Time flies with great content! Renew in to keep enjoying all our premium content.
On the cusp of a new dawn? Microsoft’s Kiswahili AI model to bridge tech inclusivity gap
Kiswahili scored a first in 2023 after it was picked as the only sub-Saharan language among 40 other global dialects to be taken on board at the launch of Google’s launch of its generative AI dubbed Bard. PHOTO | SHUTTERSTOCK
Last week’s announcement by tech giant Microsoft that it will support the development of an artificial intelligence (AI) model in Kiswahili as part of its grand plan to invest up to $1 billion (Sh132.5 billion) in Kenya’s digital ecosystem has elicited excitement around local tech circles, with experts expressing optimism that this could be the magic wand to drive uptake among native communities.
Microsoft said the initiative will be geared toward supporting Kenya’s unique cultural and linguistic needs, noting that its implementing partner, Emirati AI firm, G42, has already begun training an open-source large language AI model in Kiswahili and English through its data infrastructure in the US.
“To build on this and help accelerate advanced research in Kenya, Microsoft and G42 will increase their combined collaboration and support for local universities, through the Microsoft Africa Research Institute, the Microsoft AI for Good Lab, the Mohammed Bin Zayed University of Artificial Intelligence in Abu Dhabi,” said Microsoft.
If effectively executed, pundits argue it will translate to another major win for the Kiswahili language as well as bridge a larger part of the tech inclusivity gap by powering its roll-out to untapped markets.
“I think it’s a positive thing for the country and for a continent at large as it will enhance better inclusion within the tech space. It points to a new dawn of easier interactions between advanced technologies and local communities,” says Snehar Shah, chief executive of Edutech institution Moringa School.
“In the coming short years, tech will be for everyone. Picture a farmer in a Kenyan rural farm, for example, making customised queries in his preferred language on a generative AI tool. It goes without saying that the answer generated will be easier to comprehend which will in turn give rise to better quality decisions,” he adds.
Kathleen Siminyu, who works as an AI researcher focusing on Natural Language Processing for African languages at Mozilla Foundation notes that the government should complement the private sector’s efforts in the initiative by enacting supportive policies that could see Kiswahili usage taken a little more seriously in official engagements.
“The initiative in itself is thoughtful, but I foresee a slight challenge in the usage of the new models in the Kenyan market. Unlike countries like Tanzania where there’s a uniform Kiswahili dialect, speakers in Kenya tend to employ a mixture of many versions together and that then becomes a challenge for someone seeking to develop a standard model for that kind of population,” states Ms Siminyu.
“There’s thus room to think critically through the model that will be tailor-made for the Kenyan context as we don’t have a pure standard that we can purport to align with. However, if that hurdle is overcome, the product would drive mass uptake of AI technology among the bulk of Kenyans whose primary language of preference is not English.”
For organisational usage, the analysts point to a minimal impact on internal workings, but note that it could possibly open doors to business opportunities in new markets.
“The increased availability of Kiswahili as a language to interact with LLMs (large language models) and chatbots will not be enough to cause drastic change in organisational cultures. However, it would present some opportunities for business to address new markets and increase accessibility with a local language,” says Michael Michie, co-founder at software firm Everse Tech.
Mr Michie draws parallels with happenings in other countries like Japan where he says learners taught math and sciences in native languages outperform their counterparts who learn in English, a clear indication of the opportunities that lie within utilisation of local parlances.
“We could leverage AI to help build terminologies and explain complex math and science in local dialects which would not only make these fields of study more accessible, but also create smarter students within these fields,” he avers.
Last year, Kiswahili scored a first after it was picked as the only sub-Saharan language among 40 other global dialects to be taken on board at the launch of Google’s launch of its generative AI dubbed Bard.
Other Google services and products available in Kiswahili include Gmail, Docs and the popular mobile operating system Android.
Other internet-based platforms such as Airbnb, Uber, WhatsApp, Spotify, among others, are major global brands that support Kiswahili.
