Public health promotion campaigns are effective, but do not tend to be efficient. Most are time-consuming, expensive, and rely on the intuition of creative workers who design their messages without clearly feeling the change in behavior. New research conducted by researchers from the University of Pennsylvania, government and community institutions Dolores Albarak Ah and Man Pui Sally Chan, University of Illinois and Emory University suggests that artificial intelligence (AI) can promote the selection of message based on theory and evidence.
The research group led by social psychologist Amy Gutman Penn, is a professor at the University of Knowledge and director of the Communication Sciences at the Annenberg Public Policy Center, which has developed a series of computational processes, automatically generating HIV prevention and testing campaigns in the United States, using real-time media as a source of message. The paper, Chan, an associate professor of research at Annenberg Communications School in Penn, explains how people and institutions provide a living repository of messages that can be chosen based on team theory, as well as a living repository of AI-generated data about messages circulating on social media.
Social media provides a living repository of messages generated by the community, from which effective messages can be drawn and amplified. The researchers collected HIV prevention and test messages from US social media posts, curated “actionality” (an important characteristic of messages intended to motivate actions), and designed them to select posts that are suitable for targeted prioritization groups, in this case male (MSM) (MSM).
The researchers then conducted three studies. The first computational analysis established that AI tools successfully select messages with the desired quality. Second, online experiments with men having sex with men showed that the resulting messages were perceived as more practical, personally relevant and effective by the target audience than control messages not selected by AI tools. Third, field experiments involving competent public health agencies and community-based organizations in 42 US counties showed that utilizing the AI message selection process significantly increases the likelihood that public health agencies will post HIV prevention messages on social media.
As part of the study, researchers also tested messages examined by human researchers after being selected for messages that were not reviewed by the AI process. The AI selected messages outperformed reported validity control messages, whether reviewed or not, but the reviewed messages performed better than the unnetworked ones. Regardless of the benefits of review in terms of effectiveness, researchers warn that a short human review process should be included as part of this method to avoid harmful content and false information.
The study, recently published in PNAS Nexus, provides the first empirical evidence for successful automatic selection of public health messages for community and government dissemination. Chan says this is a promising development. “This AI process can provide cheap and creative ways for public health agencies to spread effective messages.” Albarracín agrees that “the age of AI accelerates the ability to use theory and empirical evidence in the generation of rapid and continuous campaigns.”
“Health Promotion Campaign for the US Community: AI-based Decentralized Content Extraction and Sharing” was published on PNAS Nexus in June 2025. See this paper for a complete list of authors and affiliations. doi: 10.1093/pnasnexus/pgaf171.
/Public release. This material of the Organization of Origin/Author is a point-in-time nature and may be edited for clarity, style and length. Mirage.news does not take any institutional position or aspect, and all views, positions and conclusions expressed here are the views of the authors alone.
