Breaking Algorithmic Collusion in Human-AI Ecosystems

2025-11-26

AI Summary

This paper investigates how human behavior affects the stability of algorithmic collusion among AI agents: specifically, whether introducing human participants employing no-regret learning—rather than AI agents—into a hybrid human-AI pricing ecosystem can disrupt AI-driven supracompetitive pricing. Adopting a repeated game framework, the study theoretically analyzes the interplay between equilibrium strategies and no-regret learning dynamics, and validates findings experimentally. Key results show that even a single rational human participant significantly undermines collusion stability; moreover, with only a negligible fraction of human agents, market prices rapidly converge to the competitive equilibrium. This work provides the first systematic characterization of the intrinsic fragility of algorithmic collusion in mixed human-AI environments. It offers a novel regulatory insight: rather than banning AI pricing outright, preserving minimal human autonomy in pricing decisions suffices to effectively curb tacit collusion.

Technology Category

• Humans and AI: Learning Human Values and Preferences
• Game Theory and Economic Paradigms: Coordination and Collaboration
• Multiagent Systems: Mechanism Design

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

AI agents are increasingly deployed in ecosystems where they repeatedly interact not only with each other but also with humans. In this work, we study these human-AI ecosystems from a theoretical perspective, focusing on the classical framework of repeated pricing games. In our stylized model, the AI agents play equilibrium strategies, and one or more humans manually perform the pricing task instead of adopting an AI agent, thereby defecting to a no-regret strategy. Motivated by how populations of AI agents can sustain supracompetitive prices, we investigate whether high prices persist under such defections. Our main finding is that even a single human defection can destabilize collusion and drive down prices, and multiple defections push prices even closer to competitive levels. We further show how the nature of collusion changes under defection-aware AI agents. Taken together, our results characterize when algorithmic collusion is fragile--and when it persists--in mixed ecosystems of AI agents and humans.

Problem

• Examines algorithmic collusion stability in human-AI ecosystems
• Assesses impact of human defections on price levels in repeated games
• Characterizes fragility and persistence of collusion in mixed settings

Innovation

• Human defection destabilizes algorithmic collusion
• Multiple defections drive prices toward competitive levels
• Defection-aware AI agents alter collusion nature