Knowledge that Transforms

The rapid evolution and extensive application of artificial intelligence (AI) have generated a distinct form of human-AI interaction in multiplayer online games. Using a policy in a popular online game that incorporates AI players as opponents to human players, we investigate how this policy influences player engagement, friend team-ups, and the broader game ecosystem. Our findings indicate that introducing AI players significantly increases game engagement and encourages friend team-ups. To uncover the underlying mechanism, we perform a mediation analysis to examine the roles of self-efficacy and team responsibility. AI players are intentionally designed to be less skilled while being difficult to distinguish from human players. Our mediation analysis suggests that this design can allow successful outcomes to be interpreted as signals of players’ own competence and team contribution. These interpretations imply stronger perceived self-efficacy and team responsibility, which can lead to greater engagement and more team-ups with friends. Finally, our analysis indicates that the impacts of AI players are moderated by player skill levels and teammate preferences. In particular, the positive effects of AI players on both player engagement and friend team-ups are stronger among novices. Players who initially prefer random teammates experience greater social benefits, as they become more likely to team up with friends after the introduction of AI players. Our research offers valuable implications for game developers who seek to enhance player engagement and foster social interaction in the rapidly evolving landscape of multiplayer online games.

User-generated content, for example, on Wikipedia, is easily accessed but has uncertain reliability. This makes it attractive to use but also creates risk, so there should be limits to who uses Wikipedia and for what purposes. In this paper, we use a randomized control trial to show that Wikipedia’s influence extends to judicial decision making, a field that is highly professional and supposed to follow strict procedures. This causal evidence further emphasizes the widespread influence of Wikipedia and other frequently accessed user-generated content on important social outcomes. Our findings also reveal boundaries to user-generated content’s influence. Although Wikipedia’s influence does extend to courts of “first instance” (where the case is first decided), it does not extend to higher courts (Court of Appeals, Supreme Court). These results suggest that normative prohibitions do seem to be sufficient to keep Wikipedia from influencing the most-important, well-resourced parts of law but that these prohibitions are insufficient in areas where time and resource pressures are greater. By showing that Wikipedia is influencing such an important and formal domain, our paper reinforces the importance of improving the accuracy and reliability of user-generated content, especially in domains with far-reaching societal consequences. Because there is no obvious way to prevent individuals from taking advantage of user-generated content professionally or nonprofessionally, our findings also contribute to the ongoing discussion of how to build public repositories of knowledge into more reliable storehouses.