Key Details
The study combines observed AI usage with labour market data to estimate realised productivity gains and examine how these benefits are distributed across countries and occupations.
Aspect | Key Details |
|---|---|
Evidence base | Five waves of the Anthropic Economic Index covering Claude AI conversations between January 2025 and February 2026 across 100+ countries |
Countries analysed | 86 countries with wage and employment data used to estimate productivity gains |
New metric | Labour Cost Equivalent (LCE): estimates the annual value of labour costs saved through AI-assisted task completion |
Global estimate | AI currently generates approximately US$2.7 trillion in annual labour-cost savings, equivalent to 3.4% of the combined GDP of the countries analysed |
New distribution measure | AI Concentration Index (ACI): measures whether AI productivity gains are concentrated among higher-paid occupations or broadly shared across the workforce |
Cross-country pattern | High-income economies capture 96% of estimated global AI gains; AI gains equal 4.2% of GDP in high-income countries compared with 0.6% in middle-income and 0.1% in low-income countries |
India | ACI: 0.84, indicating AI productivity gains remain concentrated among software and STEM occupations |
Principal policy message | Expanding AI adoption across occupations—not merely developing more advanced AI models—will determine future economy-wide productivity gains |
Why This Study Matters
Most studies estimate which jobs AI could affect by analysing occupational exposure or automation potential.
This IMF paper, Aggregate Gains from AI and Their Distribution: Global Evidence from Usage Data, instead measures AI's realised economic impact by analysing actual AI usage. By combining millions of observed AI interactions with wage and employment data, it estimates how much labour time AI is already saving and where those productivity gains are accruing. This makes it one of the first cross-country studies to quantify AI's current economic value using real-world usage rather than theoretical scenarios.
A New Measure of AI's Economic Value
The IMF paper introduces Labour Cost Equivalent (LCE) as a way of estimating AI's realised productivity gains. Each AI-assisted task is valued according to the labour time saved and the prevailing wage for that occupation, allowing the authors to estimate the aggregate economic value already being generated by AI.
Using this approach, the study estimates that AI currently produces approximately US$2.7 trillion in annual labour-cost savings, equivalent to 3.4% of the combined GDP of the 86 countries analysed.
Growth Is Being Driven by Wider Occupational Adoption
A central finding is that aggregate AI gains are increasing because AI is spreading into occupations employing much larger numbers of workers.
While early AI adoption was concentrated among software developers and other technical professionals, usage is increasingly expanding into occupations such as education, office administration, sales and customer services. Although the value generated by each individual AI interaction is gradually declining, wider adoption across the labour market is producing larger overall productivity gains.
The Global AI Productivity Divide Remains Wide
The report finds that AI's economic gains remain heavily concentrated among advanced economies.
High-income countries account for 96% of estimated global labour-cost savings, with AI gains equivalent to 4.2% of GDP, compared with 0.6% in middle-income economies and only 0.1% in low-income economies. Countries with stronger AI regulatory readiness, higher incomes and more service-oriented economies generally capture both larger aggregate gains and a broader distribution of benefits.
The study also finds that English-language dominance in current AI models can slow the diffusion of AI benefits in countries where institutional knowledge is less available in English.
Distribution Within Countries Is Also Uneven
To examine who benefits from AI within each country, the paper introduces the AI Concentration Index (ACI).
Developing economies generally exhibit much higher concentration, with AI productivity gains accruing mainly to a relatively small group of highly skilled professionals. In contrast, higher-income countries have seen AI spread more broadly into clerical, administrative, sales and service occupations, resulting in a more even distribution of productivity gains.
Encouragingly, the report finds that occupational concentration has begun to decline in many countries as AI adoption expands beyond its initial technical user base.
India's AI Gains Remain Concentrated
India reflects many of the patterns observed across developing economies.
The report assigns India an AI Concentration Index of 0.84, indicating that AI-driven productivity gains remain concentrated among software and STEM professionals. Adoption across occupations such as education, office work, sales and services remains comparatively limited, restricting the economy-wide productivity gains that broader occupational diffusion could generate.
The findings suggest that India's long-term AI dividend will depend not only on advances in AI capability but also on expanding AI adoption across the wider workforce.
Diffusion Is Becoming the Central Policy Challenge
The paper argues that future AI policy should increasingly focus on enabling widespread adoption rather than frontier innovation alone.
Strengthening AI regulatory readiness, expanding digital capabilities, improving access to high-quality institutional knowledge in English and local languages, and supporting AI adoption across a wider range of occupations will be critical to ensuring that AI productivity gains become both larger and more broadly shared.
What Is Labour Cost Equivalent (LCE)?
Labour Cost Equivalent (LCE) estimates the monetary value of labour costs saved through AI-assisted task completion. It measures the value of time saved using prevailing wages, providing an estimate of AI's realised productivity gains rather than its theoretical automation potential.
What Is the AI Concentration Index (ACI)?
The AI Concentration Index (ACI) measures how evenly AI productivity gains are distributed across occupations.
Higher ACI indicates gains are concentrated among a small number of high-income occupations.
Lower ACI indicates productivity gains are spread across a broader share of the workforce.
India's ACI of 0.84 suggests AI benefits remain concentrated within a relatively small professional segment of the labour market.
Policy Relevance
The paper introduces a practical methodology for measuring realised AI productivity gains using observed usage rather than estimates of automation exposure.
The findings suggest that AI diffusion across occupations will increasingly determine the scale of economy-wide productivity gains.
India's AI strategy should place greater emphasis on expanding adoption in education, administration, healthcare, retail and other labour-intensive sectors alongside frontier AI development.
Strengthening AI regulatory readiness, digital skills and high-quality multilingual knowledge resources can accelerate broader adoption and improve the distribution of AI's economic benefits.
The report highlights the importance of monitoring not only how much AI is used, but also who benefits from its productivity gains across the labour market.
Follow the Full Report Here: Aggregate Gains from AI and Their Distribution: Global Evidence from Usage Data (IMF Working Paper, 2026)