In contemporary democracies, clear information is critical for informed voting, yet citizens face overwhelming digital content, leading to misinformation and decreased accountability (Matějka and Tabellini 2020). To combat this, our research (Ash et al. 2025) explores the potential of AI tools like BallotBot, a chatbot simplifying California ballot initiative information for the November 2024 election. Utilizing retrieval-augmented generation, BallotBot improves accuracy and engagement, outperforming traditional voter guides on complex questions. Our randomized experiment with 1,500 participants revealed significant benefits in information retention and engagement, particularly for those with less prior knowledge. BallotBot reduced perceived effort for learning, enhancing curiosity around propositions. However, its long-term impacts on knowledge retention and voting behavior showed mixed results. These findings suggest that integrating AI with traditional voter resources could bolster democratic engagement and reduce representation inequalities, contingent on users’ trust in AI recommendations. Investing in transparent AI tools can enhance voter education initiatives for better-informed electorates.
Source link
