World Bank Says ‘Small AI’ Could Transform Developing Economies

‘Small AI’ models can run directly on mobile phones with low connectivity, offering tailored solutions for emerging economies.

TL;DR: Key Takeaways

• Targeted Technology: The World Bank is urging developing countries to develop 'small AI'—specialized, lightweight models designed to operate on mobile phones with limited connectivity.
• Accelerated Development: World Bank Chief Knowledge Officer Paschal Donohoe stated that adapting AI to local needs could compress centuries of economic development into decades.
• Practical Agriculture: In programs like India's AgriConnect, localized AI delivers soil analysis to smallholder farmers in their native languages, compensating for shortages of agricultural scientists.
• Infrastructure Gaps: IMF Managing Director Kristalina Georgieva emphasized basic preconditions, while the World Bank noted nearly a third of rural schools in sub-Saharan Africa lack electricity and over two-thirds lack consistent internet.
• Workforce Implications: Georgieva warned that 40% of global jobs could be augmented or displaced over time, with entry-level youth facing the steepest career hurdles.

At the annual meetings of the World Bank and the International Monetary Fund (IMF) in Bangkok, Thailand, global economic leaders have placed the developmental impact of artificial intelligence at the center of international policy discussions. The World Bank is actively encouraging emerging economies to steer away from total dependence on massive general-purpose models and instead prioritize the creation of 'small AI' systems tailored to immediate local needs.

What Is ‘Small AI’ and Why Does It Matter?

While leading Silicon Valley technology corporations focus heavily on building increasingly sophisticated consumer-facing products, the World Bank argues that emerging markets require a distinctly different approach. According to World Bank Chief Knowledge Officer Paschal Donohoe, artificial intelligence holds the potential to significantly accelerate economic progress in developing partner nations—shifting transformation timelines from centuries into decades—provided the technology is adapted to local realities.

Technically, small AI utilizes specialized artificial intelligence models that are trained on narrower, domain-specific datasets for dedicated purposes. Rather than relying on immense cloud computing facilities or continuous high-bandwidth connectivity, small AI models are engineered to operate effectively in resource-constrained environments.

Core Attributes of Small AI

• Hardware Accessibility: Engineered to run locally on standard mobile phone hardware without requiring high-end processors.
• Low Connectivity Operation: Capable of functioning in areas with intermittent, weak, or offline cellular networks.
• Localized Communication: Delivered in the native community languages and dialects of local populations.
• Targeted Expertise: Built on specific datasets designed to solve concrete operational problems rather than open-ended queries.

Real-World Deployment: Supporting Agriculture at the Grassroots

To demonstrate the practical necessity of smaller, focused models, Donohoe highlighted the agricultural sector in India, specifically smallholder farmers participating in the World Bank's AgriConnect scheme. In these rural communities, farmers require reliable, timely soil analysis presented in their native languages to optimize crop yields and manage fertilizer usage.

As Donohoe noted, there is an unavoidable human resource deficit: agricultural institutions cannot deploy sufficient agronomists and soil scientists to personally advise every small farm across India. In this context, mobile-accessible small AI can serve as a scalable bridge, delivering accurate diagnostic insights directly to farmers through handheld devices. Donohoe suggested that such solutions could be realized through structured partnerships connecting the World Bank, national governments, and private-sector technology developers.

Critical Preconditions: Overcoming the Infrastructure Deficit

Despite the substantial potential of small AI, leaders at the Bangkok meetings emphasized that software alone cannot solve foundational developmental challenges. Speaking during a panel on the technology's broader impacts, IMF Managing Director Kristalina Georgieva outlined the vital structural prerequisites that national governments must establish for artificial intelligence to deliver widespread benefits.

According to Georgieva, four foundational pillars are essential for successful implementation:

1. Access to Electricity: Consistent electrical power to charge and operate digital hardware.
2. Digital Infrastructure: Basic network connectivity and telecommunications backbones.
3. Public Adoption Readiness: Preparing communities and end-users with the digital literacy needed to adopt new tools.
4. Regulatory Protections: Government safeguards to protect citizens against malicious actors and fraudulent practices.

Georgieva observed that if governments and international partners establish these foundational basics, technology adoption has the potential to become highly inclusive, describing it as potentially 'as democratic as the cellphone.'

The Reality of Basic Public Infrastructure

The gap between technological potential and physical infrastructure remains substantial. The World Bank's annual World Development Report, released in August, highlighted acute infrastructural deficits across emerging regions. The report documented that nearly one-third of rural schools in sub-Saharan Africa operate without electrical power, and more than two-thirds lack consistent, dependable internet connections. These shortfalls present immediate obstacles to equitable technology deployment.

Workforce Challenges and Capital Mobilization

The economic transformation driven by artificial intelligence also brings complex labor market consequences. Georgieva warned that over time, approximately 40% of jobs worldwide are likely to be either augmented or eliminated by AI technologies. She noted that this shift poses acute risks for young workers entering the workforce, who traditionally build their careers by taking on routine, entry-level tasks to learn foundational skills and advance professionally.

To address development priorities amid constrained fiscal environments, the World Bank has placed a major focus on attracting private capital into emerging economies under the leadership of President Ajay Banga. Donohoe reported that the World Bank mobilised $120bn (£90bn) in private capital over the past year through guarantees and risk-mitigating financial instruments. This private-sector push comes at a critical juncture when several major donor nations, including the United States and the United Kingdom, have reduced their official foreign aid allocations.

Frequently Asked Questions

What does the World Bank mean by ‘small AI’?

Small AI refers to specialized artificial intelligence models trained on focused datasets for specific tasks. Unlike large general-purpose models, small AI is built to function locally on mobile phones with low connectivity and communicate in local community languages.

How could small AI speed up development in emerging markets?

According to World Bank Chief Knowledge Officer Paschal Donohoe, deploying targeted AI directly on mobile devices can compress economic progress that historically took centuries into decades by providing scalable expertise, such as agricultural guidance, directly to underserved populations.

What basic infrastructure is required for AI adoption according to the IMF?

IMF Managing Director Kristalina Georgieva stated that successful AI adoption requires access to electricity, reliable digital infrastructure, user readiness, and governmental protections against bad actors.

What proportion of global jobs could be affected by AI?

Kristalina Georgieva warned that roughly 40% of jobs globally are likely to be either augmented or eliminated by artificial intelligence over time, with particularly significant risks for young people entering the labor market.

Sources & Further Reading

• The Guardian