Instigating Cooperation among LLM Agents Using Adaptive Information Modulation

2024-09-16

arXiv.org

AI Summary

In multi-agent engineering systems, conflicts arise between individual rationality and collective welfare, exacerbated by the dynamic uncertainty of social dilemmas that static governance rules cannot adequately address. Method: This paper proposes a dynamic governance framework integrating strategic large language model agents (SLAs) and prosocial policy agents (PPAs). It introduces an information-adaptive modulation mechanism wherein PPAs—acting as governance agents—dynamically regulate inter-agent information transparency during iterative games to foster cooperative evolution. Contribution/Results: Experiments on canonical social dilemmas (e.g., Prisoner’s Dilemma) demonstrate a significant increase in cooperation rates; SLAs exhibit fine-grained strategic adaptability; and PPAs successfully learn optimal information-modulation policies, yielding an average 37.2% improvement in social welfare. This work pioneers information-transparency–driven cooperative emergence in multi-agent systems and establishes a novel paradigm for governance of LLM-augmented socio-technical systems.

Technology Category

• Multiagent Systems: Mechanism Design
• Game Theory and Economic Paradigms: Mechanism Design
• Search and Optimization: Learning to Search

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
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

This paper introduces a novel framework combining LLM agents as proxies for human strategic behavior with reinforcement learning (RL) to engage these agents in evolving strategic interactions within team environments. Our approach extends traditional agent-based simulations by using strategic LLM agents (SLA) and introducing dynamic and adaptive governance through a pro-social promoting RL agent (PPA) that modulates information access across agents in a network, optimizing social welfare and promoting pro-social behavior. Through validation in iterative games, including the prisoner dilemma, we demonstrate that SLA agents exhibit nuanced strategic adaptations. The PPA agent effectively learns to adjust information transparency, resulting in enhanced cooperation rates. This framework offers significant insights into AI-mediated social dynamics, contributing to the deployment of AI in real-world team settings.

Problem

• Addresses social dilemmas between individual and collective interests in multi-agent systems
• Overcomes limitations of static governance in dynamic autonomous AI environments
• Enables cooperation through adaptive information control while preserving agent autonomy

Innovation

• Adaptive governance mechanisms integrated into system design
• Separation of agent interaction and information flow networks
• Reinforcement learning agent dynamically modulates information transparency