Algorithmic Collusion by Large Language Models

2024-03-31

arXiv.org

AI Summary

This study reveals that large language model (LLM)-driven algorithmic pricing agents can spontaneously collude in oligopolistic markets—achieving supra-competitive prices and profits without explicit coordination. Method: We develop a multi-agent game-theoretic experimental framework, employing prompt engineering for controlled variable manipulation and introducing a novel “off-path behavior attribution analysis” to identify fear of price wars as the key mechanism driving collusion. Contribution/Results: We provide the first empirical evidence that LLM agents converge rapidly—within a few interaction rounds—to a stable supra-competitive equilibrium, increasing profits by 30%–200%. Crucially, minor prompt modifications alter collusion strength by over 50%, demonstrating high prompt sensitivity. Results replicate robustly across both English and Dutch auction formats. These findings uncover emergent, implicit, and instruction-sensitive characteristics of generative AI–enabled collusion, offering critical empirical grounding for AI antitrust regulation and policy design.

Technology Category

• Multiagent Systems: Mechanism Design
• Game Theory and Economic Paradigms: Coordination and Collaboration
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

The rise of algorithmic pricing raises concerns of algorithmic collusion. We conduct experiments with algorithmic pricing agents based on Large Language Models (LLMs). We find that (1) LLM-based agents are adept at pricing tasks, (2) LLM-based pricing agents quickly and autonomously reach supracompetitive prices and profits in oligopoly settings, and (3) variation in seemingly innocuous phrases in LLM instructions ("prompts") may substantially influence the degree of supracompetitive pricing. Off-path analysis using novel techniques uncovers price-war concerns as contributing to these phenomena. Our results extend to auction settings. Our findings uncover unique challenges to any future regulation of LLM-based pricing agents, and generative AI pricing agents more broadly.

Problem

• LLM-based agents autonomously achieve supracompetitive prices in oligopolies
• Minor prompt variations significantly impact supracompetitive pricing levels
• Price-war concerns contribute to LLM-based algorithmic collusion phenomena

Innovation

• LLM-based agents for pricing tasks
• Autonomous supracompetitive pricing in oligopolies
• Prompt phrasing influences pricing outcomes