A Scoping Review of Transparency and Explainability in AI Ethics Guidelines

Abstract

Transparency and explainability are crucial tenets of ethical Artificial Intelligence (AI) and are often classified as technical components of AI Ethics. Many countries and international governing bodies have developed AI guidelines and principles that are made public for respective civilians with a diverse range of expertise and knowledge. This short paper compares how explainability and transparency are presented and discussed in AI ethics guidelines developed by the top ten countries leading in AI research and development according to the AI Global Index Report in 2023 and leading efforts from governing bodies such as the EU. Methodologically, this paper presents a thematic analysis focusing on the presence and acknowledgment of various dimensions of explainability and transparency, as well as the level of detail and examples in the guidelines. The aim is to uncover how various aspects of AI ethics are presented in global guideline documents, highlight areas in which the documents converge, and discuss implications for AI stakeholders.