A country with fewer than seven million working-age residents is outpacing the United States in artificial intelligence adoption by a factor of more than two to one. That is the central finding of a research paper covering 147 economies, published on the arXiv preprint server, which places the UAE at 70.1 percent AI usage among working-age adults and the U.S. at 31.3 percent. The gap is large enough to force a basic question: does national strategy matter more than market size when it comes to getting populations to actually use AI?
Where the numbers come from
The paper introduces a metric called “AI User Share,” which divides the number of active AI users in a country by its working-age population (adults 15 to 64, as defined by World Bank data). That per-capita approach lets researchers compare a compact Gulf economy against a continental one without raw user counts overwhelming the picture.
The usage signal comes from anonymized Microsoft telemetry, meaning it captures interactions with Microsoft-powered AI features, not every generative AI product on the market. The authors adjust for device access so that countries with lower PC or smartphone penetration are not mechanically penalized. Without that correction, lower-income economies would appear to have near-zero adoption simply because fewer residents own qualifying hardware.
Two things are worth flagging immediately. First, the paper is a preprint. It has not yet cleared formal journal peer review, though its methodology is described in detail and the arXiv platform is maintained by major research institutions. Second, because the telemetry is Microsoft-only, the metric cannot capture adoption of competing tools like Google’s Gemini, Anthropic’s Claude, or open-source models. In a market where Microsoft products dominate enterprise software, the metric could overstate a country’s lead. In a market where alternatives are popular, it could understate real usage.
What the UAE has been building
The UAE’s high score did not appear out of nowhere. In October 2017, the country appointed Omar Sultan Al Olama as the world’s first Minister of State for Artificial Intelligence. A formal National AI Strategy followed, targeting integration across nine sectors including transport, health, energy, and education. Government portals began embedding AI-powered features into routine citizen services: renewing a license, paying a utility bill, navigating customs paperwork.
That top-down push sits on top of a distinctive demographic base. Expatriates make up roughly 88 percent of the UAE’s population, according to government statistics, and a large share work in white-collar sectors like finance, logistics, consulting, and technology. These are exactly the roles where AI copilots and automation tools see the fastest uptake. A young median age (around 33) and high smartphone penetration further lower the friction.
None of this proves the 70.1 percent figure is perfectly accurate. No independent UAE government survey or statistical-agency report has publicly confirmed that rate through a separate methodology. But the policy scaffolding and demographic profile make a high adoption number at least plausible in a way it would not be for every country.
Why the U.S. number is lower than you might expect
The United States builds many of the foundational AI models and platforms the world uses. Yet the 31.3 percent adoption figure, while still representing tens of millions of active users, suggests that access alone does not translate into widespread daily use.
Survey data from the Pew Research Center, published in March 2024, found that American ChatGPT usage was growing but that trust in the tool’s outputs remained low, particularly around sensitive topics like elections. That trust deficit may act as a brake. When workers are unsure whether an AI tool’s output is reliable, they are less likely to fold it into daily routines, even if the software is sitting right in front of them.
Structural factors compound the hesitation. U.S. AI deployment is fragmented across thousands of companies and agencies with different risk appetites. Regulatory uncertainty at the federal level, sector-specific compliance rules in industries like healthcare and finance, and a patchwork of state-level AI legislation all slow coordinated rollout. The result is a market-led diffusion pattern: fast among early adopters in tech and professional services, much slower in education, government, and small business.
The Stanford Institute for Human-Centered Artificial Intelligence (HAI) has documented a similar pattern in its annual AI Index Report, noting that U.S. private-sector AI investment leads the world but that adoption rates vary enormously by industry and firm size. The AI User Share metric, viewed alongside Stanford’s findings, paints a consistent picture: America’s AI ecosystem is deep but unevenly distributed.
What other countries show up in the data
The paper covers 147 economies, and while the UAE and U.S. figures have drawn the most attention, the broader ranking matters for context. Small, digitally advanced nations with strong government technology strategies, such as Singapore, Denmark, and Israel, tend to cluster near the top of per-capita AI adoption metrics across multiple indices. If those countries also score high in the AI User Share dataset, it would reinforce the idea that coordinated national strategy is a common thread. If they do not, it would raise questions about whether the UAE’s result reflects something specific to Microsoft’s market position in the Gulf.
The full country-level dataset has not been released beyond the paper’s methodology description, which limits independent replication. Researchers outside the authoring team cannot yet test alternative adjustment models or verify individual country scores. That is a meaningful gap. Until the underlying data is more widely available, the 147-economy ranking should be treated as a credible first draft rather than a settled league table.
What this means for businesses operating across borders
For companies planning regional AI rollouts, the practical signal is straightforward even if the exact percentages shift with future research. Markets where governments actively embed AI into public services and enterprise infrastructure tend to produce populations that are more comfortable using AI tools at work. That has implications for product design, training investment, and go-to-market sequencing.
A multinational launching an AI-powered workflow tool might treat the UAE, Singapore, or similar markets as early-adoption environments where users need less onboarding and more advanced features. In the U.S. or Europe, the same company might invest more heavily in trust-building, explainability features, and gradual integration with existing software stacks.
For policymakers, the comparison is less about winning a ranking and more about identifying which levers actually move adoption. Public procurement that requires AI-enabled services, national digital identity systems that reduce friction, and workforce training programs tied to specific tools all appear in the playbooks of high-adoption countries. The U.S. has the talent, the capital, and the infrastructure. What it may lack is the coordinated push that turns availability into habitual use.
Where the evidence stands as of mid-2026
The honest summary is that the AI User Share metric is one of the broadest attempts yet to measure AI diffusion on a population-normalized basis, and its finding that the UAE leads the world is striking. But the evidence base remains thin in important ways. The telemetry is proprietary and Microsoft-only. The preprint has not completed journal review. No independent audit of the UAE’s score exists. And the Pew data used to contextualize U.S. adoption is more than two years old.
What the data does establish, even with those caveats, is that scale alone does not guarantee rapid AI uptake. The world’s largest consumer market, home to the companies that built the generative AI wave, is being outpaced in per-capita adoption by a Gulf state with a fraction of its population. Whether that gap narrows as American enterprises accelerate deployment, or widens as the UAE deepens its national AI infrastructure, will depend on decisions being made right now in boardrooms and government offices on both sides. The numbers will catch up to those decisions soon enough.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
