Bridging the AI Divide

Abstract While the use of artificial intelligence (AI) technologies in Africa is growing, the disparities in use persist along various dimensions. This article aims to address how enabling governance frameworks can make large language models (LLMs) more inclusive through representation of low-resource languages in training data sets to enable equitable access to information and services. This article assesses emerging governance ecosystems in Africa from the perspective of coloniality and representation in generative AI and the extent to which LLMs can be used to tackle power asymmetries between African data subjects and AI developers to reduce inequalities that the adoption of generative AI may induce in Africa. By assessing the emerging national AI strategies in Africa, this article identifies a gap in AI governance frameworks across Africa specifically in relation to inclusivity in AI development. The countries briefly reviewed are Kenya, South Africa, and Nigeria due to the size and importance of their economies in sub-Saharan Africa and the recent efforts by the governments in these countries to adopt a governance framework to maximize AI technologies in their economies. With the use of case studies and developments across Africa, this article identifies three main data governance areas that will enable equitable generative AI in Africa. These are data generation and collection, regulatory sandboxes, and policy prototypes as well as data sharing. The issues addressed in this article center on data justice and the necessity for visibility, fairness, and representation in the adoption of generative AI across Africa.