AI's Role in Deepening Global Inequality
A growing divide reveals who actually benefits from AI progress, and who gets left behind.
Key points
- South Africa withdrew a draft AI policy in April 2026 after fake academic citations were discovered in it.
- Indonesia launched a target of training 100,000 AI-skilled workers each year, announced in August 2025.
- In 2023, the US held roughly 87% of global cloud computing and data storage exports.
- Across OECD countries, only 40% of adults have more than basic digital problem-solving skills.
How does AI increase inequality?
AI's infrastructure isn't spread evenly: the United States alone holds more than 5,000 data centers, a figure Stanford University's 2026 AI Index puts at over ten times the count of any other single country. Because so much AI processing now runs on cloud platforms rather than local hardware, that geographic concentration becomes a concentration of dependency. World Bank figures from 2023 put the US at around 87% of global cloud computing and data storage exports. For countries in sub-Saharan Africa or Southeast Asia, AI development is commercially and geopolitically outside their reach.
Are countries taking steps to bridge the gap?
Some are trying. South Africa's Department of Communications and Digital Technologies published a draft national AI policy in April 2026, proposing new oversight bodies. It was pulled days later after a journalist found that at least six academic citations in the document did not exist, apparently hallucinated by an AI tool. The minister called it "an unacceptable lapse." Indonesia's National Research and Innovation Agency has taken a different path, building practical tools for communities that commercial AI ignores: a satellite-assisted app helping artisanal fishermen locate fish, multilingual models covering Javanese and Sundanese, and AI chatbots in government services. Its national AI roadmap, released in August 2025, set the 100,000-workers-a-year training target.
What are the implications for skills and education?
The skills gap is stark. In OECD nations, 36% of adults with a university degree reported taking AI-related training in the past year; among those who stopped at upper-secondary school the figure was 18%. UNESCO sees growing efforts to bring AI into classrooms, but support at primary level remains patchy. Those with strong schooling and reliable connectivity can use AI as a tool. Everyone else is more likely to encounter it as an opaque system deciding things about them.
Common questions
How does AI affect economic opportunities?
Countries that don't build their own AI capacity risk losing local jobs and innovation. They also risk getting systems that weren't designed for their languages or priorities.
Is there any regional cooperation in AI?
Yes. African ministers adopted a Continental AI Strategy and an African Digital Compact in 2024, and the April 2025 Global AI Summit on Africa in Kigali examined how regional coordination, local-language models, public universities, and open-source tools might reduce long-term dependence on externally built systems. The divide, as this beat makes plain, is as much about who shapes AI as who can access it.
