The World Bank believes that the future of digital development in emerging economies lies not in mimicking Silicon Valley’s giant models, but in building small artificial intelligence systems that meet local needs in agriculture and education.
Small AI strategy as an alternative to giant models
Technical advantages of embedded systems and agricultural applications in India
Bangkok meetings and infrastructure challenges in Africa and the labor market
Development financing and the implications of the strategy for Arab and African economies
Frequently asked questions
Small AI strategy as an alternative to giant models
The World Bank is encouraging developing countries to pursue the development and adoption of smaller, more specialized artificial intelligence systems so that they can operate directly on mobile phones, support local languages, and perform their functions efficiently in areas where internet connectivity is limited. This strategic recommendation appeared in an interview with Pascal Donohoe, the World Bank’s chief knowledge officer, published by the British newspaper “The Guardian” on October 18.
Donohoe argued that emerging economies should not assume they are obliged to replicate the strategy of major American tech companies, which are investing massive sums in general-purpose AI models that continually become larger and more powerful. Instead, governments and businesses can prioritize systems designed around practical local requirements, including agriculture, education, and access to public information—an approach the World Bank describes by the term “small AI.”
Technical advantages of embedded systems and agricultural applications in India
The term “small AI” in this context refers to systems trained or adapted to perform narrower functions, rather than the largest multi-purpose models available. A smaller system may require fewer computing resources to run, making it practical in certain circumstances for deployment on local devices or within environments where continuous access to distant cloud computing services cannot be trusted, particularly where continuous internet connections are prohibitively expensive, unstable, or where local language support is lacking, while acknowledging that small models do not automatically achieve the same performance as larger systems, with their usefulness depending on the task and data quality.
Donohoe identified agricultural development programs in India as a clear example of the type of opportunities available; farmers may need accurate information about soil conditions and farming practices in their own languages, while the number of available specialists is insufficient to provide individual advice everywhere. The Guardian reported that Donohoe mentioned the World Bank’s “AgriConnect” initiative while explaining this potential application, which should be understood as an illustration of the strategy rather than proof that a newly launched AI service has already achieved comprehensive nationwide deployment or independently measured agricultural productivity gains.
Bangkok meetings and infrastructure challenges in Africa and the labor market
This interview was conducted against the backdrop of the World Bank and International Monetary Fund annual meetings held in the Thai capital, Bangkok. At those meetings, Kristalina Georgieva, managing director of the IMF, emphasized the critical importance of reliable electricity, digital infrastructure, human skills, and appropriate regulatory safeguards for the successful adoption of AI. The Guardian also cited previous World Bank development research showing that large parts of rural education infrastructure in sub-Saharan Africa still lack reliable electricity and internet connectivity.
The article also discussed employment disruptions, particularly for young workers entering professions involving routine tasks. IMF estimates regarding the share of jobs exposed to AI should not be characterized as projections that the same share of workers will become unemployed, as exposure encompasses the potential for capability enhancement and shifting job responsibilities alongside the possibility of displacement.
Development financing and the implications of the strategy for Arab and African economies
Donohoe stated that the World Bank mobilized approximately $120 billion in private capital over the past year through financial instruments including guarantees; this amount points to the bank’s broader development financing activity rather than a new $120 billion investment program dedicated to AI.
What is new in the October 18 interview is the presentation of the World Bank’s current focus on locally adapted, less resource-intensive AI as a developmental strategy, despite no announcement of new funding allocations for specific countries, procurement contracts, or implementation timelines. For Arab and African economies, this topic raises fundamental questions regarding the availability of training data in Arabic and local languages, offline operation, and the availability of hardware and reliable sectoral information.
Frequently asked questions
Question: What is meant by the term small AI promoted by the World Bank?
Answer: It refers to artificial intelligence systems trained or adapted to perform specific, narrow-gauge tasks and requiring fewer computing resources, enabling them to run on phones and in low-internet areas.
Question: How can these small models be employed to support the agricultural sector in developing countries?
Answer: They can provide farmers with information and guidance on soil conditions and sound agricultural practices in their local languages to make up for the shortage of agricultural extension workers, as in the AgriConnect initiative.
Question: Does the $120 billion figure mentioned by Donohoe mean there is a new AI fund?
Answer: No, this amount refers to the total private capital that the World Bank succeeded in mobilizing over the past year for all development financing activities through instruments such as guarantees.
Question: What are the most prominent obstacles facing the deployment of AI in rural African regions?
Answer: The most prominent obstacles are the absence of reliable electricity supplies, weak internet connectivity, and a lack of digital infrastructure and technical skills in rural schools and communities.
