Beyond Connection: How Does AI Widen an Old Divide?

In 2025 about 6 billion people used the internet, roughly three-quarters of humanity, while 2.2 billion remained offline, most of them in low- and middle-income countries (International Telecommunication Union, November 2025). Anyone who stops at that figure may conclude that the cure is cables and towers. Yet the same report notes that mobile broadband coverage is almost universal, and still these people remain outside the network. The distance is not only geographical.

Access Does Not Guarantee Benefit

The ITU figures show that use follows income, place and sex: 94% of people in high-income countries use the internet, against 23% in low-income countries. In towns and cities the figure is 85%, in rural areas 58%. Among men it is 77%, among women 71%. A user in a high-income country generates nearly eight times as much mobile data as a user in a low-income one. Being connected does not mean having the same opportunities as others.

Three Layers

The divide can be read in three overlapping layers. This is an analytical framing, not an official classification:

• Access: a device, a network and electricity at an affordable cost.
• Use: the skill that turns the tool into work, learning or output.
• Production: the ability to be a maker of the model, the data and the application, not merely a consumer of what others make.

The first layer is the easiest to treat and the third the hardest. Policies are often counted as successes because they addressed the first alone.

What Does AI Add?

The International Monetary Fund observes that rich countries are better placed to benefit from AI because their infrastructure and skills are stronger. It estimated that exposure to AI reaches about 60% of jobs in advanced economies, 40% in emerging markets and 26% in low-income countries. A lower share is no mark of safety; it means those countries are less prepared to benefit, and the Fund warns that inequality among nations may grow over time. Its preparedness index, applied to 125 countries, found richer countries better equipped on average, with wide variation among them.

Microsoft's reading of its own usage data points the same way. In the first quarter of 2026, the share of people aged 15 to 64 who had used a generative AI product was about 27.5% in the Global North and 15.4% in the Global South, and the company describes the gap as widening. The UAE leads the list at 70.1%. Microsoft cautions that the measure is built on its own services' data after adjustment, and that no single metric is perfect; the figures indicate a direction, not a final verdict.

Language Is a Hidden Layer

Models are built on what has been written in a given language. In the W3Techs survey (22 October 2026), 49.5% of websites whose language is known use English and only 0.6% use Arabic. That does not mean Arabic content as a whole is that small, since the measure counts websites, not texts, but it points to an imbalance in presence. Its effects are discussed in the article "Arabic Artificial Intelligence: Do We Make or Consume? Sovereignty and Arabic Language Models", and its link to fairness in "Algorithmic Bias".

What Closes the Divide?

Many hope that late-comers will leap over stages, as some countries went straight to mobile phones without landlines. But a leap does not happen automatically. It requires the following:

• Public infrastructure: connectivity and computing at a cost people can afford, treated as a public utility.
• Education for conscious use: learning to ask, verify and criticise, not memorising tools.
• Local content and data: verified Arabic corpora that make our language a maker, not a subject.
• Fair governance: that models and data not be concentrated in a few hands, and that those who hold power be held to account.

Individuals and institutions can begin with what is within reach: training staff, opening tools to the weakest groups, and measuring impact by what people gain, not by what is spent. We take the means, and we ask God for success.