AI
AI agent learns to price market externalities, boosting resource sustainability
Deep reinforcement learning offers a way to fine-tune market prices that account for environmental and social costs, outperforming traditional equilibrium in scarce-resource scenarios.
Most competitive markets ignore the hidden costs that trading partners impose on one another—such as when overusing a shared resource depletes supplies for everyone else. A new approach using artificial intelligence sidesteps this longstanding problem by letting an AI agent act as a policymaker to dynamically adjust market prices.
The core challenge#
Market failures stemming from negative externalities—costs borne by third parties rather than the transacting parties themselves—have been notoriously hard to correct. Setting the right price adjustments requires balancing competing priorities: sustainability, equity, producer and consumer welfare, and resource efficiency.
How the AI solution works#
Researchers developed a deep reinforcement learning policymaker that operates alongside other learning agents in a simulated market. Rather than relying on static formulas or complex calculations, the AI agent learns to tune prices by trial and error, optimizing toward multiple goals simultaneously. This flexibility allows the policymaker to prioritize different objectives—whether environmental protection, fairness, or economic efficiency—depending on the scenario at hand.
Key findings#
In resource-constrained environments, the AI-driven approach significantly outperformed traditional market equilibrium at maintaining resource sustainability. The policymaker successfully kept resource depletion under control while still allowing productive trade and allocation of goods.
The research was originally submitted to arXiv in June 2021 under the Multiagent Systems category in computer science and has undergone several iterations since.
