Assessing Persuasion via Non‑Conversational Channels (Social Media Flooding)

Investigate and quantify the persuasive effects of AI-generated political content disseminated through non-conversational channels, such as flooding social media at scale, and determine whether such deployments produce attitude shifts comparable to or larger than those observed in interactive chatbot conversations.

Background

The studies in the paper focus on interactive, text-based conversations between participants and LLM chatbots. However, the authors warn that persuasion can also occur through other channels, notably the mass dissemination of AI-generated content on social media.

They explicitly note this extension remains open and highlight that such non-conversational deployments could produce similar effects at potentially much larger scales, underscoring the need to measure and compare persuasion across communication modalities.