Human and Algorithmic Cooperation: Parameters Matter!
Title: Human and Algorithmic Cooperation: Parameters Matter! (w/ Steffen Eibelshäuser, Hendrik Hegemann, Victor Klockmann and Alicia von Schenk)
Abstract: Under which conditions do humans cooperate with machines? We address this question within the context of the infinitely repeated discounted Prisoner’s Dilemma. Recent theory has identified a critical parameter threshold where human behavior changes substantially from lower to higher cooperation rates. We analyze in controlled lab experiments how replacing a human with a sophisticated learning algorithm affects cooperation rates. We find that the theoretical threshold becomes sharper and more predictive once humans are replaced by profit-maximizing AI algorithms as the gap between cooperation rates above and below the threshold gets even more pronounced.