#2 Linguistic Bias is A Design Choice
Africa’s linguistic wealth is largely invisible.
Generative image inspired by J. D. ’Okhai Ojeikere’s iconic ‘Hairstyles’ series. This image reimagines African hair as both cultural archive and living architecture, each strand a story, each style a blueprint for identity in the age of AI.
At the heart of AI is data - and data is not neutral.
Large language models (LLMs) are trained on what’s most digitally available: books, websites, subtitles, social media, mostly in English, French, Spanish, German, and Mandarin.
The languages with colonial legacies and global economic clout are those most seen by machines.
View the gallery - Do Big AI Bots Speak African Languages? Nah.
Africa’s linguistic wealth? Largely invisible.
Why? Because most of its languages don’t exist in structured digital form at scale. They lack the extensive online presence and standardised datasets that AI systems rely on to “learn.”
This is the linguistic fault line in modern AI: a gap where data scarcity translates into digital silence.
