Summary
Multi-agent reinforcement learning involves training several AI agents to work together in the same environment, but the team's overall success doesn’t guarantee that rewards are shared fairly among the agents. The authors developed FairTest, a testing approach that searches for unfair outcomes by evaluating runs using multiple fairness measures and predicting where fairness problems might occur. FairTest combines these fairness scores with uncertainty in the decision-making of agents to focus testing on scenarios likely to reveal fairness issues. When tested, FairTest found more fairness failures than other methods, showing it can better expose unfairness in these AI systems.
What this means in practice
- •For ai development teams: Identify unfair reward distributions in multi-agent AI systems before deployment to improve equity and trustworthiness.
- •For robotics engineers: Test multi-robot coordination policies for fairness issues, ensuring balanced resource or reward handling among robots.
Abstract
Multi-agent Reinforcement Learning (MARL) trains a team of agents that share one environment and learn their policies together. Training maximizes the team return, and a high return does not imply that the rewards are shared fairly among the agents in every episode. Testing is an established way to discover the failures of deep reinforcement learning, yet few methods address the fairness of MARL. In this work, we propose FairTest, a search-based testing approach that seeks the unfair executions of a MARL policy. The design combines search guidance with test prioritization. The guidance scores each candidate with three fitness functions. One measures the fairness of the runs already performed, another predicts the fairness from abstract states and fairness features, and the third reads the decision uncertainty from the policy. Crossover and mutation derive further candidates from the observed executions. The prioritization ranks the candidates by the predicted fairness and the decision uncertainty, so that the runs reach the candidates where failures are expected. FairTest is evaluated on three environments and two MARL algorithms, and four baselines are given the same budget. It detects the most fairness failures compared to three baselines with statistical significance and large effect sizes. The failure count exceeds that of the strongest baseline by 221% on average and coverage improves by an average of 23%.