#human-agent

LLMs Can Infer Unspoken Intent in Collaborative Tasks

Researchers tested whether large language models can interpret incomplete instructions by reasoning about a human partner's mental state, matching human performance.