Abstract
The adoption of artificial intelligence in the public sector decision-making process is contested with transparency, accountability and trust issues, as the AI operation is underpins by the black box model. Although Explainable Artificial Intelligence (XAI) offers technical remedies to address the issues, evidence suggests that its success relies heavily on the ability of the effective data governance practices. Based on a systematic literature review on 30 key articles, this study is aimed at understanding the impact of XAI use in decision making and how the XAI dimensions impact the quality of the decision made. Findings reveal three main themes of how XAI has impacted decision making process: (1) Precise explanations due to enhanced data quality; (2) Translation of tacit knowledge into action plan for institutional legitimacy; (3) Autonomy empowerment through enhanced accountability structures. Moreover, XAI dimensions of transparency, interpretability, trust-building, and accountability significantly affect the quality of decisions. However, such impact could only be achieved through investments on organisational structures (data infrastructure and process) and actors’ capacity building. The findings underscore the idea that effective (XAI)-enhanced public administration requires the concomitant investment in technical capacities and institutional protection to enable the implementation of algorithmic transparency in a democratically accountable system for public benefit. The research signposts gaps for further studies, e.g. the mediating role of data governance in ensuring the effective use of XAI for quality decision making the public sector context.