Home

Preprint

Global Automation Atlas

Prashant Garg Prashant Garg is a Postdoctoral Researcher in Economics at Bocconi University. His research covers science, innovation, production, and media using machine learning, causal inference, and network science. He holds a PhD in Economics from Imperial College London. Tommaso Crosta Tommaso Crosta is a PhD candidate in Economics at Bocconi University. His research sits at the intersection of development and labour economics, with additional interests in Bayesian statistics applied to microeconomics and meta-analysis. Jasmin Baier Jasmin Baier is a doctoral candidate in Behavioral Economics and Public Policy at the Blavatnik School of Government, University of Oxford. Her research sits at the intersection of behavioral and development economics, focusing on skills, AI and labor markets, and human-AI interaction. (2026)

Download PDF Supplementary Materials arXiv Replication package

Abstract

Automation can displace or complement labour, but this need not be constant across economies. Existing exposure measures typically assign fixed scores to tasks or occupations and capture cross-country variation through employment structure. We posit that exposure to automation depends jointly on task content and country-level conditions. We use a large language model to classify 18,797 work tasks in 124 economies by exposure, labour margin, technology channel and artificial-intelligence materiality. Country conditioning changes the classification of the same task and the occupation exposure rankings, especially in lower-income economies. Country-conditioned scores improve out-of-country predictions of firms’ production-technology use and AI adoption; physical-execution exposure tracks industrial-robot use and density, while Atlas scores align with observed work-related ChatGPT and Claude use across activities and countries. The share of tasks exposed to automation ranges from 3.3% to 61.6% across countries. It rises with income but varies considerably within income groups. Lower-income economies are more concentrated in rule-based and labour-substituting forms of automation, whereas physical execution, planning and inference channels, together with labour-augmenting uses of artificial intelligence, become more prominent with development. Combined with employment data, we find that women are disproportionately employed in occupations with substitution-facing exposure. Across countries, higher informality is associated with a smaller exposed-task share but a larger substitution-only share among exposed tasks. Machine-learning hypothesis generation methods point to both economic and institutional drivers: automation is more feasible where firms have machinery, digital records and connected systems, but less so where tasks depend on local knowledge and trusted relationships.

Main results

Figure 1

What country conditioning changes

Country context changes task classifications and occupation rankings. These differences also improve prediction of firms' technology use.

The US benchmark applies the same task scores everywhere. Prediction tests exclude each country from estimation in turn.

Figure 2

Cross-country patterns in task exposure

Automation exposure rises with income, but countries at similar income levels still differ substantially.

Exposure is the share of tasks classified at levels 2 or 3.

Figure 3

Country-level task pathways

Substitution-only exposure exceeds augmentation-only exposure in every country. Lower-income economies have less exposed work overall, but a more substitution-heavy mix.

Labour margins classify how automation changes the worker's role inside exposed tasks. They are task-level judgments, not estimates of employment effects.

Figure 4

Automation channels and AI materiality

Rule-based workflows dominate exposed work in poorer economies. AI is more often central to information transformation, inference and planning than to rule-based workflows or physical execution.

Channel shares use exposed tasks as the denominator. AI can be material in any channel.

Supplementary Figure S23

AI materiality, AI function mix, and labour-margin composition

AI is materially involved in a larger share of exposed tasks in richer economies.

AI is material when a learned model is central to the automation route.

Figure 5

Labour-margin exposure, gender composition, and informality

Women are disproportionately employed in occupations with substitution-facing exposure. Employment sorting explains most of the occupation gap.

After accounting for income, higher informality is associated with a smaller substitution-only share among exposed tasks.

Figure 6

Country covariates and rationale concepts behind exposure differences

GDP per capita and internet use are the strongest country-level predictors of exposure. Rationales for the same task across countries point to machinery, digital records, local knowledge and trusted relationships.

These are predictive associations and hypotheses, not causal estimates.