Alifosina Mtseteka, a sugarcane farmer in Chisemphere village, Malawi, used a mobile AI chatbot named Ulangizi to describe a pest destroying her okra crop and received treatment advice in Chichewa, her native language. The guidance led to successful pest control, marking a practical win for locally relevant AI. Malawi has now launched the Malawi Low-Resource Language Data Trust Initiative, a government-backed effort to develop AI systems fluent in Chichewa, spoken by about 70% of Malawians and also used in Zambia, Mozambique and Zimbabwe. The project aims to close the AI language gap by building licensed datasets from government records and media content, with Nation Publications Limited contributing Chichewa-language material. Support comes from the World Bank and the Gates Foundation. Most large language models today are trained on over 90% English text, leaving African languages like Chichewa underrepresented due to limited digitised data. This shortage has real consequences: farmers, healthcare workers and citizens often cannot use AI tools because they do not understand local languages. Early attempts to train AI on Chichewa produced unnatural outputs, including responses with an Indian accent, but developers of Ulangizi improved the system through persistent refinement. The initiative prioritises voice and text interfaces to serve populations with varying literacy and smartphone access, relying on near-universal voice communication. Feston Kaupa, Malawi's minister of defence, spoke at the launch in Lilongwe on behalf of the ICT minister, stating that countries adopting AI strategically will boost innovation, productivity and global competitiveness. The broader goal is to ensure Malawians do not have to rely on foreign-language systems to access modern technology. Pelonomi Moiloa, CEO of the company behind Ulangizi, emphasized that people should not need to adopt foreign cultures to use advanced tools. With Africa's AI market projected to grow from $4.51 billion in 2025 to $16.5 billion by 2030, the push for local language AI is also an economic play, aiming to secure future value within the continent.
The fact that a pest advice chatbot required a breakthrough to sound natural in Chichewa exposes how deeply foreign the foundations of AI remain, even as it claims to serve local needs. Malawi's initiative depends on external funders and unproven scalability, raising doubts about whether homegrown AI can truly emerge under such conditions. If the system still struggles with accent accuracy, its real-world usability for rural users remains limited despite the success story. A technology that needed to be corrected to stop sounding Indian suggests the data pipeline may be importing biases as much as solutions.
💡 NaijaBuzz Take is AI-assisted editorial opinion, not established fact. Full disclaimer →
