Decolonising data: ensuring AI reflects Africa's rich diversity in food systems and ethics

From 7 July 2025 until 20 July, the annual Food Indaba has as this year’s theme the potential impacts and opportunities of artificial intelligence (AI) on African food systems.

On 9 July, an online Conference on AI, Knowledge & African Food Systems took place, featuring host Khanya Mncwabe, the CEO and co-founder of Matawi.

Alison Pulker, a research assistant at the African Centre for Cities, Dr Anesu Makina, Postdoctoral Researcher at the African Centre for Cities, and Gareth Haysom, senior researcher at the African Centre for Cities, as well as Associate Professor in food security from the University of Namibia, Ndeyapo Nickanor, were panellists in the discussion. 

Pulker described the food system as everything from growing food to waste. She went on to define an urban system as things like transport into urban areas, electricity needed to store food, housing and social infrastructure — how food is distributed to people within a city.

The food environment is how people can choose food; therefore power and policy come into play, with zoning laws for where food can be sold being an example used by Pulker. 

Data from African cities

Haysom introduced an AfriFOODlinks project, which looks at the city food systems in hub cities in countries like Burkina Faso, Tunisia, Uganda, Kenya and South Africa. Those cities then work with 10 more cities, and five European cities. They study what food systems in Africa need, feeding that information back to the public, and working with city officials.

A report found that the world had predominantly transitioned to an urban environment by 2007, something Haysom found striking. Of the 2.2 billion food-insecure people, 1.7 billion live in urban and peri-urban areas. 

“There’s an absence of data from African urban areas,” said Nickanor on the politics of data. She noted that it was important to look at bias, transparency, academic integrity and intellectual property when thinking about AI. 

African ethics like ubuntu could be embodied when deploying AI, mused Makina, specifically concepts like human dignity and equitability. There should be discussions around monitoring, and systems needed to be tested locally, because there was diversity between countries and linguistic diversity in Africa. 

Big AI systems did not include marginalised people, and already showed a gender bias, Makina pointed out.

Owning your data — and your life

Haysom spoke of the evolution of society; first we transformed our lives with materials, then with energy, and now with information. From fire to fossil fuels to the Fourth Industrial Revolution. 

During industrialisation, the people who became known as the luddites destroyed machines because they feared they would replace labour. Haysom questioned if we should resist in a similar way, or embrace AI. 

He added that governance had to catch up, with rules and regulations, to technological development. He also said it would be important to ensure that technology worked in our interest, and did not follow the extractive pattern that had befallen Africa before. 

Extractivism generally refers to the raw minerals and material that are minimally processed before being shipped out to other countries.

Pulker said that during her research they ran into an ethical consideration: how to protect data taken from people, and, during storage of that data, how to create long-term anonymity.

Mncwane queried Makina on how to decolonise these systems.

“When we exclude people, the information is not good. Scholars should operate on principles like fairness… and create systems with AI, asking questions rooted in our own realities,” said Makina. She cited examples of chatbots for farmers that communicate in their own languages, and a basic phone that could detect pests, but remain low-tech for accessibility. 

“At the policy level, the government should step up with infrastructure first, because people can’t participate,” said Makina. 

Haysom said we needed to understand how we brought bias to systems; if we thought someone buying amagwinya meant that people were lazy, or if we thought that people were incredible strategic decision-makers, then we were asking AI questions that reflected this bias.  

AI and extractivism in Africa 

Daily Maverick asked the panel how we could ensure that AI worked in our collective interest, and did not turn into an extractive system in Africa.

“I think it sits on what we value and devalue. Where we assign value,” said Haysom. “I think we need to work hard to amplify the value that we have; the value of our system is being eroded for a variety of reasons. 

“I also want to be pragmatic and acknowledge that people are making decisions that might seem to be undervaluing our food system, but they are making decisions because other systems are not supporting them,” he said. 

“How we as a society demand something fundamentally different in terms of governance; that governance and food systems, the laws, link to what is in the constitutions of our different countries, link to the Bill of Rights, link to economic and social justice, how do we embed AI data in the thinking of all of those processes so we can demand very different governance?”

He questioned how we challenge the disposable nature of the food system, start to see our bodies as being just as polluted as our atmosphere, and how we could start valorising local and indigenous foods in different ways — the ways that were thrown out by colonialism because they did not suit the economic model. 

Makina said there was a need for a strong governance framework, and beyond the state with organisations and individuals. 

“I know that policy frameworks are important, we’re right at the beginning of developing policy frameworks around AI systems, but we also need strong monitoring frameworks by diverse people — social justice practitioners and academics,” said Makina. 

“We need to ask the questions ‘what is our interest?’ and ‘what is our interest in the food system?’. AfriFOODlinks is a project that has managed to show that Africa’s food systems are not homogenous, in the same way that Africa is not homogenous,” said Pulker. 

“Africa’s diverse urban food systems are maybe something that we need to find out a bit more about before imposing what we think we’re protecting,” said Pulker. DM