Maintaining cooperation in complex social dilemmas using deep reinforcement learning
This addresses the challenge of designing AI agents that balance selfish interests with collective welfare in complex interactions, which is incremental as it modifies existing reinforcement learning methods.
The paper tackled the problem of maintaining cooperation in social dilemmas by developing deep reinforcement learning agents that are nice, provokable, and forgiving, showing both theoretically and experimentally that these agents can sustain cooperation in Markov social dilemmas.
Social dilemmas are situations where individuals face a temptation to increase their payoffs at a cost to total welfare. Building artificially intelligent agents that achieve good outcomes in these situations is important because many real world interactions include a tension between selfish interests and the welfare of others. We show how to modify modern reinforcement learning methods to construct agents that act in ways that are simple to understand, nice (begin by cooperating), provokable (try to avoid being exploited), and forgiving (try to return to mutual cooperation). We show both theoretically and experimentally that such agents can maintain cooperation in Markov social dilemmas. Our construction does not require training methods beyond a modification of self-play, thus if an environment is such that good strategies can be constructed in the zero-sum case (eg. Atari) then we can construct agents that solve social dilemmas in this environment.