Africa's Struggle to Bridge the AI Language Gap
Africa, home to more than a quarter of the world’s languages, faces a significant challenge in the realm of artificial intelligence (AI). Despite this linguistic diversity, many African languages are overlooked in AI development, primarily due to a lack of investment and accessible data.
Currently, most AI tools, including popular ones like ChatGPT, are trained predominantly on English, along with other European and Chinese languages. These languages have vast amounts of written data available online, which AI systems can easily use for training. However, many African languages are predominantly spoken and lack sufficient written records, making it difficult to train AI models that can serve these populations effectively.
For millions of people across the continent, this means exclusion from the technological benefits that others enjoy. However, researchers have recently made strides to address this issue, unveiling what is considered the largest known dataset of African languages.
Prof. Vukosi Marivate from the University of Pretoria, a key figure in this initiative, highlights the importance of linguistic inclusion. "We think in our own languages, dream in them, and interpret the world through them," he explains. "If technology doesn’t reflect that, a whole group risks being left behind."
The African Next Voices project, a collaboration of linguists and computer scientists, aims to close this gap by creating AI-ready datasets in 18 African languages. While this represents just a fraction of the over 2,000 languages spoken across the continent, it is a critical starting point, and the team hopes to expand further.
Over the past two years, the project has recorded 9,000 hours of speech from Kenya, Nigeria, and South Africa, covering everyday contexts like farming, healthcare, and education. The languages included in the dataset range from Kikuyu and Dholuo in Kenya to Hausa and Yoruba in Nigeria, and isiZulu and Tshivenda in South Africa.
Prof. Marivate explains that the project provides a foundation on which future innovations can build. "You need some basis to start off with, and that's what AfricanNext Voices is," he says. "Then people will build on top of that."
The team worked to ensure that the dataset was as inclusive as possible, recording voices from various regions, ages, and backgrounds to reflect the true diversity of African communities. Lilian Wanzare, a computational linguist from Kenya, emphasizes the importance of capturing the nuances of how people speak and live on the continent, noting that "Big tech can't always see those nuances."
Funded by a $2.2 million grant from the Gates Foundation, the dataset will be made publicly available, allowing developers to create tools that can translate, transcribe, and interact in African languages.
The impact of AI tools using indigenous languages is already being felt in practical applications. One example is Kelebogile Mosime, a farmer in South Africa's platinum belt, who uses an AI-powered app called AI-Farmer. The app recognizes several South African languages, including Sesotho, isiZulu, and Afrikaans, to help farmers solve problems related to crop management.
"I see the benefits of being able to use my home language, Setswana, on the app when I run into problems on the farm," says Mosime. "For someone in rural areas like me, who isn't exposed to technology, it's incredibly useful."
Another company making strides in this space is Lelapa AI, a South African startup building AI tools in African languages for industries such as banking and telecommunications. CEO Pelonomi Moiloa stresses that the current language barriers in these sectors can significantly limit access to essential services for many South Africans who don’t speak English.
"Language is a huge barrier," she says. "We're saying it shouldn't be."
For Prof. Marivate, the issue goes beyond mere convenience. "Language is access to imagination," he asserts. "It’s not just words – it’s history, culture, knowledge. If indigenous languages aren't included, we lose more than data; we lose ways of seeing and understanding the world."
The African language AI revolution is about ensuring that technology reflects the continent's diverse cultures and histories, making it accessible and beneficial to all.