Africa Demands AI Solutions That Speak Its Languages

Africa's AI needs are different from those in other parts of the world. The continent has unique challenges and opportunities that require tailored solutions. Dr Girmaw Abebe Tadesse, Principal Research Scientist and Manager at Microsoft's AI for Good Lab in Nairobi, emphasizes that language is a crucial aspect of Africa's AI story.

Currently, most AI models are trained on English and a handful of other global languages, leaving behind the diverse linguistic landscape of Africa. With over 1.5 billion people speaking more than 1,500 languages, the continent faces significant barriers to accessing AI-driven services.

To address this issue, initiatives like LINGUA Africa are making progress in developing open datasets, speech resources, and practical language tools for African languages. The program has received over 800 applications from 64 countries, with 85% of them being from the continent.

Microsoft Research Africa's Paza project is another example of efforts to improve speech recognition for low-resource languages. The project covers 39 African languages and has developed new models for Swahili and five Kenyan languages.