See, Infer, Intervene: Proactive World Modeling for Goal-Oriented Social Intelligence
Researchers have developed a framework called See--Infer--Intervene (SII) to enable multimodal retail agents to proactively assist customers. The Proactive Intent World Model (PIWM) within this framework uses psychological fields like AIDA and BDI to predict customer intent and select appropriate responses. A new benchmark, GuidanceSalesBench, was created to evaluate this system, with PIWM achieving a 0.641 macro F1 score when conditioned on ground-truth customer state, outperforming a Qwen2.5-VL-7B baseline. AI
IMPACT This research could lead to more intuitive and helpful AI assistants in retail environments, improving customer experience.