Africell has entered into a strategic research collaboration with Oxford University to examine the economic and social impact of artificial intelligence tools in West Africa, with a particular focus on AfriGPT, Africell’s innovative SMS-based AI service. The partnership represents a significant step in understanding how emerging technologies can be adapted to low-income and low-connectivity environments across the African continent.
The research project will be led by scholars from Oxford University’s Internet Institute and Department of Economics and is partially funded by Schmidt Sciences through its $3 million “AI at Work Program.” The study will explore how AfriGPT is used by subscribers in Sierra Leone and The Gambia, assessing its influence on livelihoods, access to information, education, entrepreneurship, and broader social behaviour.
AfriGPT is a low-cost, subscription-based service that allows users to interact with ChatGPT via simple text messages. Unlike most AI chat platforms, which require smartphones and stable internet connections, AfriGPT operates entirely through SMS and functions on basic 2G mobile networks. This design eliminates the need for mobile data and makes the service accessible to users who rely on feature phones, particularly in rural and underserved communities.
By removing the internet requirement, AfriGPT addresses one of the most significant barriers to AI adoption in many parts of Africa. Internet penetration in countries such as Sierra Leone remains relatively low, especially outside urban centres. As a result, many people who could benefit from AI-powered tools are excluded by cost, infrastructure limitations, or lack of access to smart devices. AfriGPT offers a practical workaround, opening new possibilities for how generative AI can be deployed in frontier markets.
Researchers involved in the project describe AfriGPT as a unique case study for examining how AI tools are adopted by unconventional or previously excluded user groups. Early evidence suggests that users employ the service for a wide range of practical purposes, including job searches, small business planning, educational support, and general problem-solving. These applications highlight the potential of AI to support economic participation and human development when designed with local realities in mind.
Sam Williams, Africell’s Group Communications Director, noted that while AI tools such as ChatGPT are increasingly commonplace in Europe and North America, their use in Africa remains limited due to economic and technological constraints. He explained that AfriGPT was developed specifically to bridge this gap.
“AfriGPT is promising because it enables mobile users in countries like Sierra Leone to access AI chatbots without internet connectivity,” Williams said. “Africell is pleased to support this rigorous research by world-class scholars, as we want to see whether AfriGPT’s perceived value and utility are supported by empirical data.”
Subscribers to AfriGPT pay a modest fee to receive AI-generated responses to questions submitted via SMS. The system is linked directly to ChatGPT, allowing users with basic phones to benefit from the same underlying technology available to global users with high-speed internet and advanced devices. This model significantly broadens access, particularly for poorer, younger, and more rural populations.
A preliminary survey conducted by Oxford researchers in April 2025 revealed notable differences between how AfriGPT is used in Sierra Leone and how ChatGPT is typically used in higher-income countries. The findings suggested distinct patterns related to timing, purpose, and user demographics. Building on this initial work, the expanded research programme funded by Schmidt Sciences will gather more detailed data across regions, seasons, and device types.
The study is being led by Johanna Barop, a DPhil researcher at Oxford’s Internet Institute, and Joseph Levine, a DPhil researcher in the Department of Economics. They explained that the project aims to understand the social, economic, and behavioural factors that shape AI use in Sub-Saharan Africa.
“There is very little existing research on how AI is used in low-income and rural African contexts,” the researchers said. “We want to understand how usage differs from global patterns, as well as the benefits and risks AI may pose in these environments.”
They added that generating evidence-based insights is essential to ensuring that AI tools are developed and distributed in ways that are inclusive, responsible, and responsive to the needs of African communities.
Through this collaboration, Africell and Oxford University hope to contribute valuable knowledge to global debates on digital inclusion, responsible AI development, and the future of work in emerging economies, positioning West Africa at the centre of conversations about how technology can drive equitable development.
