Closer to Home: Radical Theories for Africa’s Digital Realities- Data Tamasha Africa Event Recap

“If, in your lived reality as a technology and data worker, you had to walk 35 km to fetch water for every email generation prompt, for every AI poster, would the water consumption of Generative AI be an afterthought in the infrastructure that you are building?“

Nanjala Nyabola, Data Tamasha Africa 2026

Data Tamasha Africa is a continental data festival that brings together Africa’s data practitioners and enthusiasts to strengthen collaboration, mobilize investment, and support the adoption of responsible, inclusive, and sustainable data solutions across the African continent. This year’s theme was “Kujenga na Data. Building with Data,” and at this year’s Data Tamasha Africa, I moderated a funding panel, and we ran a masterclass titled “Monitoring, Evaluation, Research and Learning for Artificial Intelligence Policymaking”. This year’s 3-day packed sessions explored how African communities are reclaiming control over their digital and data futures. In this blog, I share conversations from Data Tamasha Africa that stood out to me and should spark the interest of today’s African evaluator, and how this intersects with the work that the MERL Tech Initiative is doing. 

Building with Data and Creating Value for the Data for Africans: Interrogating the design of AI Data infrastructures 

As conversations about building with data in Africa continue, digital readiness deserves a central place. It is a pathway to understanding how we build with data and how we create value from it.

The keynote session named a disconnect in how we think about digital readiness in Africa, one that extends to AI readiness. Institutions collect data and use it to make an impact, but forget the people the data was collected from. We are borrowing from these communities. As we advocate on their behalf, we need to ask whether our impact indicators reflect their lived and digital realities. That means asking harder questions. Is this only about technology? How intersectional should this work be? Is the technology even needed in this context, and if so, is it built from the ground up in a language people understand? Meaningful data value creation should start and end with the community. 

In a related lightning talk, Farhana Alarakhiya took this further. Creating value from data is not only about who makes the decision. It is also about whose decision it is, who benefits, and on whose terms. Participation matters, especially with so much underway in building digital public infrastructure. But is participation enough, and where in the value chain do we participate? Right now, Africa exports data and imports “intelligence”. If we are borrowing data from communities, it must make a return journey, and this data must carry value that informs decision-making.

Farhan argued that this return journey depends on infrastructure, and that context is its core design element. Current AI deployment in Africa treats our contexts as edge cases. We need to treat them as design inputs instead. Without a context-first approach, we risk building technically brilliant solutions that don’t fit lived realities and end up unused. This happens when we copy and paste from high-income countries or assume all Global Majority issues are the same. We assume a portability that doesn’t fit the African context. We need to go further and ask whose context defines the design, and who gets to define the theory behind it.

For Farhana, this matters even more now that generative AI is everywhere. If its foundational language is not the language we speak, can it really be inclusive? These are design questions, and they are questions of power. Monitoring and evaluation practitioners can design these indicators that reflect lived realities, ask communities what success means to them, and build in ways for findings to return to the people who shared them. If data is to make a return journey, MERL is one of the places that journey can begin, and it can help make sure it ends back with the community. As we ask what it takes to build with data, we must also ask who gets to build, whose reality gets designed, and whether the value makes it back to the original data owners.

The politics that shape Africa’s digital realities

A session that stood out for me was the lightning talk by Nanjala Nyabola, because it invited everyone to think critically about who is left out of conversations on digitisation and data infrastructure. Digitalisation conversations are all happening in translation and in foundational languages that many don’t understand. How can anyone engage meaningfully when that is the case? Nanjala posited taking a step back to think about the theory behind the intervention, so that it becomes something we engage with, co-develop and co-construct. 

This is key to reframing how we understand, use, collect and deploy data. Too often, Africans are not invited into the theoretical part of the conversation but rather as end users and consumers, mostly as part of the product. The saying goes, “If the platform is free, if the software is free, then you are the product.” Rarely are we invited to ask: if we had all the power, agency, and creative energy, would we choose the same focus that AI hyperscaling is currently marketing to us? African language AI is a great example. Wouldn’t Africans be thinking about solving the issues closest to the lived reality of their communities?

The material reality of our shared digital future in Africa is that Big Tech constantly tries to convince us that our data and digital realities are removed from our offline, in-person realities. This is especially worrying with so much unchecked expansion of data collection as AI hyperscaling continues. Data collection is not neutral. It is a theoretical reflection of our beliefs about what data is, how it can be collected, and how it can be used. Because of this, Africans owe it to themselves to sit within these questions of theory, so that we can provide answers and context that reflect the priorities we want governing our data futures.

One way to do this, suggested in the talk, is through African feminisms. African feminisms centre lived realities and offer a critical thinking method that is plural and intersectional, which opens the possibility of different outcomes in digital and data conversations. A practical example was the environmental consequences of generative AI (as quoted above). The vast majority of Africans, particularly women, are subsistence agriculturalists who work closely with and whose livelihoods depend on land and water. Nanjala pointed to Wangari Maathai and how she advocated for completing socio-economic progress through thinking about the earth and water, and how political and economic models can only be considered complete once the environment is centered. She asked the room to apply the same consideration every time we use generative AI to create a poster or a knock-off R&B song. Thinking with the lived realities of African women illuminates questions that often get left behind, and these can feed into a richer theoretical context for the data reality we want to build.

What I took away was that Africa has scope for agency and creativity to shape our data realities from a lived reality standpoint. To achieve this, we should collectively not be afraid of the foundational questions, because Africa cannot afford for some of these questions to be subsidiary. Building what makes sense for Africa means rejecting digital monocultures that negate the possibilities that exist within our societies. It means making room for different ways of thinking to challenge us to do better and reflect our lived realities better as we move forward.  

For evaluators, Nanjala’s call to think about the theory behind the intervention will sound familiar, as it echoes the logic of a theory of change. M&E practitioners are well placed to ask the foundational questions about whose priorities shape our indicators, whose languages and voices are missing, and what our tools cost communities and the environment. In doing so, we can help ensure Africa’s data futures reflect the lived realities of the communities we serve. Ultimately, the hard and organic work within our own societies, led by this framing, will lead us to a better future than what is promised by the trajectory we are on today as a continent.

Evaluating Open Datasets to Advance AI Sovereignty, Close Data Gaps and Promote Responsible Data Governance

I also took part in a workshop hosted by the GIZ African Union Office on enabling data-driven AI innovation in Africa. The session focused on one central question: how can we make data actually power AI that creates value for Africa?

What I appreciated most was how practical the discussion was. Rather than treating AI as something that simply appears once the technology is ready, the session framed AI deployment as an outcome of a whole ecosystem, and that ecosystem starts with making data available 

This is where the Africa Open Data Readiness Index for AI (AODRIA) comes in. AODRIA is a benchmarking framework that evaluates datasets on how ready they are for AI, as well as their openness, governance, sustainability, and socio-economic impact. This data accessibility initiative’s goal is to empower African businesses, researchers, and AI innovators by making high-quality, openly licensed datasets easily discoverable and usable. The index aims to:

• Provide transparent and objective scoring of open datasets relevant to AI innovation in Africa
• Set a continental benchmark for dataset quality and accessibility
• Recognise and reward strong data stewardship
• Highlight gaps and drive improvements in African open data quality

To be considered, every dataset must first meet five baseline criteria: relevance to Africa, accessibility, machine readability, active status (meaning the data is still maintained), and language.

African evaluators are not often part of AI conversations on the continent, but this session made it clear that we should be. At its core, AI runs on data, and data is our everyday work. Practitioners design surveys, collect information from communities, clean and manage datasets, and track change over time. Much of the data that could make AI useful for Africa already sits in programme reports, baseline studies, and monitoring systems that monitoring and evaluation teams build and maintain. The session was highly interactive, and working through these ideas together helped us think more concretely about the steps needed to move from simply having data available to actually deploying AI.

AI Positionality for evaluators in the age of AI: MEL Capacity and Ecosystem Strengthening as a  practical first step 

My biggest takeaway from the conference was about where African evaluators currently stand in the AI conversation, and where we should stand. The work that we are doing with the Made in Africa AI in MERL approaches is a practical way for African practitioners to create our own theories that are radical and rooted in lived realities. AI is already reaching into African evaluation practice. It is changing how data is collected and analysed, and more humanitarian and development programmes and interventions are  AI-enabled. 

This is where the Made in Africa AI approaches in MERL come in. It gives African practitioners a practical way to build our own theories about AI, and it makes a strong case for strengthening MEL capacity and the wider ecosystem as the first step. When evaluators ground the standards, requirements, and approaches for adopting, integrating, or even resisting AI in MERL in theory, it creates room for homegrown definitions and solutions rather than borrowed ones.

We do not have to start from scratch. The Made in Africa Evaluation scholarship already gives us a theoretical base, and this is the foundation from which standards, requirements, and approaches for AI in MERL in Africa should be drawn. That is why positionality matters so much for the African evaluator. The modern African evaluator is well placed to guide, among other things, whether traditional evaluation methods and metrics should stay the same as AI becomes part of evaluation practice and as AI-enabled humanitarian interventions grow.

African evaluators should not be seen as sitting on the periphery of the African AI conversation. We sit at the centre of it, with the experience needed to define what Made in Africa AI looks like in practice.

We will be continuing this conversation in person at the European Evaluation Society Conference in Lille this October and at the African Evaluation Association Conference in Morocco this November . Be sure to catch us there and continue these conversations in person with us!