Exploring ethics risk in South Africa’s artificial intelligence industry : towards a risk governance framework

Abstract

The ubiquitous and swift growth of artificial intelligence (AI) coupled with the nature of the technology raises ethical risks for enterprises and for their stakeholders. Artificial intelligence's ethical risks are unlikely to be uniform across or within societies. The literature on AI ethics is, however, dominated by a universalistic, Global North outlook. This exploratory study aims to add to the discourse by providing a Global South perspective: examining domain-specific AI ethics risks in South African industry from an ethics risk governance perspective. The study uses an inductive, qualitative methodology to explore industry practitioners and related experts' views and approaches to AI ethics. A novel research instrument was used, and data was collected through semi-structured interviews. The data was thematically analysed, and several salient themes identified. Theoretically, the study relates AI ethics to business ethics, ethics risk governance, and the King Code. Empirically, the study provides a multi-level (universal, country, and industry) view of AI ethics risks, and identifies relevant external and internal industry governance factors. It maps the prevailing AI ethics management practice in South Africa, which is found to be informal and ad hoc, albeit with nascent signs of a more structured, tailored approach. It also compares the South African findings to that of the dominant Global North literature. The study proposes a South African-centric, high-level conceptual framework for AI ethics risk governance, which can be contextually adapted for wider relevance. The study makes policy recommendations to industry and government to control, govern, and manage AI ethics risks.