Africa and AI: Decolonising Datasets
Ask anyone about AI and they will immediately rattle off the usual suspects: ChatGPT, Gemini, Copilot, Claude, Perplexity.
Now ask them to name an African AI platform… and watch the silence stretch longer than a buffering on Smart4U data bundle that’s reached the Fair Usage Policy limit. It’s not that they are not there, but they are mostly known in niche markets in dev communities and test groups with the tech broskis..
I’ve personally come across a few, like AfricaAI (which currently seems to be in powersaving mode) and Vambo AI, on a mission to make AI accessible in every African language, has earned a spot in the Google for Startups cohort.
As a digital storyteller, I often think of myself as a digital archivist. And we’ve said this countless times: if you don’t tell your own story, someone else will tell it for you… badly. The same logic applies to AI.
Right now, by default, we are feeding global AI systems with our data while remaining largely invisible in how that data is interpreted. These large language models are learning from us, yes but as outsiders looking in. And like any outsider, they inherit bias, assumptions, and sometimes outright stereotypes.
I notice this particularly in image generation. Ask for a generic scene and you’ll likely get something Eurocentric. Specify “African city,” and suddenly you’re toggling between safari aesthetics and adobe clichés.
The question is no longer whether we should use AI. That ship has sailed. The real question is: what frameworks, policies, and ownership structures are we putting in place as Africa?
For the most part, African governments are treating AI as a buzzword, something to sprinkle into a speech, a reason to convene government officials at a conference resort somewhere to talk about AI systems, where half the room’s entire AI experience is a ChatGPT prompt. Meanwhile, the young generation is already building, automating systems aunching startups, and scaling solutions with AI at the centre while the government policy is yet to finish tying its laces.
On the 13th of March, Zimbabwe’s president officially launched the Zimbabwe National Artificial Intelligence Strategy, themed Harnessing AI for Inclusive National Development. The country has set its sights on becoming the hub for AI-driven development in Southern Africa.
The strategy is organised around five flagship pillars and includes introducing AI in schools from primary level through to university, alongside workforce upskilling initiatives.
On paper, it reads well. In practice, it reads more like a wishlist than a roadmap. The document says all the right things but sidesteps the hard questions:
How much will this cost and Where exactly will the funding come from? Will a new tax be introduced to finance the strategy? What happens when the budget does not match the ambition? Who owns the datasets? And most importantly what about the electricity challenges?
Still, it’s a start. Having a strategy is better than none
What’s your go to AI?
WinterABC26 Africa and AI
Leave a Reply