Human-centered mechanism design with Democratic AI

TL;DR

Democratic AI uses reinforcement learning with human input to design social mechanisms that align with human preferences, demonstrating improved fairness and collective benefit in an online investment game.

Contribution

This paper introduces Democratic AI, a human-in-the-loop reinforcement learning pipeline for designing social mechanisms aligned with human values.

Findings

AI-designed mechanism redressed wealth imbalance

AI mechanism sanctioned free riders effectively

AI mechanism won majority vote in experiments

Abstract

Building artificial intelligence (AI) that aligns with human values is an unsolved problem. Here, we developed a human-in-the-loop research pipeline called Democratic AI, in which reinforcement learning is used to design a social mechanism that humans prefer by majority. A large group of humans played an online investment game that involved deciding whether to keep a monetary endowment or to share it with others for collective benefit. Shared revenue was returned to players under two different redistribution mechanisms, one designed by the AI and the other by humans. The AI discovered a mechanism that redressed initial wealth imbalance, sanctioned free riders, and successfully won the majority vote. By optimizing for human preferences, Democratic AI may be a promising method for value-aligned policy innovation.