Humans shape the behavior of artificially intelligent algorithms. One mechanism is the training these systems receive through the passive observation of human behavior and the data we constantly generate. In a laboratory experiment with a sequence of dictator games, we let participants' choices train an algorithm. Thereby, they create an externality on future decision making of an intelligent system that affects future participants. We test how information on training artificial intelligence affects the prosociality and selfishness of human behavior. We find that making individuals aware of the consequences of their training on the well-being of future generations changes behavior, but only when individuals bear the risk of being harmed themselves by future algorithmic choices. Only in that case, the externality of artificially intelligence training induces a significantly higher share of egalitarian decisions in the present.

No abstract

If individuals tend to behave like their peers, is it because of conformity, that is, the preference of people to align behavior with the behavior of their peers; homophily, that is, the tendency of people to bond with similar others; or both? We address this question in the context of an ethical dilemma. Using a peer effect model allowing for homophily, we designed a real-effort laboratory experiment in which individuals could misreport their performance to earn more. Our results reveal a preference for conformity and for homophily in the selection of peers, but only among participants who were cheating in isolation. The size of peer effects is similar when identical peers were randomly assigned and when they were selected by individuals. We thus jointly reject the presence of a self-selection bias in the peer effect estimates and of a link strength effect..