Not lost in translation: training AI to speak African languages

As artificial intelligence transforms communication, education and business worldwide, many African researchers have become increasingly concerned that hundreds of African languages remain largely invisible to the world’s leading AI platforms. In response, a team at the University of Cape Town has developed new language tools aimed at ensuring African languages are not left behind in the global AI revolution.

A digital divide in language technology

Modern AI systems rely on enormous amounts of text and speech data to learn how languages are spoken and written. While English, Chinese, Spanish and other major global languages benefit from vast digital resources, many African languages have comparatively little online content available for training AI models.

As a result, popular AI assistants, translation tools and voice-recognition systems often struggle to understand or accurately generate African languages. Researchers warn that this technological gap risks excluding millions of people from the benefits of rapidly advancing digital services.

The challenge is particularly significant in Africa, where more than 2,000 languages are spoken across the continent, representing one of the world’s richest linguistic landscapes.

Building African solutions

Frustrated by the lack of progress from major technology companies, researchers at the University of Cape Town decided to take matters into their own hands. Their project focuses on collecting language data, developing specialised models and creating AI systems capable of understanding and communicating in languages that have historically been overlooked by global technology firms.

The initiative involves collaboration with local communities, linguists and technology specialists to ensure that language resources are both accurate and culturally relevant. By building datasets directly from native speakers, researchers hope to improve the quality and reliability of AI-generated content.

More than translation

The benefits extend far beyond simple translation. AI systems capable of understanding African languages could improve access to education, healthcare, government services and financial products. Voice-enabled tools may be particularly valuable in regions where literacy rates vary or where people prefer to communicate in their native language rather than a colonial language such as English, French or Portuguese.

Researchers also believe that language-focused AI can help preserve cultural heritage by digitally documenting languages that may otherwise face decline as younger generations increasingly adopt dominant global languages.

Africa’s growing AI ecosystem

The project reflects a broader trend across the continent as African universities, start-ups and research institutions invest more heavily in artificial intelligence. Rather than relying entirely on imported technologies, many organisations are seeking to develop AI systems designed specifically for local needs and realities.

This approach has attracted growing attention from international investors and development agencies, which increasingly view AI as a tool that can accelerate economic growth and improve public services throughout Africa.

Challenges remain

Despite promising progress, significant obstacles remain. Developing high-quality language models requires substantial computing power, funding and long-term commitment. Many African research institutions operate with limited resources compared with global technology giants.

There are also concerns about ensuring that AI systems accurately reflect the diversity of dialects, cultures and linguistic variations found across the continent.

Giving every language a digital future

For the researchers involved, the goal is about more than technology. It is about ensuring that speakers of African languages can participate fully in the digital economy without abandoning their linguistic identity.

As artificial intelligence becomes increasingly integrated into everyday life, projects such as the one at the University of Cape Town may play a critical role in ensuring that Africa’s languages are not lost in translation, but instead become part of the next generation of global technological innovation.

Newshub Editorial in Africa – 4 June 2026