講演情報
[5O2-IS-5a-02]Can Space-Perspective Priming Mitigate Conflict Between AI Agents? An Empirical Study of Overview Effect Interventions in Cross-Cultural DialogueA Multi-Condition Comparative Experiment Using Structured Knowledge from Public Space Tourist Testimonies
〇Noriaki Inami1 (1. Astraporta, Inc.)
regular
キーワード：
Overview Effect / Space-Perspective Priming、Cross-Cultural AI Agent Dialogue、Conflict Escalation Mitigation、Multi-Agent Simulation、Cognitive Framing Intervention
LLM-based AI agents are increasingly used in negotiation and decision-making tasks. However, interactions between agents conditioned with differing cultural norms and value priorities may escalate into conflict, reflecting patterns seen in cross-cultural human disputes. This study examines whether adopting a “space perspective,” inspired by astronauts and commercial spaceflight participants, can mitigate such escalation. Recurring themes in public testimonies—such as Earth’s fragility, the thinness of its atmosphere, and a sense of shared human destiny—were structured and introduced as an intervention variable termed space-perspective priming. We conducted multi-condition experiments comparing no intervention, unilateral intervention, and bilateral intervention settings. Conflict was quantitatively assessed using indicators including aggressive language, accusatory framing, and dialogue breakdown. Results indicate that space-perspective priming reduces escalation and encourages constructive negotiation behaviors. The study offers an experimental proxy for exploring whether a shared planetary perspective may help reduce global conflict dynamics.