Assistive Large Language Model Agents for Socially-Aware Negotiation Dialogues
This work addresses the challenge of enhancing socially-aware dialogues in business negotiations for AI agents, representing an incremental advancement in applying LLMs to specific domain tasks.
The authors tackled the problem of improving business negotiation outcomes by developing assistive LLM agents, including a remediator agent that rewrites norm-violating utterances, and introduced a tuning-free In-Context Learning method with a novel value impact criterion for selecting exemplars, achieving effectiveness across three negotiation topics as evidenced by empirical results.
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.