ILO AI Divide Study 2026: 135 Countries Face Disruption First

What the ILO AI Divide Study 2026 Actually Found

The paper, titled Disruption without Dividend? How the Digital Divide and Task Differences Split GenAI's Global Impact, was authored by Paweł Gmyrek, Mariana Viollaz, and Hernan Winkler. It serves as a background study for the World Development Report 2026, making it one of the most trusted studies on how AI reshapes jobs.

The key findings push back on the idea that AI hits all economies the same way. In high-income countries, 30% to 32% of jobs are directly exposed to generative AI. In low-income countries, that number drops to 10% to 15%. That sounds like good news for developing economies. But look closer at which jobs are exposed.

The jobs most vulnerable to AI automation in developing nations are clerical and administrative positions. These are not low-skill roles. In lower-income economies, they represent some of the best available pathways to decent work, stable income, and career growth. They have historically been especially important for women and young workers entering formal employment for the first time.

The Connectivity Paradox: Online Enough to Lose, Offline Enough to Miss Out

Perhaps the most striking insight from the ILO AI divide study 2026 is what the researchers call the connectivity paradox. Workers in jobs at risk of AI automation tend to have enough internet access — even in low-income settings — to face displacement. Their employers can adopt AI tools that cut clerical tasks. It does not matter whether the broader economy is digitally advanced.

Meanwhile, workers in roles that could benefit from AI-powered productivity gains often lack reliable internet access. Think of a farmer who could use AI weather forecasts. Or a teacher who could tap AI tutoring tools. Or a small business owner who could use AI-powered accounting. These workers are cut off from the gains. Their digital systems simply cannot support the technology.

This creates a cruel irony: developing economies are connected enough to experience AI's destructive effects but not connected enough to capture its productive benefits. As the Campus Technology analysis put it, "richer countries face greater exposure to AI-driven changes than developing countries, which are less exposed to AI but risk being left behind."

That same gap between awareness and access already plays out across generations. In wealthier nations, baby boomers are learning AI. In developing economies, their peers remain locked out entirely.

The Global Numbers Tell a Troubling Story

Data from multiple sources paints a consistent picture of a widening AI divide. Microsoft's AI Diffusion Report found that generative AI adoption in the Global North reached 24.7% of the working-age population by the second half of 2025. In the Global South, that figure stood at just 14.1%. The gap is growing: adoption in the Global North increased by 1.8 percentage points in six months, compared to just 1.0 percentage points in the Global South.

The Digital Cooperation Organization's Digital Economy Trends 2026 report identifies the AI divide as an "emerging trend" with a projected 2.5-year timeline. That means the lasting damage of unequal AI access could become locked in within a few years if leaders do not act now.

An International Monetary Fund (IMF) staff discussion note published in January 2026 adds another dimension. It found that about 1 in 10 job vacancies in advanced economies now demands at least one AI-related skill. In emerging market economies, that ratio is roughly half. Jobs that require AI skills tend to pay higher wages. But the spread of those skills is tied to fewer jobs in roles with high automation exposure. In short, AI skills create opportunity and destroy it at the same time — depending on where you live.

"Workers in positions vulnerable to automation typically maintain sufficient internet connectivity to experience displacement effects even in low-income settings, while those who could benefit from GenAI augmentation face substantial digital gaps that may block them from reaching productivity gains." — ILO Working Paper 166

Why This Matters for Technology Democratisation

The ILO AI divide study 2026 strikes at the heart of a question that defines MW3.biz's mission: who gets to benefit from technology? For years, the technology industry has promised that AI would be the great equaliser. Cheaper than hiring consultants. Faster than traditional training. Available to anyone with a smartphone. The ILO data suggests that promise is being broken.

The problem is not that AI itself is exclusionary. The problem is that what you need to benefit from AI — reliable broadband, digital skills training, affordable devices, and support from institutions — stays locked up in wealthy nations. As we explored in our coverage of Geoffrey Hinton's UN speech on AI regulation, the question is not whether AI should exist but who gets to steer it. The ILO data adds a further question: who even gets to ride in the car?

The concept of vibe coding and AI democratisation only works if people have the connectivity and tools to access these platforms. A no-code AI builder is meaningless if the internet connection drops every 20 minutes. A free AI productivity assistant does not help a worker who has never been trained to use one.

The MW3.biz approach to democratising the digital economy has always centred on making enterprise-grade technology accessible to individuals and small businesses. The ILO findings confirm that this mission is more urgent than ever.

What Needs to Happen Next

The study outlines several policy recommendations that governments and organisations should prioritise:

Expand digital connectivity. The most basic requirement is ensuring that workers who could benefit from AI augmentation — using AI to enhance rather than replace human work — actually have reliable internet access. The 66.9 million workers currently locked out of AI productivity gains represent an enormous untapped resource.

Invest in skills development. Digital skills and analytical thinking — problem-solving, critical thinking, data reading — form the base for AI-powered productivity. Without training aimed at workers in developing economies, the skills gap will grow right alongside the tech gap.

Strengthen labour market institutions. Safety nets need to be updated for an AI-disrupted world. Workers who lose office jobs to automation need real support and retraining paths — not just sympathy in the statistics.

Implement gender-sensitive policies. Because women hold an outsized share of clerical and administrative jobs in developing economies, AI-driven displacement will have an outsized impact on women's economic participation. Policies must account for this.

Adapt exposure measures to local realities. The study found that standard AI exposure measures consistently overstate AI's impact in developing countries. They assume workers do the same tasks in the same job title no matter where they are located. In reality, workers in lower-income economies perform fewer complex problem-solving tasks and rely less on computers, reducing the scope for both automation and augmentation.