Keywords: deliberative poling, meta, trust, generative ai

The impact of participation in deliberative mini-publics on trust—both social trust in others and institutional trust in the sponsoring government or organization—is not yet fully understood. While participation is often theorized to foster trust through fair and transparent processes, empirical findings reveal mixed results, influenced by context, participant satisfaction, and prior attitudes. This study extends existing research on deliberative democracy and trust to a new context: an online deliberative mini-public sponsored by an AI company. Drawing on data from Meta’s recent 1,500-person Community Forum on generative AI chatbots, we examine the causal impact of online deliberation on these two dimensions of trust using a difference-in-differences methodology. In addition, we analyze the role of sociodemographic predictors in shaping trust outcomes. We contextualize our findings within existing political theory and empirical evidence on democratic innovation and trust, highlighting those theories our data support and those they challenge.