MERIT Feedback Elicits Better Bargaining in LLM Negotiators

2026-02-11

AI Summary

This work addresses the limited strategic depth of current large language models in negotiation tasks, their difficulty in adapting to complex human behaviors, and the absence of effective benchmarks for evaluating alignment with human preferences. To this end, the authors propose a utility-feedback-driven framework for enhancing negotiation capabilities, introducing AgoraBench—a new multi-scenario negotiation benchmark—and designing utility-theoretic evaluation metrics such as agent utility, bargaining power, and acquisition ratio. The approach integrates prompt engineering, supervised fine-tuning, and a dataset aligned with human preferences. Experimental results demonstrate that the proposed method significantly improves the model’s strategic reasoning and opponent awareness, yielding negotiation strategies that better reflect human preferences and outperform existing baselines.

Technology Category

• Multiagent Systems: Agreement, Argumentation & Negotiation
• Humans and AI: Learning Human Values and Preferences
• Knowledge Representation and Reasoning: Preferences

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Bargaining is often regarded as a logical arena rather than an art or a matter of intuition, yet Large Language Models (LLMs) still struggle to navigate it due to limited strategic depth and difficulty adapting to complex human factors. Current benchmarks rarely capture this limitation. To bridge this gap, we present an utility feedback centric framework. Our contributions are: (i) AgoraBench, a new benchmark spanning nine challenging settings (e.g., deception, monopoly) that supports diverse strategy modeling; (ii) human-aligned, economically grounded metrics derived from utility theory. This is operationalized via agent utility, negotiation power, and acquisition ratio that implicitly measure how well the negotiation aligns with human preference and (iii) a human preference grounded dataset with learning pipeline that strengthens LLMs'bargaining ability through both prompting and finetuning. Empirical results indicate that baseline LLM strategies often diverge from human preferences, while our mechanism substantially improves negotiation performance, yielding deeper strategic behavior and stronger opponent awareness.

Problem

• bargaining
• Large Language Models
• strategic depth
• human factors
• negotiation

Innovation

• utility feedback
• AgoraBench
• human-aligned metrics
• LLM negotiation
• preference-based learning