Towards Machines that Trust: AI Agents Learn to Trust in the Trust Game

TL;DR

This paper presents a theoretical and simulation-based analysis of how AI agents can learn to trust in the trust game, providing a mathematical foundation for trust emergence.

Contribution

It introduces a theoretical framework and simulation results demonstrating how reinforcement learning enables AI agents to develop trust in the trust game.

Findings

Theoretical analysis supports trust emergence in AI agents.

Simulation results validate the mathematical framework.

Provides insights into trust development in AI systems.

Abstract

Widely considered a cornerstone of human morality, trust shapes many aspects of human social interactions. In this work, we present a theoretical analysis of the $\textit{trust game}$, the canonical task for studying trust in behavioral and brain sciences, along with simulation results supporting our analysis. Specifically, leveraging reinforcement learning (RL) to train our AI agents, we systematically investigate learning trust under various parameterizations of this task. Our theoretical analysis, corroborated by the simulations results presented, provides a mathematical basis for the emergence of trust in the trust game.