Leveraging Large Language Models for Personalized Public Messaging

Abstract: We present a novel methodology for crafting effective public messages by combining large language models (LLMs) and conjoint analysis. Our approach personalizes messages for diverse personas – context-specific archetypes representing distinct attitudes and behaviors – while reducing the costs and time associated with traditional surveys. We tested this method in public health contexts (e.g., COVID-19 mandates) and civic engagement initiatives (e.g., voting). A total of 153 distinct messages were generated, each composed of components with varying levels, and evaluated across five personas tailored to each context. Conjoint analysis identified the most effective message components for each persona, validated through a study with 2,040 human participants. This research highlights LLMs’ potential to enhance public communication, providing a scalable, cost-effective alternative to surveys, and offers new directions for HCI, particularly for the design of adaptive, user-centered, persona-driven interfaces and systems.

Teaser: This small mutiples plot shows the strong correlation between pairs of personas in the COVID-19 messaging study, showing both LLM results (top) and human results (bottom). Each panel shows how one persona relates to all others.

To create this figure we conducted a Spearman’s correlation for feature level importance values across all persona pairs in the COVID-19 health messaging study. The figure shows LLM-simulated personas (top) and human participants (bottom). Each panel focuses on one of the five designed personas - one row in the correlation matrix -, comparing its response similarity to the other four personas. The panels consistently reveal that response similarity decreases with stance similarity. For the two extreme personas, a clear unilateral drop-off is observed, while for the others there is a bilateral drop-off. Notably, human participants exhibit very similar patterns and trends to LLMs. These findings suggest the strong potential of our method of using LLMs to capture meaningful persona-driven variations.

Video: Watch it to get a quick overview how a crowd participant would build a small causal network with our interface:

Paper: A. Das, C. Xiong Bearfield, K. Mueller, “Leveraging Large Language Models for Personalized Public Messaging,” ACM CHI Conference on Human Factors in Computing Systems Late Breaking Work, pp. 1-7, Tokyo, Japan, May 2025. PDF