[4G4-OS-26a] OS-26

2:15 PM - 2:30 PM JST(5:15 AM - 5:30 AM UTC)

[4G4-OS-26a-03] A Negotiation Game for Experiential Learning of AI Governance in Human-Centered SocietiesUnderstanding Fairness, Accountability and Transparency through an AI Incident Scenario

〇Arisa Ema¹, Fumiko Kudo² (1. The University of Tokyo, 2. The University of Osaka)

Keywords:

AI Governance,Simulation,Bias

AI systems are increasingly used in socially sensitive domains such as recruitment and credit assessment, making public understanding and acceptance of governance issues—fairness, accountability, transparency, and trust—crucial. These challenges are intensified as practical and societal considerations intersect with technical and ethical concerns. In many cases, AI services involve both model developers and external service operators, creating conflicting interests and ambiguous responsibility boundaries. This study proposes a negotiation-based simulation game centered on a social media controversy involving an AI-powered job-matching service. Participants assume the roles of a service provider and an AI developer, negotiating responses to issues such as fairness measures, accountability, and service continuation under time pressure. The presentation explains the game’s design rationale and why negotiation-based gameplay supports understanding and acceptance of AI governance. Preliminary results from test plays with students and professionals suggest improved awareness of governance challenges and decision-making under uncertainty. This study contributes to the design of human-centered AI education and dialogue frameworks.