Data, AI and elections in Africa: working towards a collaborative African response

Artificial intelligence (AI) and other digital technology tools are increasingly being deployed to influence African politics. Civil society, government, the media, and other stakeholders must work together to understand how these tools manipulate public opinion and election outcomes to mitigate these risks. 

RIA, in collaboration with International Media Support (IMS) and the Africa democracy team of the Swedish International Development Cooperation Agency, held two workshops in mid-November with a wide range of stakeholders including electoral experts, diplomats, African media, data scientists and African digital rights organisations, among others. The goal of the workshops was to build dialogue across silos and identify points for co-operation to support African democracy in the time of AI. 

The first workshop focused on finding the nexus between these diverse areas of work, while the second sought to unearth how African researchers can access data held by tech companies to conduct research on disinformation distributed on digital platforms.

AI, disinformation and African elections

The discussions in the first workshop highlighted the challenges that each stakeholder group is facing in their work to enhance African democracy, as well as some of the initiatives they are undertaking. 

Media stakeholders spoke of how the capacity of traditional media to fight disinformation in Africa’s information sphere is affected. Much of media advertising is now moving to platforms, and journalists face harsh working environments; with some of them continually fighting for press freedoms. Coalitions are widely seen as the way ahead in the new content ecosystem.

The Association of Electoral Management Bodies in Africa (EMB) highlighted that misinformation discredits their work by undermining public trust in the electoral process. The organisation has come up with social media guidelines, seeking to address the behavior of state actors, social media companies as well as the public, in their social media interactions during the elections period. The guidelines call for tech companies to release their data to African researchers so that they are able to conduct research into the elements of information disorders affecting election processes. The EMBs hope to popularise the social media guidelines so that some of the provisions go into local laws and regulations.

“The right of access to public interest information and data is critical to the issue of misinformation,” said Professor Guy Berger, distinguished fellow at RIA, speaking at the event. “If we want to understand what is going on in social media platforms, we need to access their data. Some big tech companies offer data access to Europe and America, but not to Africa, and this needs to be addressed.”

Liz Orembo, a research fellow at RIA, discussed the role of the Africa Internet Governance Framework (AfIGF) in shaping policies in Africa. The discussions from the AfIGF inform a unified African voice that is then brought forwards at the global IGF. RIA, in partnership with the African Union – the AfIGF convening body – and the German Agency for International Cooperation (GIZ) has worked on an African data governance framework that seeks to address how data use and generation in Africa can address the continent’s development priorities while protecting the rights of citizens. 

The demand, supply, and use of data by African researchers to understand disinformation during elections

While extensive research on data’s role in elections is common in the West, it is nascent in Africa. The second workshop was part of RIA’s advocacy for an  African Alliance for Data Access. It was informed by RIA’s research that looked at the demand by African researchers for the data held by platform companies. This project was conducted in partnership with the Action Coalition on Meaningful Transparency (ACT) of which RIA is a member. The workshop involved hands-on assessment about using threat analysis and data to research information disorders during elections. 

What emerged is how  the importance of qualitative and quantitative data, analysed within the specific local contexts, is crucial to really understand the nuances of disinformation, as well as its manifestation. This means researchers need greater access to platform data, as well as data collected through more traditional means such as interviews. On a positive note, the interactions between data scientists, journalists and digital rights organisations revealed some low hanging fruit for future collaborations. 

The way forward

Tackling the challenge of disinformation will require evidence-based knowledge to reveal whether and how social media and AI can damage electoral integrity and credibility and how to mitigate the threats. This will will require a multidimensional understanding of forces and patterns in the production and spread of competing narratives. RIA intends to build on the foundations laid in this two workshops to fill the gaps in knowledge in this important area.