Grace Ashiru
Machine learning and AI are ushering in novel opportunities. However, as a recent study suggests, these benefits mainly cater to speakers of predominant languages like English and French, such as in products like ChatGPT.
There’s a push to guarantee that those who speak Shona, Hausa, Xhosa, Kiswahili, and other African languages can equally tap into these technological advancements.
“It’s perplexing that AI tools for African languages are so scarce,” remarks Kathleen Siminyu, a tech leader based in Kenya. Through her work at the Masakhane Research Foundation, she’s dedicated to providing accessible AI tools for speakers of African languages.
“Inclusion and participation in the development of language technology shouldn’t be an afterthought; it should be a top priority from the start..”
Siminyu leads a recent study, released Friday in the Cell Press journal Patterns, that presents a strategy for enhancing AI tools dedicated to African languages. Central to this initiative is the availability of specific language datasets for training computers to handle African languages.
The study’s authors highlight the importance of championing African content. They emphasize the necessity of crafting essential resources like dictionaries, spell checkers, and keyboards for these languages and easing the financial and administrative hurdles to translate government messages into various national tongues, encompassing African languages.
The group intends to broaden its research and tackle obstacles that could impede access to the technology. Their discoveries could also contribute to the preservation of native African languages.
They acknowledged shortcomings within their team, noting, “All the researchers involved in this study are Anglophone. Therefore, the inclusion of Francophone and Lusophone African stakeholders from pertinent groups is essential,” they stated.
