilities in areas like natural language processing, code generation, data analysis, and creative content production. For organizations trying to understand how AI is reshaping work, the ILO’s index provides the most rigorous global benchmark available. You can explore similar research findings through Libertify’s interactive library, which transforms dense reports into engaging experiences.

Methodology: How 52,558 Data Points Reveal AI’s Workforce Impact

The strength of the ILO’s generative AI occupational exposure index lies in its methodological rigor. Unlike many AI impact studies that rely on expert opinion alone or narrow technological benchmarks, the ILO employed a multi-stage approach that combines large-scale worker input with structured expert deliberation and AI-assisted prediction modeling.

The research team began with the Polish occupational classification system, which contains 29,753 individual work tasks — one of the most granular task-level databases available for labor market analysis. From this universe, they selected a representative sample of 2,861 tasks for detailed assessment. A survey of 1,640 workers, drawn from each of the ten major ISCO-08 occupational groups, then evaluated these tasks for their perceived automation potential using generative AI tools. This process yielded an impressive 52,558 individual data points about task-level AI exposure.

But worker perception alone can underestimate or overestimate technological capabilities. To calibrate these findings, the ILO assembled a panel of international experts who completed their own independent assessments, followed by several rounds of Delphi-style structured discussions to build consensus. The Delphi method — widely used in technology forecasting since the RAND Corporation pioneered it in the 1950s — reduces individual bias through iterative anonymous feedback rounds.

The combined knowledge from workers and experts was then synthesized into a repository designed to transcend national specificities. The research team used this repository to develop an AI assistant capable of predicting automation scores for any task described in ISCO-08 technical documentation. This means the index can be extended to any country that uses or maps to the ISCO-08 classification system — essentially the entire world.

From Raw Data to Four Exposure Gradients

Rather than using a simple binary classification of “exposed” versus “not exposed,” the ILO created four progressively increasing exposure gradients. Gradient 1 represents minimal exposure — tasks where generative AI has limited applicability. Gradient 4 represents the highest exposure — tasks where GenAI could substantially perform or transform the work. This nuanced framework acknowledges that AI exposure exists on a spectrum and that most occupations contain a complex mix of differently exposed tasks.

Generative AI Exposure Across Four Gradient Levels

The four-gradient framework reveals a nuanced picture of generative AI occupational exposure that defies simplistic narratives about robots replacing workers. At the global level, the distribution across gradients shows that while exposure is widespread, the highest levels of exposure remain concentrated in specific occupational categories.

At the broadest level, one in four workers (25%) occupies a role with at least some generative AI exposure — meaning their occupation includes tasks in Gradient 1 or above. However, the critical figure for understanding labor market disruption is the 3.3% of global employment that falls into Gradient 4, the highest exposure category. These are roles where a substantial proportion of core tasks could potentially be performed or fundamentally altered by current generative AI capabilities.

The progression across gradients matters enormously for policy. Workers in Gradient 1 may experience AI as a productivity-enhancing tool that handles peripheral tasks, freeing them for higher-value work. Workers in Gradient 2 and 3 face more significant changes to their daily workflows, potentially requiring reskilling or role redefinition. Those in Gradient 4 face the most profound transformation — their core professional competencies overlap significantly with what generative AI can now do.

Understanding which gradient a workforce segment falls into is essential for organizations designing AI adoption strategies. A company employing primarily Gradient 2 workers needs a fundamentally different change management approach than one with significant Gradient 4 exposure. Explore how leading organizations are adapting to these shifts through our collection of AI workforce transformation research.

Clerical Occupations and the Highest AI Exposure Risks

Among all occupational groups analyzed in the ILO’s generative AI occupational exposure index, clerical workers continue to face the highest levels of exposure — a finding consistent with the original 2023 study and confirmed with stronger evidence in the 2025 update. This result has significant implications for millions of workers worldwide, particularly in sectors like financial services, government administration, healthcare administration, and corporate back-office operations.

Clerical occupations — which include roles such as general office clerks, secretaries, data entry operators, bookkeepers, accounting clerks, and administrative assistants — are characterized by tasks that are heavily text-based, rule-governed, and repetitive. These are precisely the task types where current large language models and generative AI systems excel. Document drafting, email correspondence, data compilation, report formatting, meeting scheduling, and basic financial record-keeping all fall within the demonstrated capabilities of commercially available AI tools.

The concentration of high-exposure tasks in clerical roles creates a compounding vulnerability. Unlike professional occupations where a subset of tasks may be automatable but others require deep expertise, creative judgment, or interpersonal skills, many clerical positions have a higher proportion of tasks that cluster in Gradients 3 and 4. This doesn’t necessarily mean these jobs will disappear — but it does mean they will likely undergo the most dramatic transformation in terms of how work is organized and what skills are valued.

Organizations that rely heavily on clerical workforces should be actively planning for this transition. The OECD’s artificial intelligence and work research program provides complementary guidance on managing workforce transitions in high-exposure sectors.

Beyond Clerical: Expanding Exposure in Digital Professions

A notable change from the 2023 index is the increased exposure detected in strongly digitized occupations. Software developers, data analysts, graphic designers, copywriters, market researchers, and financial analysts have all seen their exposure levels rise as generative AI capabilities have expanded. This shift reflects the technology’s growing ability to handle specialized, knowledge-intensive tasks that were previously considered safe from automation.

The Gender Gap in Generative AI Occupational Exposure

One of the most consequential findings of the ILO’s generative AI occupational exposure research is the pronounced gender disparity in who faces the highest levels of AI-driven job transformation. The data reveals that women are disproportionately concentrated in occupations with the highest exposure levels — a pattern that could either exacerbate existing workplace inequalities or, with thoughtful policy intervention, create pathways to more equitable outcomes.

At the global level, 4.7% of female employment falls into Gradient 4 — the highest exposure category — compared to just 2.4% of male employment. This means women are nearly twice as likely as men to work in roles facing the most significant generative AI impact. The disparity is not accidental; it directly reflects long-standing patterns of occupational segregation that channel women disproportionately into clerical, administrative, and support roles — precisely the categories where GenAI exposure is most intense.

In high-income countries, the gender gap becomes even more dramatic. The ILO data shows 9.6% of female employment in Gradient 4 versus just 3.5% for male employment — a ratio approaching 3:1. This amplification in wealthier economies occurs because high-income countries have larger service sectors, more formalized administrative structures, and higher rates of female participation in clerical and office-based work.

“These differences increase with countries’ income level, and so does the overall exposure.” — ILO Working Paper 140

The gender dimension of generative AI occupational exposure demands urgent attention from policymakers. Without proactive intervention, AI-driven job transformation could disproportionately displace or downgrade women’s employment. Conversely, targeted reskilling programs that help women in high-exposure roles transition to AI-augmented or AI-management positions could turn this challenge into an opportunity for greater workplace equity. The UN Women research library provides additional perspectives on gender-responsive approaches to technological change.

Income Disparities: AI Exposure in Rich vs. Developing Nations

The ILO’s generative AI occupational exposure index reveals a stark divide between countries at different income levels, with profound implications for global economic development and inequality. The data shows that generative AI exposure scales dramatically with national income — high-income countries face roughly three times the exposure levels of low-income countries.

In high-income countries (HICs), 34% of total employment has some level of generative AI exposure, with significant shares in the higher gradients. In contrast, low-income countries (LICs) show just 11% of employment with any AI exposure. This difference reflects fundamental structural distinctions in economic composition: wealthier nations have larger knowledge-based service sectors, more digital infrastructure, higher rates of formal office-based employment, and greater penetration of the technology platforms through which generative AI is typically deployed.

This income-exposure relationship creates a complex policy dilemma. On one hand, high-income countries face more immediate disruption to their labor markets and greater urgency for transition planning. Their workers in clerical, professional, and technical roles are already experiencing AI’s effects in daily work tasks. On the other hand, low-income countries face a different kind of risk: as generative AI enables companies to automate tasks previously outsourced to lower-wage economies, the traditional development pathway of moving workers from agriculture into manufacturing and then into services could be fundamentally disrupted.

The development implications extend beyond direct employment effects. Countries that successfully manage the generative AI transition — by investing in digital skills, updating educational curricula, and building adaptive social protection systems — may gain competitive advantages. Those that fail to prepare could find their workforce increasingly misaligned with the demands of a global economy where AI-augmented productivity sets the baseline.

Professional and Technical Roles Face Growing AI Disruption

While clerical occupations retain the highest overall exposure levels, the 2025 update to the ILO’s generative AI occupational exposure index highlights a significant and accelerating trend: strongly digitized professional and technical roles are seeing increased AI exposure compared to the 2023 assessment. This shift reflects generative AI’s expanding capabilities beyond basic text processing into specialized domains.

Software developers, for example, now work alongside AI coding assistants that can generate functional code, debug programs, write documentation, and even architect system designs. Data analysts find that large language models can query databases, produce visualizations, and draft analytical narratives. Financial analysts see AI tools generating investment reports, running scenario analyses, and summarizing complex regulatory filings. Marketing professionals use generative AI for campaign copywriting, audience segmentation analysis, and competitive research synthesis.

The pattern is clear: as generative AI models become more capable and as workers and employers become more familiar with their application, the frontier of exposure is expanding beyond routine clerical tasks into knowledge-intensive professional domains. This doesn’t mean professionals are facing imminent job loss — in fact, many professional tasks in Gradients 2 and 3 represent augmentation opportunities where AI handles routine analytical work while human professionals focus on judgment, creativity, client relationships, and strategic decision-making.

However, the trajectory is unmistakable. Professionals who fail to develop competency with AI tools risk finding their productivity increasingly outmatched by AI-augmented peers. Organizations that help their professional workforce embrace AI-assisted workflows — rather than resist them — are likely to see significant competitive advantages in quality, speed, and cost efficiency.

The Augmentation Frontier

The ILO research suggests a critical distinction between tasks where AI serves as a productivity multiplier and tasks where it serves as a potential substitute. For most professional roles, the majority of tasks fall into the augmentation category — AI handles data gathering, first-draft generation, and routine analysis, while the professional provides expertise, judgment, and interpersonal skills. Understanding this distinction is essential for workforce planning. Learn how leading institutions are communicating AI strategy through our future of work interactive experiences.

Job Transformation vs. Job Replacement: What the Data Shows

Perhaps the most important conclusion from the ILO’s generative AI occupational exposure research is that job transformation — not wholesale replacement — is the most likely outcome for the vast majority of affected workers. This finding pushes back against both the utopian narrative of AI freeing all workers from drudgery and the dystopian narrative of mass technological unemployment.

The evidence supports transformation over replacement because most occupations consist of a heterogeneous bundle of tasks with varying levels of AI exposure. A legal professional, for instance, might find that contract review (high exposure) is increasingly automated, but client counseling, courtroom advocacy, and strategic case planning (low exposure) remain firmly in human hands. An accountant might see routine bookkeeping and tax preparation (high exposure) handled by AI, while advisory services, audit judgment, and relationship management (low exposure) become more central to their role.

This task-level heterogeneity means that even in occupations classified as having high overall exposure, significant components of the work resist automation. The ILO framework explicitly accounts for this by measuring exposure at the task level rather than the occupation level, then aggregating upward. This granular approach reveals that pure replacement scenarios — where every task in an occupation can be handled by AI — are extremely rare.

The transformation model has several practical implications for organizations and workers. First, job descriptions and performance expectations will need to evolve as AI handles certain tasks. Second, the skills that define professional value will shift toward areas like critical thinking, ethical judgment, creative problem-solving, emotional intelligence, and the ability to effectively direct and quality-control AI outputs. Third, education and training systems need to prepare workers not for a world without AI, but for a world where working alongside AI is the norm.

Research from the MIT Department of Economics broadly supports this task-level transformation thesis, with studies showing that historical technology adoption has generally transformed rather than eliminated occupations, though the pace and distribution of benefits vary significantly.

Policy Frameworks for Managing Generative AI Workforce Transitions

The ILO’s generative AI occupational exposure index isn’t just an academic exercise — it’s designed to serve as a foundation for evidence-based policy responses. By linking the refined index with national micro data, policymakers can generate precise projections of how AI-driven transformation will affect specific sectors, regions, and demographic groups within their countries. This capability enables targeted rather than one-size-fits-all policy interventions.

The ILO emphasizes social dialogue as the cornerstone of effective transition management. This means bringing together governments, employers’ organizations, and workers’ representatives to negotiate the terms of AI adoption. Experience from previous technological transitions — from industrial automation to computerization — consistently shows that managed transitions produce better outcomes for all stakeholders than unmanaged disruption.

Several policy priority areas emerge from the data:

• Reskilling and upskilling programs: Targeted training initiatives for workers in Gradient 3 and 4 occupations, focused on developing complementary skills that AI cannot easily replicate — critical thinking, interpersonal communication, creative problem-solving, and AI tool proficiency.
• Gender-responsive interventions: Given the disproportionate impact on women, policies must explicitly address gender equity in reskilling access, career transition support, and representation in AI-augmented roles.
• Social protection modernization: Updating safety nets to support workers during transitions, including portable benefits, income smoothing programs, and flexible retraining support that doesn’t require workers to lose their current employment before accessing help.
• Educational system reform: Aligning curricula — from primary school through university and vocational training — with the skill requirements of an AI-augmented economy, emphasizing human-AI collaboration rather than competing with AI on its strengths.
• Development-sensitive approaches: Recognizing that low-income and high-income countries face fundamentally different exposure profiles and require different policy tools.

The ILO’s framework allows each country to map its specific occupational structure against the global exposure index, producing customized risk assessments that account for local labor market conditions, industrial composition, and existing policy infrastructure.

Preparing for the Future of Generative AI Occupational Exposure

The ILO’s 2025 Global Index of Generative AI Occupational Exposure represents the most comprehensive, methodologically rigorous assessment available of how generative AI is reshaping the global workforce. Its core findings — that one in four workers face some exposure, that 3.3% are in the highest exposure category, that women and high-income country workers face disproportionate impact, and that transformation rather than replacement is the dominant pattern — provide essential grounding for organizational and policy decision-making.

For business leaders, the index provides a data-driven framework for assessing which parts of their workforce face the greatest transformation pressure and where AI-augmented productivity gains are most achievable. For policymakers, it offers the foundation for evidence-based, targeted intervention programs. For workers, it clarifies where the most significant changes are likely and what types of complementary skills will remain valuable.

Perhaps most critically, the ILO’s research demonstrates that the generative AI transition is manageable — but only with deliberate, coordinated action. The window for proactive planning is narrowing as AI capabilities continue to expand. Organizations and governments that act on these findings now — investing in reskilling, redesigning workflows, strengthening social protections, and building inclusive dialogue mechanisms — will be far better positioned than those that wait for disruption to force reactive responses.

The full ILO Working Paper 140, including detailed methodological documentation and country-level analysis tools, is available through the ILO Publications portal. As the global conversation about AI and work intensifies, this research provides an indispensable evidence base for navigating one of the most significant workforce transitions in modern history.