Assistive Large Language Model Agents for Socially-Aware Negotiation Dialogues

2024-01-29

Conference on Empirical Methods in Natural Language Processing

AI Summary

Social norm violations in business negotiations frequently lead to agreement failure. Method: This paper proposes a fine-tuning- and annotation-free LLM-assisted negotiation framework featuring dual-role agents simulating negotiation dialogues and a third-party “Repairer” agent that detects and rewrites socially norm-violating utterances in real time. Innovatively, a “value impact” criterion guides in-context learning (ICL) example selection to enable socially aware, dynamic intervention. The approach integrates role-aware dialogue modeling, unsupervised ICL, explicit social norm encoding, and utterance-level rewriting. Contribution/Results: Evaluated on three real-world business negotiation tasks, the framework significantly improves both agreement rates and bilateral satisfaction. High-quality multi-turn negotiation datasets and implementation code are publicly released.

Technology Category

• Multiagent Systems: Agreement, Argumentation & Negotiation
• Natural Language Processing: Safety and Robustness
• Humans and AI: Learning Human Values and Preferences

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Search and Retrieval-Augmented AI: Agentic search

Abstract

We develop assistive agents based on Large Language Models (LLMs) that aid interlocutors in business negotiations. Specifically, we simulate business negotiations by letting two LLM-based agents engage in role play. A third LLM acts as a remediator agent to rewrite utterances violating norms for improving negotiation outcomes. We introduce a simple tuning-free and label-free In-Context Learning (ICL) method to identify high-quality ICL exemplars for the remediator, where we propose a novel select criteria, called value impact, to measure the quality of the negotiation outcomes. We provide rich empirical evidence to demonstrate its effectiveness in negotiations across three different negotiation topics. We have released our source code and the generated dataset at: https://github.com/tk1363704/SADAS.

Problem

• Enhancing business negotiation dialogues
• Role-playing agents for negotiation practice
• Improving outcomes with remediator agents

Innovation

• LLM-based negotiation agents
• Remediator agent rewrites norms
• In-Context Learning method