Mark Suzman sees UAE path to bridging AI's global wealth divide

Gates Foundation chief executive Mark Suzman pointed to the UAE as a model for how artificial intelligence can serve people across the world rather than deepen the divide between rich and poor nations, as the foundation released its latest report on AI.

The foundation’s concerns run deep. Bill Gates recently published a lengthy essay warning of the dangers that unregulated AI poses to the social fabric. Yet Suzman’s comments suggest the organization believes the trajectory is not fixed. The choices made now about how AI is developed and deployed will determine whether the technology narrows or widens the gap between nations.

Additional reference context is available at https://www.thenationalnews.com/future/technology/2026/09/15/gates-foundation-ai-uae-access/.

A concrete partnership illustrates this thinking. During the COP28 climate conference in 2023, the Gates Foundation and the UAE announced a joint initiative called Aim for Scale, each committing $100 million to spur agricultural innovation and reduce hunger. Suzman highlighted that collaboration as evidence that the UAE understands how technology can address real human needs. The country’s Mohamed bin Zayed University of Artificial Intelligence has undertaken work he described as particularly promising, including research into AI’s capacity to prevent disease and improve weather forecasting.

That commitment to practical application shapes the foundation’s broader strategy. The organization announced it would spend at least $1 billion over the next two years to expand access to AI and the solutions and knowledge that have remained out of reach for much of the world. The new report, titled Make this Matter AI, lays out which priorities must guide that spending.

The stakes are substantial. Unlike previous technological revolutions that typically compressed inequality over time, AI risks moving in the opposite direction. The specific areas requiring urgent attention include language diversity in AI systems, technology access in lower-income regions, basic infrastructure development, and the application of AI to agriculture and environmental challenges.

“The risk is, rather than narrowing, which is usually what happens with a new technology, it might widen,” Suzman said. Without deliberate action, inequities in access to health, education and agricultural productivity services could intensify across the developing world.

Language presents an immediate problem. Despite advances in multilingual AI language models in recent years, a 2025 study from the UAE’s Mohamed bin Zayed University of Artificial Intelligence found that these systems still perform better in English than in other languages. This gap matters enormously for regions where English is not the primary tongue. Africa alone contains more than 2,000 languages; if AI is to improve health outcomes on that continent, the technology must understand not just formal language but dialects, accents and slang as well.

The UN estimates that Arabic is spoken by approximately 400 million people worldwide, yet it was not a priority during AI’s initial development phase. In 2025, the UAE’s government-backed Technology Innovation Institute moved to address this gap by introducing Falcon Arabic, a model described as fluent in the language. The Gates Foundation’s report notes that models continue to show lower performance in low-resource languages like Amharic, Arabic and Yoruba when handling questions about science and technology.

Suzman believes philanthropic partnerships can help ensure languages are not left behind. The Gates Foundation works with Microsoft, Google and others to assemble language datasets, and through its new report the foundation is calling on additional technology companies to invest in broadening language availability. Once a language is properly integrated into AI systems, Suzman noted, the work need not be repeated. It becomes a public good available to the entire field.

Meanwhile, in developed nations like the United States, considerable debate surrounds whether open-source or closed AI models represent the superior approach. Suzman suggested that developing countries cannot afford to spend time on that debate. What matters instead is whether AI tools use educational datasets safely and securely, and whether people can access AI at the lowest possible cost. Outcomes and access, he said, are what truly count.

The foundation may announce additional initiatives on the language front in coming weeks, Suzman indicated. More broadly, he argued that technology companies and governments share an interest in getting involved.

The open question hanging over all of this is whether the next two years of spending, partnerships and model-building will move fast enough to outpace the widening that Suzman fears is already under way.