Adaptive Mechanism Design: Learning to Promote Cooperation

TL;DR

This paper introduces a learning-based mechanism for external agents to promote cooperation among artificial learners in social dilemmas by dynamically adjusting incentives, leading to stable and high-welfare outcomes.

Contribution

It proposes a novel rule for automatically learning incentive schemes based on anticipated parameter updates, enhancing cooperation in complex multi-agent settings.

Findings

High social welfare achieved in matrix games with cooperation

Stable cooperative outcomes in certain games even after intervention stops

Reduced incentives needed over time to maintain cooperation

Abstract

In the future, artificial learning agents are likely to become increasingly widespread in our society. They will interact with both other learning agents and humans in a variety of complex settings including social dilemmas. We consider the problem of how an external agent can promote cooperation between artificial learners by distributing additional rewards and punishments based on observing the learners' actions. We propose a rule for automatically learning how to create right incentives by considering the players' anticipated parameter updates. Using this learning rule leads to cooperation with high social welfare in matrix games in which the agents would otherwise learn to defect with high probability. We show that the resulting cooperative outcome is stable in certain games even if the planning agent is turned off after a given number of episodes, while other games require ongoing intervention to maintain mutual cooperation. However, even in the latter case, the amount of necessary additional incentives decreases over time.