Governing artificial intelligence ethically: A fuzzy DEMATEL analysis of Australia’s AI principles in the Indonesian policy context

Abstract

The rapid diffusion of artificial intelligence (AI) in Indonesia has outpaced the development of robust ethical governance mechanisms. Although Indonesia has issued ethical guidelines for AI, these remain largely declarative and lack operational enforcement. In contrast, Australia’s AI Ethics Principles provide a more structured and implementable framework. However, the interactions among these ethical principles and their relative importance within the Indonesian policy context remain insufficiently understood. This study examines the causal relationships among AI ethical principles adapted from Australia’s framework to identify key governance drivers for strengthening ethical AI implementation in Indonesia. A quantitative perception-based study was conducted using survey data from 109 Indonesian AI stakeholders, including developers, practitioners, and users. The Fuzzy DEMATEL method was employed to analyze causal relationships among eight AI ethical principles. The results indicate that Accountability, Transparency and Explainability, Contestability, and Privacy Protection and Security act as causal drivers within the ethical governance system, while Fairness, Human-Centered Values, Reliability and Safety, and Human, Societal, and Environmental Wellbeing function as outcome-oriented principles. These findings suggest that ethical outcomes depend primarily on strong upstream governance mechanisms. The study provides empirical guidance for localizing international AI ethics frameworks and advancing actionable AI governance in Indonesia.