labor-market

ILO Brief: Developing Economies Risk AI "Disruption Without Dividend"

Only 11% of jobs in low-income countries are exposed to generative AI, against 34% in rich ones. An ILO brief explains why that gap does not shield them.

PorEditor e autor
Publicado:
Análise assistida por IA

About 11% of employment in low-income countries is in occupations exposed to generative AI, against 34% in high-income countries. [Fact] A November 2026 policy brief from the International Labour Organization warns that lower exposure can still leave these economies worse off, because AI may reach automatable jobs there before workers and firms have the electricity, connectivity and skills to gain from it. The brief names this outcome "disruption without dividend."

Lower exposure, weaker capacity to gain

The exposed share of employment is 34% in high-income countries, 25% in upper-middle-income, 20% in lower-middle-income and 11% in low-income countries. [Fact] A study of 135 countries cited in the brief finds that developing economies have lower exposure to automation by generative AI than advanced economies and comparable potential for augmentation. [Fact] The ILO authors argue that weak digital infrastructure can let automation reach workers before augmentation does. [Claim]

In 2025 about 2.2 billion people had no internet access. The online share of the population was 36% in Africa and 92% in Europe. [Fact] World Bank Enterprise Surveys for 2022–25 classified 27 of 7,016 firms in the Middle East, North Africa, Afghanistan and Pakistan as AI adopters. [Fact] One projection cited in the brief puts ten-year output gains from AI at about 5% in advanced economies and 2.7% in low-income countries under a high-productivity scenario. [Estimate]

Job effects so far are small and fall on new entrants

The brief describes the labour market effects of AI to date as modest and slow-moving, including in advanced economies. An ILO analysis of 40 emerging digital technologies found that the employment-to-population ratio rose by 1.39 percentage points in low-income countries, 0.76 in high-income, 0.65 in upper-middle-income and 0.39 in lower-middle-income countries. [Fact] Danish administrative data showed no measurable effect of AI chatbots on earnings or hours two years after ChatGPT's release. [Fact]

The losses that have been measured are concentrated among young workers. In the United States, employment of workers aged 22 to 25 in the most AI-exposed occupations fell 6% between 2022 and 2025, while employment of older workers in the same occupations rose 6% to 9%. [Fact] The decline was concentrated in software development and customer service and came through reduced hiring of new entrants. The brief reports a similar, more preliminary pattern in the Philippines and Thailand. [Fact] On online freelance platforms, demand for writing and translation skills fell 20% to 50% relative to trend after ChatGPT's release, and demand for novice workers declined. [Fact]

The people who train AI have thin benefits

Data annotation, data labelling and content moderation for AI systems are often done by workers in developing economies, through microtask platforms or AI business process outsourcing firms. Among such workers surveyed in 2022 and 2023, 50% reported access to paid holidays, 65% to sick leave and 50% to maternity leave. [Fact] ILO research counted at least 653 active digital labour platforms as of October 2025. In June 2026 the International Labour Conference adopted the Decent Work in the Platform Economy Convention (No. 193), the first international labour standard dedicated to platform work. [Fact]

What this means for workers, and how to prepare

The occupations the brief names correspond most closely to customer service representatives, software developers, writers and authors and interpreters and translators. [Estimate]

  • If you are entering customer service or software work, expect fewer junior openings. Experienced workers in the same occupations kept gaining jobs, so build experience through apprenticeships and vocational programmes, which the brief asks governments to expand.
  • If you freelance in writing or translation, add work that clients still buy from people. Demand rose in some skill clusters that AI complements.
  • If you do annotation or moderation work, ask for a written contract and a pay schedule before you accept tasks, and check whether paid leave is included. The brief recommends that governments require written contracts and predictable, timely pay.
  • Learn AI tools alongside skills specific to your sector. The brief lists digital literacy, AI literacy and domain training as priorities for women, young people and informal workers.

The brief closes with nine policy recommendations for governments, including shared compute access, AI modules in labour force surveys and consultation with workers before algorithmic management systems are deployed.

Limits

The brief is a review of published studies and contains no new survey or data. Most of the evidence it cites on adoption and employment comes from advanced economies, and its exposure figures measure potential, not observed job changes. The occupation links to our job pages are our match.

Sources

  • International Labour Organization, "AI and developing economies: Promoting decent work in a time of rapid technological change," Policy Brief, November 2026: https://www.ilo.org/publications/ai-and-developing-economies-promoting-decent-work-time-rapid-technological
  • Full text (PDF): https://www.ilo.org/sites/default/files/2026-11/Policy%20Brief_AI%20and%20developing%20economies.pdf

About this analysis

This article was written with AI assistance from the full text of the brief. Figures marked [Fact] are reported in the brief; the item marked [Estimate] in the last section is our occupation match.

Histórico de atualizações

  • Publicado pela primeira vez em 17 de novembro de 2026.
  • Sem atualizações substanciais desde a publicação inicial.

Etiquetas

#ILO#developing economies#generative AI#AI exposure#platform work#data annotation

Fontes

  1. ilo.org
  2. ilo.org