AI promises large productivity gains but risks widening the global income gap: advanced economies appear to capture disproportionately larger growth benefits while middle‑ and low‑income countries lag, making urgent reskilling, infrastructure investment and international cooperation essential.

Main Finding

AI is a powerful engine for economic growth but also a force that can widen global disparities. Recent international evidence (IMF, World Bank, ILO, 2025–26) suggests advanced economies capture substantially larger growth gains from AI than low‑income countries, even as adoption expands in middle‑income countries. The dominant near‑term labor outcome is occupational transformation rather than wholesale job loss, creating a pressing need for large‑scale reskilling and coordinated policy action to share AI’s benefits equitably.

Key Points

• Growth differential: An IMF working paper (WP/25/076) estimates the GDP growth impact of AI in advanced economies may be more than twice the impact observed in low‑income countries, signalling a risk of widening the global income gap.
• Adoption pattern: The World Bank (2025) reports >40% of global ChatGPT traffic comes from middle‑income countries — evidence that generative AI use is spreading beyond high‑income countries but not necessarily translating into comparable economic gains.
• Labor market exposure: The ILO (2025) finds roughly 1 in 4 jobs worldwide are exposed to transformation from generative AI. The likely outcome is task and occupational transformation, not uniform displacement, implying large retraining and reallocation needs.
• Multi-dimensional impacts: The study examines AI’s macro effects on output, productivity, and investment and links these to distributional issues (within and across countries), development trajectories, and the digital divide.
• Policy imperative: Without policy intervention (skills, infrastructure, fiscal and regulatory measures, and international cooperation), AI could exacerbate inequality between advanced and developing countries.

Data & Methods

• Methodological approach: Descriptive‑analytical and comparative. The study synthesizes recent international reports and integrates quantitative estimates with qualitative interpretation to assess cross‑country patterns.
• Primary sources cited:
  • IMF Working Paper WP/25/076 — quantitative estimates of AI’s growth impact by country income group.
  • World Bank 2025 report — digital usage metrics (e.g., ChatGPT traffic share) as proxies for adoption.
  • ILO 2025 report — job exposure metrics and occupational transformation analysis.
• Analytical focus:
  • Macroeconomic channels: output, total factor productivity (TFP), investment flows, and sectoral reallocation.
  • Labor channels: task changes, occupational transformations, demand for new skills, potential displacement risk by sector and skill level.
  • Development channels: digital infrastructure, absorptive capacity, access to AI tools, and the digital divide between advanced, middle‑income, and low‑income countries.
• Limitations noted:
  • Reliance on cross‑sectional and modelled estimates from international reports rather than longitudinal causal identification.
  • Rapidly evolving technology and usage patterns mean estimates are provisional and sensitive to policy and investment choices.

Implications for AI Economics

• Distributional risk vs. opportunity:
  • AI can raise aggregate productivity and growth, but the realized gains depend on countries’ human capital, digital infrastructure, institutions, and adoption policies — advantaging already advanced economies unless targeted interventions occur.
• Policy priorities to promote inclusive gains:
  • Human capital: large‑scale reskilling/upskilling programs focused on AI‑complementary skills and occupational mobility support.
  • Digital infrastructure: invest in broadband, cloud access, and affordable computing in developing countries to raise absorptive capacity.
  • Technology diffusion: policies to foster technology transfer, local AI R&D, public–private partnerships, and affordable access to AI tools for SMEs and public services.
  • Redistribution and safety nets: strengthen social protection, wage insurance, and targeted redistribution to mitigate transition costs.
  • International cooperation: coordinate on AI governance, financing for digital infrastructure in low‑income countries, and norms for data flows and intellectual property that do not unduly concentrate rents.
  • Measurement and monitoring: improve cross‑country data on AI adoption, task changes, productivity effects, and distributional impacts to guide policy.
• Research implications:
  • Need for causal micro‑ and macro‑studies on AI adoption effects in developing country contexts.
  • Better metrics linking digital usage (e.g., platform traffic) to economic outcomes, and studies on how policy interventions alter the distribution of AI gains.
• Short takeaway for policymakers:
  • Treat AI as both an opportunity and a structural challenge: act now to build skills and infrastructure, design policies to capture and share productivity gains, and engage in international efforts to prevent an AI‑driven widening of global inequality.