Microsoft Corporation
Challenges in Human-Agent Communication
Pages
27
Time to read
68 mins
Publication
Language
English
Pages
27
Time to read
68 mins
Publication
Language
English
This document is a research article that identifies and analyzes twelve key communication challenges associated with modern autonomous agents. These agents, powered by generative foundation models, can interact with users through natural language but introduce complexities in human-agent communication. The challenges are categorized into three main areas: issues related to information that agents need to convey to users, difficulties in enabling users to communicate their goals and preferences to agents, and overarching challenges that affect all aspects of human-agent interaction. The article emphasizes the importance of establishing common ground through effective two-way communication to enhance transparency and control over agent behavior. It discusses the implications of these challenges for user trust and the need for new design patterns and guidelines to improve human-agent collaboration. The findings highlight critical gaps in current research and suggest directions for future studies aimed at addressing these communication issues.