AI, Big Data and the African Food System: Why Public Accountability and Shared Infrastructure Matter More Than Ever
The Ukudla African German Centre for African German Centre for Sustainable and Resilient Food Systems and Applied Agricultural and Food Data Science hosted a robust dialogue moderated by CoE-FS co-director Prof Lise Korsten,. Here, eading researchers unpacked one of the most urgent questions in African food governance today: How can AI and big data support food-system resilience without deepening inequality, dependency and corporate control?Drawing on insights from; CoE-FS PI Dr Marc Wegerif, CoE-FS digital innovation lead Prof Clement Nyirenda, Ukudla postdoc fellow Dr Adele Barker and Ukudla PhD fellow Mercy Bwanaisa . Yhe session painted a sharp picture of data and the food system, a landscape rife with both possibility and peril.
The Limits of Big Data
A central theme was the mismatch between AI-driven systems and the realities of African food economies. As Mark highlighted, satellite imagery and large-scale datasets simply cannot “see” the human relationships that hold informal markets together. African food systems — especially the networks of street traders, public markets, small farmers and micro-enterprises — rely on face-to-face trust, relational credit, hyperlocal knowledge, and community networks.
These forms of intelligence are powerful, sophisticated and economically essential — but they are offline. As a result, they remain invisible to the very technologies now shaping global food policy.
Prof Clement Nyirenda Photo by Alaistar Russell
The Power Behind the Tools: Who Owns Africa’s Data Future?
A major concern raised throughout the dialogue was ownership. AI infrastructure, data platforms and software tools are overwhelmingly controlled by a handful of global corporate actors, none of whom are African. This creates a structural dependency that undermines sovereignty.
Prof Nyirenda illustrated the danger through a striking example: governments may fund data-science projects, only to receive final reports rather than the underlying software, models or data pipelines. They are then forced to return — and pay again — each time they need updates. Without access to the tools, raw data is effectively unusable.
Across the panel, there was clear consensus: Africa must be able to own, govern and reuse the tools that process its data. Software, code and analytical pipelines are themselves forms of data — and must be treated as public goods.
Dr Adele Barker at the Ukudla Panel.
Photo by Alaistar Russell,
Indigenous Knowledge, Bias and Incomplete Data
Complementing this Dr Barker stressed that big data systems privilege what is easily measurable and online. Indigenous knowledge, local experience, oral histories and community practices — the core of African adaptation strategies — rarely appear in these datasets. When AI systems are built on incomplete data, their outputs reinforce existing vulnerabilities and obscure the realities of those most reliant on informal markets.
This does not mean AI has no role. Rather, the panel urged the development of hybrid models that intentionally bridge ethnographic insight, indigenous knowledge and computational capacity.
Critical Infrastructure: High-Performance Computing for All
Prof Korsten steered the conversation toward a structural challenge: research infrastructure. Omics research, climate modelling and large-scale data analysis all rely on high-performance computing (HPC). Yet HPC is unevenly distributed and extremely expensive. Prof Nyirenda argued that South Africa’s fragmented institutional investments — each university trying to build its own HPC facility — are inefficient and exclusionary. He called for a national strategy
where institutions pool resources, including historically disadvantaged universities, to create shared, world-class infrastructure.
Such an approach, the panel suggested, is essential if African researchers are to compete globally and produce data-driven insights that reflect African priorities rather than external agendas.
A Changing Skills Landscape
The discussion also highlighted how rapidly skill demands are shifting. Coding and data science now attract participants far beyond traditional STEM fields, with lawyers, humanities scholars and social scientists joining training programmes.
The Ukudla Science Forum Stand Photo by Alaistar Russell
This shift is positive, but underscores the need for curriculum reform. The country also faces a shortage of system administrators capable of maintaining HPC environments. Without investment in these human capabilities, infrastructure alone cannot close the gap.
Citizen Science, Stokvels and Community Intelligence
A powerful intervention from a township economy leader emphasised the importance of citizen science and stokvel-based financial systems. Mark reflected on how micro-traders possess deep, relational knowledge that corporate data systems cannot capture — and may, in fact, seek to mimic or dominate.
This raised a broader concern: not all knowledge should be absorbed into digital systems. Some of the most resilient structures — such as cash-based lending, stokvels and community witnessing — derive their strength precisely from being offline, decentralised and resistant to surveillance or extraction.
Historical and Community Data for Resilience Planning
When Prof Korsten invited Ukudla fellow Mercy Bwanaisa to speak on resilience, she linked historical data and community-generated information to proactive policymaking. Long-term datasets reveal recurring shocks, changing intensities and land-use trends, while community-based data highlights how households adapt and cope.
Together, these forms of knowledge can guide governments toward long-term resilience planning rather than short-term crisis response.
Toward an African Vision for Data Justice and Food Sovereignty
Throughout the discussion, a shared conclusion emerged: if Africa is to use AI and big data to strengthen food systems, it must do so on its own terms. That means:
-
Owning tools, infrastructure and analytical pipelines
-
Integrating indigenous and community knowledge into data narratives
-
Building shared HPC systems across institutions
-
Growing the next generation of African data scientists and system administrators
-
Embedding public accountability into every layer of the data ecosystem
The dialogue, guided throughout by Prof Lise Korsten, underscored that Africa’s food systems cannot be shaped by tools that fail to see them. Responsible, accountable and locally grounded data innovation is essential — not only for resilience, but for sovereignty.
Photos by Alaistar Russell
