Views: 0
The first ever comprehensive AI assessment was made in the World Development Report, revealing the latest knowledge in the digital transitions and AI for economic progress in national growth and business. The report assesses the ways the AIs can, mainly, help low- and middle-income countries in accelerating growth and public services, provided they bridge foundational gaps rather than trying to build costly frontier models; the assessments are vital for the rest of the world too.
Background
At a time when development progress has dipped to its weakest pace in 75 years, artificial intelligence (AI) offers developing economies a rare path to greater prosperity. World Development Report-2026 (WDR, titled “The Promise of Artificial Intelligence”) provides the first comprehensive assessment of what AI means for these economies: i.e. how firms and governments are using the digital tools, what is holding them back, where the largest gains may lie and how the risks can be managed.
AI could help lift global growth to its strongest pace since the golden years of the 2000s, delivering tangible benefits to people in all states, and in developing economies as well.
The recent WDR’s edition shows that even if AI fails to live up to all its hype, developing economies would still be better off than they are today: the worst-case estimates suggest that potential growth rates in developing economies would rise above the dismal average of the first half of the 2020s. At a time when development progress globally has dipped to a 75-year low, the digital transition seems the only feasible solution to perspective growth: “such an opportunity might not come again any time soon”, notes the report.
AI is spreading much faster and becomes more context-specific than earlier general-purpose technologies like steam power, electricity and the internet. That acceleration calls for policies that poor countries could presently afford and implement; however, most developing economies do not yet have the needed facilities to develop AI technologies of their own, so advancement cannot be a big part of an AI strategy for them. Hence, the context-specificity of AI-based technologies defies simple import-and-adopt approaches; and the WDR recommends strategies in which adaptation is the main direction.
To extend medical, judicial, educational and agricultural extension services, the Report points to predictive AI in back-end functions instead of generative AI at the frontline of service delivery. To mitigate against the danger of becoming overly reliant on technologies being developed in a single country, the report advises making interoperability a feature instead of aiming for sovereignty. To balance the imperatives of innovation and protection, the report advocates encouraging industry to debate and develop voluntary standards. And perhaps most important, it recommends an emphasis on small AI applications diffused widely across the economy instead of big investments in large language models concentrated in a few firms; i.e. using text messages, voice calls and a variety of other low-cost but sharp instruments for clinics, courts, schools, farms and small businesses.
General source and citations from: https://www.worldbank.org/en/publication/wdr2026
WDR’s outcomes
Countries in the world must adapt AI to local languages, institutions, data and development needs; they have to ensure that national systems can work seamlessly with multiple platforms and providers; and steadily build the skills and infrastructure needed to do more. Low-cost tools, so-called “small AIs”, can put scarce expertise within reach of millions, through text messages, voice calls, basic phones, and other technologies that work even where electricity, computing power, and internet access are limited. The gains are already visible. AI is helping to accelerate medical screening, assisting farmers with more accurate weather forecasts, and aiding teachers in creating better lessons for students.
As AI models are becoming more powerful and pervasive, their values as tools for development depends on three factors. The first is capabilities: unlike common computers and/or internet, the AI can perform complex tasks that guide decision-making, though usually require human expertise, such as diagnosing diseases, forecasting weather or advising farmers. These skills are often in short supply in most developing countries, so AI has the potential to fill important gaps.
The second is the digital concentration: most advanced AI models (as well as chips and data centers, etc.) are controlled by a small number of companies in just a few states. Such a concentration creates dependency risks and affects the AI tools’ availability in meeting a country’s needs. However, this concentration also allows some states to customize AI models without having to spend billions in creating the most advanced AI systems from scratch.
The third is the digital complementation: GenAIs and agentic AIs can be used safely and effectively in national growth patterns. However, the AI works best with the reliable infrastructures, good education systems and strong governing institutions; developing countries often have shortages in all three. Moreover, the AI models have to be customized to the local context because the models are trained by the nation-wide data; hence, the quicker the countries build these “complements”, the faster AI improves development.
Among key WDR’s implications and findings are the following:
= In employment and productivity: only 4.5% of jobs in low- and middle-income countries face automation risk from generative AI, compared to over 14 percent in high-income countries.
= About 16 percent of jobs in developing economies can see meaningful productivity boosts through AI augmentation (the figure is about 19 percent in wealthy nations).
= The triple “adopt, adapt and advance” strategy is mostly applied for developing countries; they do not need to spend billions building massive, all-purpose AI models. Instead, they should adopt existing off-the-shelf and low-cost tools (so-called “small AIs”), adapt them to local languages, data and regional needs, and gradually advance capabilities over time.
= In crucial infrastructure gaps: AI benefits are not automatic; thus, growth depends on fixing foundational constraints like reliable electricity, high-speed connectivity, digital skills, local-language datasets and institutional governance.
= In national policy and preparedness: while over 80 countries had published national AI strategies by mid-2026, only one of the world’s 25 low-income countries (Rwanda) has done so.
= Governments must act as enablers of digital foundations, targeted tool users and regulators using voluntary standards to manage risks like bias and market concentration.
Digital governance
The report includes passages in AI governance: most states around the world increasingly see digital technologies as a “foundational force” shaping economic activity, national security and global influence. Control over key parts of such AI value chains as raw materials, advanced chips and data centers in most leading powers, especially the US and China, creates dependency risks for other countries.
The availability of open AI models and the falling price of proprietary AI models (albeit for older versions) make advanced AI more accessible; however, using them at scale still depends on data centers built on chips and commercial cloud services that are largely controlled by the major digital powers. As to developing countries, they can reduce dependencies by “sourcing these AI products and services” from different countries, choosing the most suitable available technologies.
But, as the AI systems depend on combining “different layers”, such a strategy can create interoperability problems and expose countries to conflicting regulations tied to technologies from different jurisdictions.
Bottom-line
It is vital still to remember that the AI tools and models reflect a general connotation on any computer systems that manage to perform human-like tasks: hence the evolution of GPT (generative pretrained transformer) originated in 2018, includes three key words assembled in a digital process that learns and crates new content. Thus, generative AI is a specific digital subset designed to create such contents – texts, images, music, codes, etc.; and most recognised GenAI is the ChatGPT by OpenAI. Some global digital gurus use their own models: e.g. Elon Musk uses Grok, developed by xAI, but he also uses third-party AI models like Anthropic Claude, Midjourney and Suno.
The WDR offers a clear framework in three AI’s approaches: adopt, adapt and advance for the road ahead to help countries capture AI’s benefits without wasting scarce resources or deepening dependence on any single supplier. Earlier technological revolutions left much of the developing world behind. This landmark edition of the World Development Report offers developing economies a roadmap to prevent a repeat of that outcome. Generally, regardless of national political economies, the AI models may in reality widen gaps in productivity, specifically between developing countries and high-income countries, where AI adoption is higher.
https://ppp.worldbank.org/sites/default/files/2026-08/World%20Development%20Report%202026%20The%20Promise%20of%20Artificial%20Intelligence.pdf
The issue of retaining digital specialist “at home” is vital around the world: e.g. the AI graduates in China are increasingly choosing to stay home instead of pursuing opportunities abroad, a move that could help the country catch up with the US in the digital transition and research. China-based AI scientists now dominate global AI-related publications, and the rise of firms such as DeepSeek and Moonshot AI has proven the capability of Chinese AI models. Competition for jobs is fierce and the pressure to publish is intense; however, if China would retain its top talents, it is feasible that the country could become the world AI-leader by the end of the decade. The AI ecosystem in China differs in several ways from that of California’s Silicon Valley. In China, the line between academia and industry is more porous, and access to computing resources is much lower; this presents both opportunities and challenges: e.g. Chinese firms have been accused of copying US advances in their race to catch up, but they also have found techniques that make AI models more efficient. Fundamentally, the US still makes the world’s most advanced AI: i.e. models from OpenAI and Anthropic, both based in San Francisco, California, top the leader boards for system capabilities.
Source: https://www.nature.com/articles/d41586-026-03223-8
More in: Hook D. DeepSeek and the New Geopolitics of AI (Digital Science, 2025).
AI is transforming national well-being in numerous ways: a) economically (by affecting what people do and how much they earn); b) socially (by changing how people access public goods and services); and c) politically (by reshaping power relationships between countries, between governments and corporations, between corporations and individuals and between citizens and states). Thus, for developing countries, the priority should be to use AI to expand people’s capabilities and improve opportunities; governments can help by making AI easier to access, ensuring that it works in local languages and is suited to local conditions; while investing in the digital infrastructure, skills and institutions needed to support it. At the same time, most states must reduce the risks of AI so that its benefits exceed possible risks and harms.