People increasingly turn to AI for health information, but little is known about how that shapes their trust and understanding.

Generative AI is rapidly changing how people access health information. NORC data released in late 2025 found that more than a quarter of U.S. adults had used AI tools to seek health information, and about 4 in 10 trust AI-generated content as much as or more than traditional media sources. Yet despite this shift, significant gaps remain in understanding how AI-mediated delivery affects the way people interpret, trust, and act on what they learn.

For public health organizations, health systems, and funders, this creates an urgent challenge: how do you develop effective health communication strategies when the medium itself, AI, is reshaping the message? Answering that question requires not just better data, but new research tools designed to study human-AI interaction in health contexts.

NORC built a research-ready AI chatbot that creates controlled environments for studying health communication.

THIA, NORC’s Trusted Health Information Assistant, is a generative AI chatbot designed specifically for health communication research. Unlike general-purpose AI tools, THIA draws its responses from a curated knowledge base of vetted health sources using retrieval-augmented generation (RAG), giving researchers control over the information participants receive and the ability to trace every response back to its source.

THIA’s modular architecture allows research teams to customize the platform for specific health topics, populations, and study designs. Its first deployment focused on caregiver well-being for older adults, but the system can be adapted to address topics ranging from immunization confidence to mental health information-seeking to digital health literacy. THIA can be embedded within surveys, interviews, or experimental designs, enabling researchers to test how variations in message framing, tone, and interaction design influence comprehension, engagement, and trust.

NORC built THIA on its own secure infrastructure with responsible AI safeguards, including transparency features that display source citations alongside chatbot responses. A back-end analytics dashboard captures engagement patterns, conversational dynamics, and sentiment data, supporting both quantitative and mixed-methods analysis.