Governance of artificial intelligence: A risk and guideline-based integrative framework

This study addresses the growing challenge of governing artificial intelligence (AI) arising from the risks that it increasingly poses to the public sector and society. Based on an in-depth literature analysis, we first identify AI risks and guidelines and classify them into six categories, including technological, data, and analytical risks and guidelines, informational and communicational risks and guidelines, economic risks and guidelines, social risks and guidelines, ethical risks and guidelines, as well as legal and regulatory risks and guidelines. These risks and guidelines are then elaborated and transferred into a four-layered conceptual framework for AI governance. The framework interrelates AI risks and AI guidelines by means of a risk management and guidance process, resulting in an AI governance layer depicting the process for implementation of customised risk mitigation guidelines. The framework constitutes a comprehensive reference point for developing and implementing AI governance strategies and measures in the public sector.