Agents refuse some gifts to balance cooperation and competition
AGiR: Mitigating Gift Over-Reliance in Mixed-Motive Games
Multiagent SystemsMachine Learning
Summary
Sometimes, computer agents that play games or solve problems together rely too much on gifts from others instead of learning how to earn rewards themselves. The authors found this makes agents vulnerable and less effective. They designed a way for agents to sometimes refuse gifts, which helps them stay strong and fair while still cooperating. This method works with many existing systems without needing big changes.
What this means in practice
- •For multi-agent system developers: Improve agent training in mixed cooperation and competition settings by integrating refusal of some gifts to prevent dependency and encourage reward-earning behaviors.
- •For game ai designers: Create more balanced AI opponents or collaborators by using refusal-augmented gifting to avoid exploitation and support fairer gameplay strategies.
Authors
Woohyeon Byeon, Seongmin Kim, Jiwon Jeon, Woojun Kim, Youngchul Sung
Abstract
In this paper, we consider mixed-motive games (MMGs) which model various real-world multi-agent scenarios where self-interested agents must both cooperate and compete. We identify an issue with existing gifting methods, where agents become overly dependent on incoming gifts and fail to learn reward-yielding behaviors, a phenomenon we call gift over-reliance. To address this, we propose a refusal-augmented gifting framework that balances cooperation and competition in MMGs. Our approach introduces an adaptive refusal mechanism that allows agents to selectively reject a portion of received gifts from their policy-update signals, thereby mitigating gift over-reliance and preventing the agents from being exploited by other agents. This mechanism is gifting-method-agnostic, serving as a plug-in for existing schemes while enabling decentralized learning. We demonstrate that across diverse MMG environments and gifting schemes, our mechanism consistently reduces gift over-reliance and prevents lock-in to sacrificing policies, improving $α$-fairness as a result, without necessitating major changes to existing training pipelines.