Evaluating the Impact of AI-Driven Solutions on the Efficiency and Equity of Social Grant Distribution in South Africa

Abstract: This study evaluates the impact of Artificial Intelligence (AI)-driven solutions,specifically the Electronic Know Your Client (EKYC) and biometric systems, on the social grantdistribution process in South Africa. Piloted in Gauteng, the Eastern Cape, and the North West,these AI-driven technologies were introduced to combat fraud and streamline the identificationprocess. However, current challenges with system integration, particularly with SASSA’s legacySOCPEN system, hinder full effectiveness. This study proposes a multi-layered AI modelstructured around Data Collection, AI Processing, Decision-Making, and Feedback Mechanismsto enhance system functionality, improve fraud detection, and support equitable grant distributionacross South Africa. Through semi-structured interviews with key stakeholders, including SASSAofficials and IT personnel, the study assesses the effectiveness of these solutions in reducing fraud,improving system efficiency, and ensuring that grants reach legitimate beneficiaries. Findings from the pilot reveal varying levels of success across the three provinces. In Gauteng,fraud detection improved by 5%, but system integration issues hindered further progress.Stakeholders believe that fraud detection could rise to 30% with better system interfaces. EasternCape saw minimal progress due to frequent system downtimes, with stakeholders highlighting theneed for infrastructure upgrades and system reliability improvements. Awareness campaigns inthe province are helping to familiarise beneficiaries and officials with the system, but manualprocessing remains prevalent. North West showed a 10% reduction in identity theft, but fraudtransactions and incorrect payments continue to be major issues, indicating the need for strongeraccount verification measures.