Risk preferences of learning algorithms

• 成果类型: Article
• 署名作者: Haupt, Andreas; Narayanan, Aroon
• 署名单位: Massachusetts Institute of Technology (MIT)
• 刊物名称: GAMES AND ECONOMIC BEHAVIOR
• ISSN/ISSBN: 0899-8256
• DOI: 10.1016/j.geb.2024.09.013
• 发表日期: 2024
• 页码: 415-426
• 关键词: Online learning; Behavior attribution; fairness

摘要

Many economic decision-makers today rely on learning algorithms for important decisions. This paper shows that a widely used learning algorithm-epsilon-Greedy-exhibits emergent risk aversion, favoring actions with lower payoff variance. When presented with actions of the same expectated payoff, under a wide range of conditions, epsilon-Greedy chooses the lower-variance action with probability approaching one. This emergent preference can have wide-ranging consequences, from inequity to homogenization, and holds transiently even when the higher-variance action has a strictly higher expected payoff. We discuss two methods to restore risk neutrality. The first method reweights data as a function of how likely an action is chosen. The second method employs optimistic payoff estimates for actions that have not been taken often.