An Environmental Review of the Generative Artificial Intelligence Policies and Guidelines of South African Higher Education Institutions: A Content Analysis

Abstract

Accompanying the inroads generative artificial intelligence (GenAI) models such as ChatGPT have made into the higher education sector, an urgent need has arisen to investigate the types of GenAI policies South African higher education institutions (HEIs) have developed in response to GenAI. To date, no study has explored this aspect of South African higher education. With this lack in mind, this paper reports on an online rapid environmental review of the GenAI policies of 26 South African HEIs that were freely available on the websites of these HEIs or otherwise online. The main purpose of the paper is to establish whether these HEIs had institution-wide GenAI policies, what types of policies they were and what constituted their contents. The study employed a critical-ethics-based framing comprising six dimensions: the Siyavuma, semi-Siyavuma, critical, semi-critical, uBuntu and semi-uBuntu dimensions. It analyzed data through content and thematic analyses. Some of its findings are worth mentioning. Firstly, it discovered that five of the 26 South HEIs had their institution-wide GenAI policy documents freely available on their websites or online; one HEI had four such policy documents. The retrieved GenAI policy documents were mainly guides or guidelines. Secondly, academic staff and students were the main target audiences of the GenAI policy documents. Thirdly, ChatGPT was the most mentioned and the most cited GenAI tool in the reviewed policy documents. Fourthly, the responsible use of AI tools, GenAI and academic integrity, and risks and concerns of using GenAI tools featured as one instance of the main convergence points for the GenAI policy documents that spelled out their aims and their main foci. Lastly, six of the GenAI policy documents manifested elements of the critical dimension, whereas one GenAI policy document has features of the uBuntu dimension. The paper also makes relevant recommendations.