Public sector 1

Stanley Mpofu, University of the Witwatersrand

The role of AI in the public sector is still uncertain as policy, capability and trust are still trying to catch up with the technology. This is not a problem exclusive to South Africa; most governments are still defining the rules and using AI in narrow internal functions rather than as core technology on the frontline. The technology is perceived as strategically important, but uneven skills, legacy systems and low data readiness are slowing down meaningful and scaled adoption. The data challenge is particularly important because AI requires training data to learn from and this data must be clean, well-governed, shareable and ethical, which means that data management must evolve in the public sector so AI can benefit its citizens.

South Africa’s Digital Transformation Roadmap frames data as a strategic asset and positions Digital Public Infrastructure – digital IDs, payments and data exchange rails – as the backbone for AI-ready government services. The roadmap emphasises moving from fragmented departmental systems to interoperable platforms so that citizen data can move securely across departments, which is a precondition for any cross-cutting AI use case. If, for example, social grant eligibility or means tests were to be automated and interpreted accurately with AI, data needs to be accessible and interoperable across SASSA, SARS, Home Affairs, banks and other custodians through governed data-sharing mechanisms, not manual extracts. It also needs to be clean and held within clean governance lines.