Despite being rich in linguistic diversity, Africa struggles to see its native languages reflected in artificial intelligence (AI) technologies. This oversight leaves millions of speakers without access to the benefits of modern technology as most AI systems are built primarily on English and a handful of other widely used languages.
In a bid to rectify this imbalance, researchers recently unveiled what is believed to be the largest dataset of African languages to date. This groundbreaking project, driven by the African Next Voices initiative, enlists linguists and computer scientists to augment the online presence of these languages.
Prof. Vukosi Marivate from the University of Pretoria articulated the urgency of integrating African languages into AI, pointing out that technologies reflecting local languages are essential for millions who think and dream in these languages. The project recorded 9,000 hours of native speech across 18 languages spoken in Kenya, Nigeria, and South Africa, focusing on real-life contexts such as healthcare, agriculture, and education.
Among the beneficiaries is Kelebogile Mosime, a farmer utilizing the AI-Farmer app to troubleshoot agricultural challenges in her mother tongue, Setswana. Such initiatives demonstrate the transformative potential of language-accessible technologies, bridging gaps in vital services.
While this project marks significant progress, there is a broader implication that language inclusivity in AI can preserve cultural identities and perspectives, fostering a more comprehensive understanding of the world. The African Next Voices initiative serves as a vital foundation for future innovations, promoting the preservation and celebration of Africa's linguistic and cultural heritage.
