World Bank jobs council urges 'crisis-era' support measures as IMF warns of AI ‘tsunami’ hit

Global tech giant and financial institution the World Bank has warned that African languages are inadequately represented in artificial intelligence (AI) systems, which could leave millions of Africans unable to fully benefit from the technology. In its Africa Economic Update: Building AI Readiness report published in October 2026, the bank highlighted that the underrepresentation of African languages in AI training data could limit the effectiveness of AI models in understanding diverse African languages and cultural contexts.

The report states that 56% of open-source training data is exclusively in English, leaving a significant portion of Africa's 2,000 languages underrepresented in AI development. This imbalance poses a challenge not only for African access to AI but also for the technology's ability to serve African communities reliably in their native languages.

The bank argues that adapting AI to local languages and conditions can provide significant opportunities for Sub-Saharan African countries. It cites the example of InkubaLM, a language model developed for African languages by African AI company Lelapa AI. This model reportedly outperforms larger models on African languages like isiZulu, Yoruba, Hausa, Swahili, and isiXhosa.

However, the report emphasizes that performance claims should be evaluated carefully, as success in selected tests does not guarantee superior performance in every real-world application. Moreover, the bank raises concerns about the ethical implications of language data collection, control, and use in AI development. It calls for clear rules to ensure permission for data use, protect privacy, and establish fair compensation for local researchers and communities whose language resources contribute to AI development.

The report also stresses the importance of addressing infrastructure, skills, and computing access barriers in Sub-Saharan Africa for local researchers and businesses to develop, test, and maintain AI systems. The World Bank's findings suggest that while African-language AI holds promise for improving education, information, and digital services, it must be accompanied by investments in language resources, technical skills, infrastructure, and meaningful participation from the communities who will use the technology.

The ultimate measure of AI progress in Africa should be the reliable performance of AI tools in enabling people to learn, work, and participate in the digital economy, rather than merely the number of languages a system supports.