AI and Inequality: How Automation Is Deepening the Global Wealth Divide

By Jingyi Wu, China Sep. 24, 2025

The promise of artificial intelligence (AI) was once bold: to elevate productivity, democratize opportunity and unlock the next frontier of global growth. But a troubling reality is emerging — AI is not just transforming work, it's deepening the divide between the winners and losers of the digital economy. Wealth is consolidating at the top, labour income is stagnating, and entire regions are being left out. Unless policy intervenes decisively, the AI revolution risks becoming the next chapter in the story of global inequality.

Uneven Gains: Who Benefits from AI?

Much of the value created by AI flows to those who own the algorithms, the data infrastructure and the capital — not necessarily those who perform the work. According to a working paper by the International Monetary Fund, “The Global Impact of AI: Mind the Gap,” AI adoption could allow advanced economies to grow more than twice as fast as low-income countries, all else equal. IMF Within countries, the same pattern holds: workers in high-skill roles, with access to AI tools, stand to see large productivity gains; low-skill, routine workers may see their tasks automated or downgraded. IMF+1

The result: instead of broadly shared prosperity, we're seeing concentration. A 2025 study found a significant correlation between AI capital-stock accumulation and increasing wealth disparity. ScienceDirect In plain terms — if you invest in AI, you get richer; if you depend on manual or routine tasks, you might get poorer.

Between Countries: The New Digital Divide

The digital divide is no longer just about connectivity; it’s about capacity to harness AI. Countries with advanced tech ecosystems, plentiful data, strong institutions and venture capital are poised to reap disproportionate gains. On the flip side, many developing economies are stuck in low-value production, service outsourcing or manual labour — all highly exposed to automation. The IMF paper warns these differences in AI preparedness, data access and institutional capacity “are unlikely to fully offset” the inequality gap. IMF

In fact, a piece from the World Economic Forum notes that nearly one-third of the global population — about 2.6 billion people — lack reliable internet access, cutting them off from the core of the AI economy. A mere 10 percent increase in broadband penetration in a developing country boosts GDP growth by about 1.4 percent. World Economic Forum Without infrastructure, skills and capital, many nations face a future where they supply raw data and labour — while the AI value accrues elsewhere.

Within Countries: Capital vs Labour in the AI Era

One of the defining features of the AI economy is that capital (machines, algorithms, data centres) is increasingly substituting or augmenting labour in tasks that used to be human-driven. The result: a shift in income from labour to capital owners. Economists term this the declining labour share. In recent years, labour’s share of income has been trending downward globally — and AI may accelerate that trend. Massachusetts Institute of Technology

Moreover, jobs that are high wage and knowledge-intensive are getting enhanced by AI, further boosting inequality. A commentary from the Center for Global Development highlights that AI may not just increase within-country inequality — it could hinder between-country convergence too. World Socialist Web Site When AI tools replace routine outsourcing jobs in developing countries, the ladder to middle-income status is removed. The “work your way up” path becomes blocked.

Automation Risk: Who’s Vulnerable?

In many low- and middle-income countries, labour markets are heavily weighted toward routine and low-skill service jobs — call centres, data entry, outsourcing workflows. These are exactly the tasks most exposed to AI and automation. For example, a report from AP News at the 2025 Global AI Summit for Africa found that women in Africa’s outsourcing sector are 10 percent more likely than men to have their tasks automated by 2030. AP News

The danger is clear: entire workforces, entire regions, could see their livelihoods vanish or be drastically degraded as AI matures. And unlike previous waves of automation, the transition may be faster and less accompanied by new job creation.

Policy Gaps: Untouched and Unprepared

Despite the scale of the threat, policy response remains weak. Some countries are scrambling to introduce AI strategies, but many lack the institutional capacity to regulate effectively, ensure equity, or protect vulnerable workers. The IMF note points out that when firms decide how much AI to adopt, the wealth inequality effect becomes especially pronounced — because firms with deeper pockets adopt faster and capture more of the surplus. IMF

What’s missing: mechanisms to ensure that AI doesn’t merely replace human labour with machine capital without shared gains. What’s also missing: coherent global coordination to ensure developing countries are not permanently left behind in the AI race.

Toward a Fairer AI Economy: Key Principles

If left unchecked, AI may lock in a system of winners and losers that spans both countries and classes. But it doesn’t have to be this way. Policymakers, international institutions and the private sector can act now. Three principles should guide the effort:

1. Shared ownership and redistribution. Because so much value accrues to capital owners, taxation of AI gains — especially monopoly rents — could provide revenue for universal or targeted redistribution. Some scholars propose that AI capital profits could fund universal basic income (UBI) or social dividends. arXiv

2. Digital and AI inclusion. Ensure that countries and communities currently excluded from the AI economy get access to infrastructure, data, skills training and institutional support. A 10 percent increase in broadband in developing economies raised GDP growth by 1.4 percent, showing what inclusion can do. World Economic Forum

3. Labour-centric regulation. Strengthen rights, retrain workers for higher-value tasks and regulate algorithmic bias. AI governance must include labour voices, transparency, and ensure technology augments rather than replaces human work. The Brookings Institution found that high-skilled workers appear most likely to benefit from AI; that means the lower skilled need protection. Brookings

Conclusion

The AI revolution holds extraordinary promise — but promise alone is not enough. Without conscious policy choices, AI risks becoming the engine of a new era of inequality. Wealth will accumulate where infrastructure, data and capital are concentrated. Workers everywhere — particularly those lacking high skills or located in regions without digital readiness — may find themselves left behind.

In the end, the question is not whether AI will reshape the economy; it will. The question is: who will benefit from that reshape. Will the rewards be broadly shared — lifting incomes, improving livelihoods and narrowing divides? Or will they go to a narrow elite, deepening global fault lines?

The answer will depend on the choices we make today. If inclusion is ignored and redistribution delayed, AI may become the newest amplifier of inequality. If not, it may instead become a tool of shared prosperity.

Works Cited

Cerutti, Eugenio, et al. “The Global Impact of AI: Mind the Gap.” International Monetary Fund Working Papers, no. 25/76, Apr. 2025, https://www.imf.org/publications/wp/2025/04/04/ai-adoption-and-inequality-565729.

“How AI Can Enhance Digital Inclusion and Fight Inequality.” World Economic Forum, 2025, https://www.weforum.org/stories/2025/06/digital-inclusion-ai/.

“The Simple Macroeconomics of AI.” Shaping Work (MIT), 2025, https://shapingwork.mit.edu/research/the-simple-macroeconomics-of-ai/.

“AI and Income Inequality in the United States.” Brookings Institution, 2024, https://www.brookings.edu/articles/ais-impact-on-income-inequality-in-the-us/.

“AI Boom Risks 40% of Jobs, Deepens Inequality — UN Report.” eWeek, 7 Apr. 2025,

 https://www.eweek.com/news/un-ai-report-inequality-jobs/.