Conversational AI systems represent both opportunity and vulnerability for democratic discourse as they are frequently embedded in interactive systems, such as instant messenger or games. With millions engaging in AI-mediated conversations daily, these interactions shape moral reasoning and political deliberation in unprecedented ways. Research demonstrates that AI advice profoundly influences users' ethical thinking. While AI conversations offer potential for inclusive democratic participation—enabling personalized deliberation, countering misinformation, and moderating polarized discussions—they simultaneously introduce asymmetrical power relations through opaque data extraction and algorithmic curation. Recent evidence reveals significant shifts in AI systems' normative orientations, e.g. exhibiting stronger utilitarian biases. This raises critical questions about consent, transparency, and legitimacy in AI-mediated democratic processes. Based on recent empirical studies by the authors examining AI-supported conversations and different conversational styles, we discuss how conversational AI can be employed to establish democratically legitimized frameworks for governing these influential technologies while maintaining pluralistic online discourse.

Rieger, D., Kobilke, L., Haim, M., & Reinemann, C. (6/2026). AI and Democratic Processes Online. Presented at the 76th Annual Conference of the International Communication Association, Cape Town. (content_copy)