The potential for artificial intelligence (AI) to create both benefits and risks is increasingly the subject of international politics. The field of International AI Governance explores the opportunities and challenges of a world where AI is becoming increasingly important for many societies. The primary themes of this work are the sharing of AI’s benefits, the mitigation of AI’s risks, and the prevention of war.
International AI Governance faces additional challenges beyond those in domestic AI governance, including interstate competition and the heterogeneity of states. Interstate competition tends to circumscribe the potential for international cooperation due to concerns about unfair outcomes or the possibility of violence. State heterogeneity can be a problem because states can differ dramatically in many ways, including their governance style, governing ideology, and economic system. To improve humanity’s prospects for the international governance of AI, work in this field needs to navigate through these challenges.
The Oxford Martin AI Governance initiative has completed some work in this domain and more is forthcoming. Prior work includes a potential approach for harmonising civilian AI governance regulations and explorations of the roles that can be played by compute providers and the emerging network of AI Safety Institutes. Ongoing work includes explorations of how international AI governance agreements can be verified and how the benefits of AI can be shared.
