Keywords: Deliberation, Generative AI, Deliberative Polling, Meta

Previous research has shown that Deliberative Polling can increase optimism toward artificial intelligence, yet the extent to which these shifts in perception vary across demographic groups remains unclear. Building on a recent multi-country deliberative poll conducted by Stanford's Deliberative Democracy Lab in collaboration with Meta and the Behavioral Insights Team, which engaged participants from Brazil, Germany, Spain, and the United States, this paper analyzes demographic subsamples—such as age, gender, education, and income strata—within the collected datasets. The Deliberative Poll explored public attitudes toward AI chatbots, examining issues such as personalization, information sources, and the human-like design of AI. Employing robust statistical methods, this paper disaggregates and compares attitudes both before and after deliberative interventions to reveal whether certain subpopulations exhibit stronger optimism gains or express more nuanced concerns than others. Ultimately, this paper contributes to a richer understanding of how deliberation shapes public perspectives on AI, informing both policy design and the effective communication of emerging technologies.