Lost in translation - How Africa is trying to close the AI language gap

Africa is a continent with a huge variety of languages, with more than two thousand spoken across its countries. However, when it comes to artificial intelligence, most African languages are not included. This is mainly because AI tools like ChatGPT are trained using English and other major languages from Europe and China. These languages have a lot of written content online, which makes it easier for computers to learn from them. In contrast, many African languages are mostly spoken and do not have enough written material for AI to use. As a result, millions of Africans are left out when it comes to using AI in their own languages.

Professor Vukosi Marivate from the University of Pretoria explains that people think, dream, and understand the world through their own languages. If technology does not support these languages, entire groups of people risk being left behind. To address this problem, researchers have created what is believed to be the largest dataset of African languages. The African Next Voices project brought together linguists and computer scientists to record 9,000 hours of speech in 18 African languages across Kenya, Nigeria, and South Africa. These recordings include everyday situations in farming, health, and education. Languages like Kikuyu, Dholuo, Hausa, Yoruba, isiZulu, and Tshivenda were included, and the team hopes to add more languages in the future.

The project was funded by a $2. 2 million grant from the Gates Foundation, and the data will be open for anyone to use. This means developers can create tools that translate, transcribe, and respond in African languages. There are already examples of how AI using local languages can help solve real problems. For instance, Kelebogile Mosime is a farmer in South Africa who uses an app called AI-Farmer. The app recognizes several South African languages, including Setswana, and helps her with farming challenges like pest control and diagnosing sick plants. She finds it very useful because she can ask questions in her home language and get helpful answers.

Pelonomi Moiloa, CEO of Lelapa AI, is working to build AI tools in African languages for banks and telecom companies. She points out that English is often seen as the language of opportunity, but for many South Africans who do not speak it, this can mean missing out on essential services like healthcare and banking. She believes language should not be a barrier to access. Professor Marivate adds that language is not just about words; it is about history, culture, and knowledge. If African languages are not included in technology, people lose important ways of seeing and understanding the world. This is why it is so important to make sure AI can work with African languages.