Navigating Rifts in Human-LLM Grounding: Study and Benchmark

Omar Shaikh, Hussein Mozannar, Gagan Bansal +2

Language models excel at following instructions but often struggle with the collaborative aspects of conversation that humans naturally employ. This limitation in grounding -- the…