No dividends interruption? How does the digital divide and task differences segment the global influence of GenAI (English) 2026

11328 Disruption without Dividend? How the Digital Divide and Task DifferencesSplit GenAI’s Global ImpactPublic Disclosure Authorized Paweł GmyrekMariana ViollazHernan Winkler A verified reproducibility package for this paper isavailable athttp://reproducibility.worldbank.org,clickherefor direct access. Policy Research Working Paper11328 Abstract This article examines how generative artificial intelligence(GenAI) could affect labor markets globally, with particularattention to the uneven distribution of risks and oppor-tunities between advanced and developing economies.Cross-country differences in occupational structure suggestthat developing economies face lower aggregate automationexposure than advanced economies but comparable poten-tial for task augmentation. However, disparities in digitalinfrastructure create an asymmetry: workers in positions gains. This finding suggests that developing countries mayexperience the disruptive effects of GenAI faster than itsproductivity benefits. At the same time, conventional occu-pational exposure measures systematically overestimate theimpact of GenAI in developing countries by assuming uni-form task content across economies. Using data from skillssurveys, the article demonstrates that workers in developingcountries perform substantially fewer non-routine analyt- The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about developmentissues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry thenames of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely thoseof the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and Disruption without Dividend? How the Digital Divide and Task Differences Split GenAI’sGlobal Impact∗ Pawel Gmyrek† JELcodes:J24,O33,J21,O15,L86Keywords:Generativeartificialintelligence,Automation,Digitaldivide,Occupational 1Introduction A rapidly growing body of literature documenting the potential labor market impacts of Genera-tive Artificial Intelligence (GenAI) has emerged ever since the launch of ChatGPT in November2022. While early experimental studies focused on specific occupational segments reveal sub- stantial heterogeneity in impacts, the aggregate implications for labor demand, job quality, andearnings distribution remain highly uncertain.1Estimates of broader labor market exposure suggest that GenAI’s distinctive capacity to impact high-skilled, cognitive work sets it apart from previous waves of automation,2with advanced economies facing both greater risks andopportunities compared to emerging markets.3Yet a critical gap persists in our understanding Our previous research on Latin America and the Caribbean (LAC) (Gmyrek et al., 2024) re-vealed a fundamental constraint for global AI diffusion:nearly half of the jobs that couldtheoretically benefit from GenAI augmentation are unable to realize this potential due to alack of access to basic digital technologies at work. This underscores that the digital divide is not only a constraint on technology adoption but also a powerful amplifier of global inequality: To further explore how digital infrastructure constraints interact with occupational exposureto GenAI to shape the distributional impacts of this technology across all income levels andgeographic contexts, this study updates and extends our 2024 methodology, incorporating recentadvances in GenAI capabilities and providing the first comprehensive, cross-regional analysis. This article yields four principal findings.First, a phenomenon that we described as“smallbuffer, big bottlenecks”in our earlier study concerning LAC (Gmyrek et al., 2024) is equallyobserved in the global context, demonstrating an inherent asymmetry of threats versus benefitsof GenAI technologies in developing countries. Specifically, workers in roles susceptible to au-tomation possess sufficient internet connectivity for relatively rapid displacement even within swiftly than its productivity benefits in developing nations. Second, our analysis of detailed country-level data from multiple survey sources confirms apositive correlation between GenAI exposure and economic development, with high-incomeeconomies demonstrating higher average exposure rates (about 32 percent of total employ- ment) than low-income economies (about 15 percent).5We show that this correlation is mainlydriven by occupations facing a higher risk of automation, since the share of jobs amenable to Third, while low- and lower-middle-income countries exhibit lower aggregate automation risksdue to their distinct employment structures, GenAI poses a considerable threat to office-basedoccupations, representing a disproportionate share of the “good” skilled jobs in these economies.T