It seems AI flunked geography class. In this DALL-E-created version of Africa, the geniuses behind AI morphed the postage stamp-sized central African country of Rwanda into a North African powerhouse once known as Algeria and Libya, making Rwandan territory Africa’s largest. The Atlantic Ocean mysteriously has transformed into Lake Tanganyika. Ethiopia is suddenly “Burandi” (must be Burundi’s lost twin), while Uganda/Ugandi, once the Pearl of Africa, has vanished into the Indian Ocean. Unnamed Madagascar has a cluster of mini-islands to the north. Then there’s DR Congo, or now “Angavika”, which, Google informs me, is an Indian term related to a deficiency or disability in Hindi.
A better view of the map
And that’s when it hit me. Is this absurd mess of disinformation really that different from the infamous 1884-85 Berlin Conference? It’s often called the ‘Scramble for Africa’ because it represents the ruthless and exploitative way European powers determined the fate of the African continent, driven solely by their own interests.
Back then, a cabal of privileged, wealthy white men, including royalty like King Leopold of Belgium, wore medal-heavy military uniforms. The others, mainly diplomats, were decked out in tailcoats, top hats and cravats. Sporting bushy sideburns, clutching pocket watches and brandishing imperial arrogance, they sat around a large table for months, carving up a continent most had never set foot on.
Image source: Anton von Werner, Public domain, via Wikimedia Commons
These colonial European powers horse-traded amongst themselves with the casual ease of someone slicing a pie; they claimed territories, renamed cities and laid down borders with no regard for the people living there. This arbitrary external imposition of borders (many with straight lines) and governance structures in the colonial language have had long-lasting negative impacts on its people.
Today, hoodies, jeans and smartwatches have replaced topcoats, cravats and pocket watches. Instead of maps, we have algorithms developed and trained by companies and researchers outside Africa. These algorithms shape narratives and perceptions about the continent, reinforcing stereotypical or Western-centric views rather than authentic, diverse African voices. In place of colonial governors, we have tech developers. But the outcome is disturbingly similar. Outsiders, most of whom have limited understanding of or appreciation for Africa and its nuances, are shaping how the world sees it, often inaccurately and sometimes harmfully, creating 21st century “digital colonialisation”.
AI content, developed outside Africa, continues to reflect the continent like a monolith despite 54 countries, an estimated 3,000 languages and incredibly diverse beauty and cultures. In an article published by the International Trade Centre, Africa is a hotbed of innovation with a booming tech start-up scene with more than 1,000 tech hubs. Further, according to the Global Entrepreneurship Monitor (GEM), African women have the highest rate of entrepreneurship in the world.
Most AI technologies and datasets are developed without direct input from Africans, resulting in systems that lack African perspectives and often miss the nuances specific to local African contexts. This issue was highlighted at the African Philanthropy Conference in Victoria Falls, Zimbabwe, in July, where Ethiopian scientist and AI expert Dr Abebe Birhane, a fellow at the Mozilla Foundation, underscored this issue with a striking comment: only 3% of all AI content is generated in Africa.
What needs to change?
For AI to more accurately represent Africa, Dr Birhane suggested that AI developers should have incentive structures that reward equity and accountability for cultivating systems and solutions that shift power. She noted that these problems are not annoying obstacles but fundamental flaws that must be addressed and African communities must provide input and active participation.
In an interview, UNESCO Co-chair for AI and Human Rights, Dr Susan Perry said, “Dr Abebe Birhane is absolutely right. I would go one step further than Dr Birhane to identify missing data sets (97% of data on Africa produced outside of Africa) and provide tax incentives to AI developers to produce those data sets with Africans in Africa. This can be done through cooperation with the 1000 estimated data hubs on the continent and would bring in much-needed cash to digitise existing (paper) data on Africa produced by African governments, businesses and civil society actors. AI can only be as good as the data; digital invisibility impacts all Africans. I agree with you; we are witnessing the digital colonisation of Africa.”
Serial tech-for-good social entrepreneur and founder of Tech Matters, Jim Fruchterman, noted, “It is essential that Africans retain control over their data and how it is used, rather than depending on AI products trained on data from other parts of the world with only the tiniest bit of data from Africa. Only then will we have AI tools which useful for Africans, and actually reflect the cultures, languages, and even geography of the continent.”
Like the decisions in the 1884-85 Scramble for Africa, the development of AI that impacts Africa without African involvement perpetuates and reinforces a troubling and ongoing power imbalance. This exclusion strips Africa of its agency and control over its digital and technological representation. The narratives created by AI can shape global understanding of Africa for years to come, like how entrenched colonial legacies have influenced Africa’s post-colonial reality.
At the very least, AI developers should prioritise getting the fundamentals right, such as ensuring an accurate map of Africa. It is, in my opinion, profoundly irresponsible for companies to propose creating tools or resources based on flawed and incorrect data from the outset. Surely, they must be held to a higher standard, particularly in an era so heavily influenced by #fakenews and #disinformation?
The Berlin Conference
If you’re interested in learning more about the Berlin Conference, you might be surprised at who had a seat at the table and the countries that signed the Berlin Conference Act –
