Adaptive Algorithms and Collusion via Coupling

Abstract

This paper studies the strategic interaction of algorithms in economic games. We analyze games where learning algorithms play against each other while searching for the best strategy. We first establish a fluid approximation technique that enables us to characterize the learning outcomes in continuous time. This tool allows to identify the equilibria of games played by Artificial Intelligence algorithms and perform comparative statics analysis. Thus, our results bridge a gap between traditional learning theory and applied models, allowing quantitative analysis of traditionally experimental systems. We describe the outcomes of a social dilemma, and we provide analytical guidance for the design of pricing algorithms in a Bertrand game. We uncover a new phenomenon, the coordination bias, which explains how algorithms may fail to learn dominant strategies.

Emotional Eater Questionnaire--Turkish Version

“…Apart from such vulnerabilities, Q-learning dynamics can also lead to peculiar outcomes in the game settings by leading to tacit collusion that can undermine the competitive nature of the markets [Calvano et al, 2020, Klein, 2021, Hansen et al, 2021, Banchio and Mantegazza, 2022. For example, Banchio and Mantegazza [2022] study the collusive behavior of Q-learners in the widelystudied prisoner's dilemma game (where agents have two actions: 'cooperate' and 'deflect'). They observe that Q-learners can learn to collude in cooperation even though 'cooperate' is always an irrational choice as 'deflect' dominates 'cooperate' strategy.…”