A new benchmark, ACII-DaiKon, has been introduced to evaluate models on interpersonal dynamics within dyadic conversations. This benchmark focuses on directional influence, turn-taking, and rapport development, moving beyond speaker-centric approaches. It utilizes the Hume-DaiKon dataset, comprising over 700 hours of audiovisual data across five languages, to support multimodal and temporal reasoning. AI
Summary written by gemini-2.5-flash-lite from 2 sources. How we write summaries →
IMPACT Establishes a new evaluation standard for conversational AI, pushing research towards more nuanced understanding of interpersonal dynamics.
RANK_REASON This is a research paper introducing a new benchmark and dataset for conversational AI.