So what else can I tell you about AI? Listen to the small builders and shapers from unlikely parts of society!

Prof Vukosi Marivate, Director African Institute of Data Science and AI (AfriDSAI), University of Pretoria, South Africa; Co-Founder Lelapa AI; Co-Founder Deep Learning Indaba; Member of the Independent Scientific Panel for AI for the UN; Member of the African AI Council at Smart Africa.

AI presents a transformative socioeconomic opportunity for South Africa, Africa and beyond, provided it is developed as an instrument of local sovereignty, grounded in indigenous languages, local data, and national institutions.

A picture I took while visiting Nyeri , Kenya earlier this year. Copyright Vukosi Marivate

You have likely heard a great deal of debate and discussion about AI over the last few years;

• AI is an amazing technology;
• AI is disrupting much of our society, directly or indirectly;
• AI has been sucking up not just attention from the world, but also capital, national policy, and power concentration. 
• There is a tussle between regulators, civil society, private companies and governments. Who gets to shape the future and not feel like they are left behind? How do we reduce the downsides?

We know all of this, one way or another. Let’s take a step back and consider what these debates and discussions look like from the perspective of the majority of the world, and what their consequences might mean for those who are too often absent from the centre of the conversation.

As Kentaro Toyama notes in his book Geek Heresy, technology is an amplifier. It can amplify both the good and the bad. AI and its related fields are not new; they are the result of decades of research and development. I say this as an AI researcher based in South Africa, on a continent of 55 countries that has spent decades grappling with the opportunities and consequences of technological change. The question, then, is not simply what AI can do, but whose priorities it amplifies, whose interests it serves, and who gets to shape the systems being built.

When the early waves of the current AI revolution were starting to hit the shore, going back to around 2010–2015 with the scaling of deep neural networks using specialised computing systems running on Graphics Processing Units (GPUs), many across the continent started preparing. We were preparing for a future in which we wanted to be active shapers of this technology: to harness its benefits while reducing its harms and downsides, rather than increasingly being relegated to simply being users of technologies developed elsewhere. I was fortunate to be one of those preparing. In late 2016, a group of AI researchers and developers came together to start the Deep Learning Indaba, to strengthen machine learning on the continent and enable Africans to become active shapers and contributors to the coming AI revolution, rather than simply adopters. Since then, the Indaba has grown into one of the continent’s most consequential capacity- and capability-building programmes for AI and data science, alongside other movements such as Data Science Africa and Data Science Nigeria, which started in 2015, and many others. The Deep Learning Indaba, with “Indaba” meaning an important meeting in isiZulu, has grown into much more than an annual gathering of the African AI community. It brings together an academic and research conference, tutorials, workshops, community spotlights that connect people across countries and sectors, including climate, local languages and governance, grassroots movements, startup pitching and much more. 

There is already a growing ecosystem of people across the continent doing this work precisely. As you read this in mid-August 2026, the eighth Deep Learning Indaba has just taken place in Lagos, Nigeria, bringing together more than 1,000 researchers, developers, policy specialists and others. Its smaller, country-level IndabaX meetings have now reached 47 of 55 African countries, while Data Science Africa 2026 was held just weeks earlier in Kampala, Uganda. These communities have been working alongside one another for close to a decade, creating spaces for serious discussion and exchange on issues ranging from climate and local languages to AI policy, law, startups, trust and emerging technologies. I am still struck each year by the depth of these conversations, often in spaces far removed from the places we are accustomed to being told are where the “real” discussions about AI take place. Africans are not waiting for others to decide what the future of AI should look like. They are building, debating, refining and making choices about where these technologies can serve their societies, from creating local economic opportunities and improving public services to developing systems that can recognise and work with the languages people actually speak. 

Why do I highlight the above? When it comes to policy, technical ability, capacity and capability to build such technology locally, we tend to be sold the idea that it is too hard and it should be left to just a few powerful countries or companies. The word “sovereignty” is dealt with as a swear word. This should not distract us. For me, sovereignty means choice: the ability to understand our own needs, take in information, assess our options, and decide what is right for our people and our societies. It means having the choice to decide where we should build local capacity and capability, and where we may strategically choose not to. It means not thinking only about short-term gains or reaching for FOMO (the Fear Of Missing Out) but making deliberate investments in our own long-term development. How do we make sure that someone who speaks isiNdebele in Southern Africa can be properly served by these LLMs? The same question applies to those who communicate in Swiss German. These are the kinds of choices that can ultimately amplify the voices of the majority of the world, rather than allowing the future to be shaped only by a powerful, and often loud, minority.

I put my hand up to join the UN Independent International Scientific Panel because I wanted to share my experiences, learn from others, and contribute to shaping how we think about AI globally. I am a Natural Language Processing researcher at heart, with a focus on African languages, and it has been meaningful to be part of this community nationally, across the continent and globally, helping shape its development over the past 10 years. The journey has had its ups and downs, and those experiences are ones I can now bring to the table as we think about developments in South America, Asia, Europe and beyond. I have also had the privilege of working with people across fields such as agriculture, health and education who have spent decades trying to ensure that technological progress serves everyone. Their work may not always be flashy, and much of it is done by people who simply believe in something bigger than themselves and in the idea that progress should benefit all of humanity. They believe that knowledge is meant to be shared, that iron sharpens iron. They may not be part of trillion-dollar deal announcements, but their communities feel the impact of their work in their everyday lives.

I urge you to look for these people, the “Probonarii”, a term used by  Bright Simmons, and give them the space to share their knowledge and experience. We will not think-tank or “hire consultants” our way into the future. We will do so by recognising, supporting and investing in local ecosystems, and by giving the people within them the resources and space to build responsibly and sustainably for the benefit of all. 

Note: This post represents my original, unedited English draft. A translated and edited version of this piece was recently published in the Süddeutsche Zeitung Dossier. If you read the German publication, you will notice a few editorial changes made to fit the spatial constraints of a standard op-ed. The editors condensed the opening discussion, reordered some of the arguments regarding local capacity building, and omitted a few of my more personal reflections, such as my work with the UN Independent International Scientific Panel and the references to Kentaro Toyama and Bright Simmons. Despite these cuts, the core message remains exactly the same: AI must be developed as an instrument of local sovereignty. You can read the published German version here: Künstliche Intelligenz muss zum Wegbereiter lokaler Souveränität werden.