The Global AI Summit on Africa, held in Kigali, Rwanda, brought together policymakers, innovators, researchers, and civil society leaders to discuss the future of artificial intelligence on the continent. While much of the conversation centered on the potential of AI to drive economic and social progress, important reflections also emerged on the need for regulatory innovation. In particular, the discussions highlighted the urgent need for new, context-appropriate models of data governance to ensure that AI innovation is responsible, equitable, and impactful. In the reflection below, incoming Executive Director Pria Chetty shares her thoughts on how data governance surfaced as a critical theme at the Summit.
At the recent Summit, AI innovation conversations had a way of drifting to conversations on the need for regulatory innovation, specifically the need for new options for data governance that enable AI innovation — and build into AI innovation mechanisms for responsible, equitable, and relevant AI. Only then can AI innovation deliver on the promise of economic and social dividends for our continent. Here’s how data governance came up.
At the heart of the data governance challenges were the conditions for data access and use, referred to as so-called legitimate data access. Legitimate data access and use respects the limits of sovereign data rights. Data access and use are legitimated by a good cause, such as pandemic responses. Data access and use pursue economic or social value for the data subject, or offer beneficial participation in digital opportunities such as AI-enabled opportunities.
Assurance, predictability and safety
Largely, we need data governance frameworks to assure us of the legitimacy we so desperately seek. Data governance should bring predictability to the conditions for data access and support conformance with legal and best practice standards. Add to this embedded rules that facilitate data access and use that are safe, ethical, and inclusive.
Interoperability for data exchange with purpose
Frameworks of interoperability — technical and otherwise — are needed for meaningful access and data use for economic and social value extraction, across interests.
We need to enable engagement with data subjects
Why? Legitimate data access alone does not equate to positive data outcomes. We need citizens to have agency over their data to maintain the quality of the data and solve challenges related to the currency, accuracy, and trustworthiness of data. In this sense, a public-private-people partnership should be advocated in data governance, where people (citizens) participate and have increasing levels of agency over the use of their data.
I’ll add my tick to reflections on the AI Summit calling for actions we can take forward. What are the potential regulatory innovations to take forward?
- Relationality of data principles
It strikes me that data protection principles of accountability, transparency, and data participation, currently conceived as distinct principles, need to be adjusted as having relational qualities. Access to data should be premised on transparency regarding its use, exchange, and reuse to enable meaningful data participation. This relationality may be developed into techno-legal solutions where transparency, participation, and evaluation and reporting of accountability co-exist. - Our data regulators are at a point of existential reflection.
For the individual who demands access to the data economy, what is the real mandate of the data regulator (data protection authorities) whose reason for existence is actually grounded in meaningful rights of privacy and access to information? Data regulators must bring clarity and cohesion to their mandates for their continued relevance and utility. Let’s be honest, their work is at a distance from the data activity we are seeing on the ground. - Data discovery should be a value in relevant data governance solutions
Where AI is ready to problem-solve access to healthcare, data — in the form of diagnostic data and data that evidences workable solutions — improves the AI proposition, learning, and innovation loops. Similarly, understanding dynamic data access and use alongside dynamic attributes of data is key to adaptable data governance for meaningful data access and use. How do we ensure that the data governance model is subject to dynamic evaluation and innovation? - Fully utilise licences and contracts
We’ll need a stronger utility from licences and contracts that govern data exchange. Existing licences and contracts need a makeover, to say the least — designed with current and future use cases in mind — as a unique opportunity to build and embed the understanding and impact of data exchanges. There are new templates for African data exchange that deliver data-driven impact just waiting to be conceived.
While the Summit sparked important dialogue, some of the conversations felt abstract, with actionable outcomes remaining just out of reach. Still, we leave with a few clear compass points to guide us forward: a willingness to engage deeply, a commitment to collaboration, and a shared responsibility to navigate the complex road from where we are to where we aspire to be — building meaningful, just AI for Africa.
