Abstract
The integration of Artificial Intelligence (AI) into educational assessment has accelerated in recent years, where AI tools are being incorporated into assessment as learning (AaL), assessment for learning (AfL) and assessment of learning (AoL). The study is contextualised within South Africa’s developing higher education AI policy environment, with particular attention to ethical governance, data sovereignty and the mitigation of historical socioeconomic disparities. The theoretical framework guiding the study includes the intelligent Technological Pedagogical Content Knowledge (TPACK framework and constructivist theory, which enabled the study to examine how personalised learning pathways, immediate feedback and predictive analytics are facilitated by AI tools such as generative models and intelligent tutoring systems. Grounded in a structured literature review approach, the study adhered to the guidelines and employed thematic synthesis of peer-reviewed articles published between January 2020 and July 2026. The findings indicate a distinct implementation gap, where knowledge of AI systems does not necessarily translate into pedagogical integration, highlighting the need for a shift towards authentic assessment methods and targeted teacher professional development. The study concludes that while AI presents significant potential for curriculum reform and decolonisation, its effective implementation in South African higher education depends on context-specific innovation that ensures equitable access and upholds human-centred learning principles.
Keywords: AI in Educational Assessment, Artificial Intelligence, Constructivism, Curriculum Practice, TPACK.