The AI Economist: Taxation policy design via two-level deep multiagent reinforcement learning

• DOI: 10.1126/sciadv.abk2607
• Publication Date: 2022-05-04T17:57:55Z
• Authors: Stephan Zheng; Alexander Trott; Sunil Srinivasa; David C. Parkes; Richard Socher
• Keywords: Social planner; Social Learning; Planner; Mechanism Design

Artificial intelligence (AI) and reinforcement learning (RL) have improved many areas but are not yet widely adopted in economic policy design, mechanism or economics at large. The AI Economist is a two-level, deep RL framework for design which agents social planner coadapt. In particular, the uses structured curriculum to stabilize challenging coadaptive problem. We validate this domain of taxation. one-step economies, recovers optimal tax theory. spatiotemporal substantially improves both utilitarian welfare trade-off between equality productivity over baselines. It does so despite emergent tax-gaming strategies while accounting labor specialization, agent interactions, behavioral change. These results demonstrate that complements theory unlocks an AI-based approach designing understanding policy.