From Overload to Convergence: Supporting Multi-Issue Human-AI Negotiation with Bayesian Visualization
A new research paper explores the challenges humans face in multi-issue negotiations mediated by AI, finding that performance degrades beyond three issues due to increased cognitive load. To mitigate this, the paper introduces a Bayesian visualization tool that estimates agreement probability and visualizes the narrowing space of mutually acceptable outcomes. Experiments with 32 participants demonstrated that this visualization improved negotiation outcomes, efficiency, and maintained human control without altering value distribution. AI
IMPACT This research offers design guidance for AI systems mediating complex negotiations, potentially improving user experience and outcomes in such scenarios.