“This Is Africa” Cites MwanaSayansi Newspaper as an Example
Should we bridge local understanding of Artificial intelligence by creating technological terms that make current scientific developments accessible, or do we teach and help computers understand local languages? The answer is both.
African languages steadily evolve to accommodate current realities such as advances in technology, science, and the technical aspects of industry. For those that can remember
reading the book “Walenisi” by Prof. Katama Mkangi, you can recall the amazement of consuming complex sci-fi in Kiswahili. This sociology professor, who was detained for being a part of the underground socialist movement “Mwakenya” went about world-building an alien utopia with such expert storytelling and linguistic finesse that anyone who read the book talks about it with quiet awe.
He is but one exemplary example that Africans cannot only conceptualise other worlds or futuristic science and technology, space exploration, time travel, parallel universes, and extraterrestrial life in local languages but are also an integral part of creating a dynamic landscape where life imitates art. Which if we are honest is how AI started- by science fiction pushing scientists to make the ‘unimaginable’ happen.
Artificial intelligence (AI) through an African Lens
As with many developments of the 4th industrial revolution, the world deems Africa behind the curve. The one size fits all and Eurocentric metrics do not give room for heritage-centred design (HCD) or localised adaptation.
But we cannot disregard the digital divide that language perpetuates. Artificial intelligence (AI) makes it possible for machines to learn from experience, adjust to new inputs and perform human-like tasks. The majority of data that is coded, collected, selected, or used in machine learning is in English. Which of course creates a bias in favour of English speakers. Other aspects deepen this digital divide including training exclusion and geographical limitations such as access to quality networks and digital tools among others.
The bright side is linguistics may not be a barrier for much longer. People like Kenyan researcher and political analyst Nanjala Nyabola are making digital access and utilisation of the new terminologies easier for non-specialist or non-English speaking communities. Quartz reports, “With funding and support from the Stanford Digital Civil Society Fellowship, she (Nyabola) partnered with a team of Kiswahili scholars to translate key words in technology and digital rights into the most widely spoken African language.” The resource that is freely available online features words such as ‘Akili unde’ (Artificial Intelligence), ‘Mtambowavu’ (Bot), and ‘Ufichamishaji Kamili’ (End-to-End Encryption).
She explained in the interview, “The problem with Swanglish (Kiswahili mixed with English or English words adapted for local use) is that you get the word that fits in that moment but you’re not necessarily building a lexicon.”
Other Africans, Bonaventure Dossou (Benin) and Chris Emezue (Nigeria) have developed an Artificial Intelligence (AI) language translation model, named ‘FFR’ that is similar to Google Translate to bridge gaps in AI-first. The creation uses African languages in Natural Language Processing (NLP) which is a branch of AI that teaches and helps computers understand human languages. They are making some of the African languages that are collectively categorised as “low-resourced” easier to access, index, and use.
Now Tanzanian journalist and medical doctor, Syriacus Buguzi, has started ‘MwanaSayansi’, the first science and technology newspaper written in Kiswahili. Buguzi started the publication to make sure information in science is shared in a familiar and local language.
“We started it because we saw a gap in science communication in Tanzania. And previously when we tried to think of how to bridge the gap between researchers and the community, I tried to look at our audience in Tanzania and what they are lacking… If you have a class of scientists who are communicating their research findings, they present their stuff at conferences, in scientific journals they will always be in English… For Swahili speakers, it’s like denying them an opportunity if you are not communicating to them regularly about Science-you are denying them scientific information,” Buguzi told the BBC.
The momentum with which the linguistic gap in science is being bridged by innovative Africans paints a hopeful outlook for the future of AI on the continent.