Governance for Artificial Intelligence (AI) and Interoperability: Questions of Trust

Abstract

Although the rapidly emerging capabilities of AI bring potential benefits that could be transformative for cyber security, significant threats have emerged that continue to grow in impact and scale. One proposed solution to addressing important risks in AI is the emergence of strategies for AI governance. Yet, as this conceptual early-stage research argues, what is crucial for individuals, businesses, public institutions, including the military, and for high-risk environments, are questions concerning trust in AI governance. Will governance of AI be trusted? As an example, during 2023, several AI governance initiatives and strategies emerged, with some nation states proposing legislation while others looked to treaties and collaboration as solutions. Indeed, at a supra-national level, the United Nations expert multinational stakeholder Policy Network on AI (PNAI) formed to examine key issues in current AI governance. These include the interoperability of governance, data governance mechanisms, AI in supporting inclusion and the transition of nations. To help our understanding of trust in AI governance, the focus for this paper is limited in scope to interoperability in AI governance. Interoperability encompasses different aspects, policy initiatives (such as frameworks, legislation, or treaties), systems and their abilities to communicate and work together. The approach taken in this early-stage research is framed as questions of trust in AI governance. The paper therefore reviews the nature of different AI governance strategies developed and implemented by a range of key nation states and supra-national actors. This is followed by an evaluation of the role of trust, focused on AI governance strategies, in the context of interoperability in AI governance. Trust-building strategies are also considered, with a focus on leveraging the separate elements involved in trust-building to assist our understanding of the implementation of trusted AI governance. The contribution of this early-stage research is to highlight issues that may not be considered by the technical community and to contribute to developing a platform and a research approach that informs policy- learning for institutions, practitioners and academics.