Insert-expansions for Tool-enabled Conversational Agents

TL;DR

This paper investigates how integrating tools into large language models' reasoning improves conversational agents, proposing a 'user-as-a-tool' approach to enhance response relevance and effectiveness.

Contribution

It introduces the insert-expansion concept and empirically evaluates the 'user-as-a-tool' method to mitigate tool-induced distractions in conversations.

Findings

Benefits observed in recommendation tasks

Reduced sidetracking in tool-enabled conversations

Enhanced response relevance through user-as-a-tool approach

Abstract

This paper delves into an advanced implementation of Chain-of-Thought-Prompting in Large Language Models, focusing on the use of tools (or "plug-ins") within the explicit reasoning paths generated by this prompting method. We find that tool-enabled conversational agents often become sidetracked, as additional context from tools like search engines or calculators diverts from original user intents. To address this, we explore a concept wherein the user becomes the tool, providing necessary details and refining their requests. Through Conversation Analysis, we characterize this interaction as insert-expansion - an intermediary conversation designed to facilitate the preferred response. We explore possibilities arising from this 'user-as-a-tool' approach in two empirical studies using direct comparison, and find benefits in the recommendation domain.