Kara Luo

Online communities have enabled major collective achievements by allowing many individuals to contribute with little coordination or centralized control. Increasingly, autonomous AI agents are being deployed alongside human contributors to support these communities. Prior research suggests that such agents can improve efficiency and expand the scope of tasks that online communities can undertake. However, there is limited longitudinal evidence on whether these benefits extend to the highest-quality outcomes, and whether they can be realized without costly changes in how community members coordinate their work. Here we analyze over two decades of activity across 763 Wikipedia subcommunities, where large numbers of autonomous agents (“bots”) have performed critical tasks since early in the platform’s history. Greater bot activity is associated with more articles advancing from the lowest quality tiers each month, but shows little to no association with advancement to the highest quality tier. These divergent patterns are shaped by human coordination: at all stages, articles improve most when bot activity coincides with article-level discussions, and advancement to the highest tiers is especially likely when bot activity aligns with both article-level and community-level discussions. Overall, these findings highlight both the promise and the limits of autonomous agents: while AI can augment collective production, achieving the best outcomes may require greater coordination—and potentially less decentralization—than has historically characterized online communities. Because Wikipedia’s bots represent an early generation of AI, our results provide a baseline for evaluating how more capable agents may reshape online communities in the future.

Heated exchanges over controversial topics can harm relationships while failing to change minds. Across five pre-registered experiments (N = 3,963), we tested a brief intervention to mitigate the interpersonal costs of disagreement and increase the likelihood of future conversations around vaccine hesitancy. Vaccine-supportive participants randomly assigned to training in conversational receptiveness were seen as more reasonable and more trustworthy than those writing in their natural tone, and evaluated their (untrained) counterpart more favorably. They were as persuasive as those who were incentivized to be persuasive only, and their counterparts were more interested in learning their views on other topics. Finally, receiving training in conversational receptiveness and learning that one’s counterpart was similarly trained increased participants’ willingness to discuss vaccines by 50%. We discuss the implications of these findings in the context of the many goals individuals pursue in contentious conversations and the outcomes conflict scholars should consider in future research.

Scholars warn that partisan divisions in the mass public threaten the health of American democracy. We conducted a megastudy (n = 32,059 participants) testing 25 treatments designed by academics and practitioners to reduce Americans’ partisan animosity and antidemocratic attitudes. We find that many treatments reduced partisan animosity, most strongly by highlighting relatable sympathetic individuals with different political beliefs or by emphasizing common identities shared by rival partisans. We also identify several treatments that reduced support for undemocratic practices—most strongly by correcting misperceptions of rival partisans’ views or highlighting the threat of democratic collapse—which shows that antidemocratic attitudes are not intractable. Taken together, the study’s findings identify promising general strategies for reducing partisan division and improving democratic attitudes, shedding theoretical light on challenges facing American democracy.

As individuals and political leaders increasingly interact in online social networks, it is important to understand the dynamics of emotion perception online. Here, we propose that social media users overperceive levels of moral outrage felt by individuals and groups, inflating beliefs about intergroup hostility. Using a Twitter field survey, we measured authors’ moral outrage in real time and compared authors’ reports to observers’ judgements of the authors’ moral outrage. We find that observers systematically overperceive moral outrage in authors, inferring more intense moral outrage experiences from messages than the authors of those messages actually reported. This effect was stronger in participants who spent more time on social media to learn about politics. Preregistered confirmatory behavioural experiments found that overperception of individuals’ moral outrage causes overperception of collective moral outrage and inflates beliefs about hostile communication norms, group affective polarization and ideological extremity. Together, these results highlight how individual-level overperceptions of online moral outrage produce collective overperceptions that have the potential to warp our social knowledge of moral and political attitudes.