From AI consumers to AI shapers: What the World Bank’s warning means for the Western Balkans

The World Bank’s Digital Progress and Trends Report 2025: Strengthening AI Foundations warns that countries without strong digital foundations risk being “passive consumers rather than active shapers of the AI economy.” At NarativAI, we look at what this means for regions like the Western Balkans: how prepared are our institutions, media, education systems and policymakers to move from using AI to actually shaping it?

Author: NarativAI

As artificial intelligence becomes a core driver of economic growth and public sector transformation, middle-income regions such as the Western Balkans face a strategic choice: remain passive users of foreign technologies or build the capacity to adapt and shape AI according to local needs.

The World Bank’s Digital Progress and Trends Report 2025: Strengthening AI Foundations warns that countries without strong digital foundations risk being “passive consumers rather than active shapers of the AI economy.” The report classifies most middle-income economies in a “medium readiness” category, meaning they have achieved basic digital adoption but still lack the infrastructure, data, and skills needed to influence how AI is developed and deployed.

“Compute is becoming the new electricity of the AI era,” the report states, pointing out that high-income countries host 77 percent of global data-center capacity. By contrast, middle-income regions depend largely on foreign cloud providers, placing critical public, economic and research data under external legal and commercial control. For smaller economies such as those in the Western Balkans, this raises questions of digital sovereignty and long-term strategic autonomy.

The World Bank frames AI readiness around four foundations, known as the “4Cs”: Connectivity, Compute, Context and Competency.

On connectivity, the report notes that while mobile coverage is nearly universal, gaps in broadband speed, reliability and data consumption between high- and middle-income countries continue to widen. AI applications, particularly those using real-time data and large models, require low-latency, high-capacity networks that many regions are still struggling to provide, especially outside major urban centers.

On compute, the concentration of data centers and high-performance computing in a handful of countries creates what the Bank calls a new strategic vulnerability. Middle-income governments are advised to assess options ranging from domestic data-center investment and regional cooperation to secure “data embassy” arrangements that allow sensitive public data to be stored abroad under national legal jurisdiction.

The third pillar, context, refers to locally relevant data, languages and institutional settings. The report highlights that more than half of open-source AI training datasets are in English, leaving many languages with smaller digital footprints under-represented in the systems increasingly used in education, administration and media. “AI model capabilities depend on the quantity, quality and diversity of training data,” the World Bank notes, warning that tools trained primarily on data from high-income, English-speaking environments may fail to reflect local realities in multilingual, middle-income societies.

Finally, competency—digital and AI skills—remains a major bottleneck. Although demand for AI-related jobs is growing faster in middle-income countries than in rich economies, the supply of advanced skills is limited and talent outflows toward global technology hubs persist. The report stresses that “competency is a prerequisite for meaningful AI participation,” calling for education reform, industry-aligned training and policies to retain skilled professionals.

At the same time, the World Bank identifies “small AI” as a realistic entry point for regions that cannot compete in building large, frontier models. These affordable, task-specific tools—often running on mobile phones or standard computers—are already being used in developing and middle-income countries to support teachers, farmers, nurses and small businesses. In such contexts, AI acts as a “co-worker or coach rather than a replacement,” delivering immediate productivity gains without massive infrastructure investments.

However, the report cautions that small AI alone is not enough to transform economies. Without sustained investment in the four foundations, countries may remain dependent on external platforms and vendors, unable to influence standards, governance or the direction of technological development.

“AI’s benefits will not be automatic,” the World Bank concludes. “They will depend on deliberate public policy choices to strengthen digital foundations, ensure local relevance, and manage the concentration of power.”

For the Western Balkans, this implies that the challenge is not simply to adopt AI tools, but to build the connectivity, compute capacity, local data ecosystems and skills that would allow the region to move from being a user of imported technologies to an active participant in shaping how AI serves its economies, institutions and languages.

(This text was written and reviewed by the editor with support from artificial intelligence tools for language editing and stylistic refinement. More on how NarativAi uses AI — Link)