AI Is Both the Problem and the Promise for Health Communication

The use of artificial intelligence (AI) among colleagues has quickly gone from “Have you heard of Claude?” to trading examples of how we use AI to work, plan, and communicate more clearly. With access to these tools, we would be remiss not to use them to strengthen our work, increase efficiency, and widely disseminate information that can improve health and inform decision-making.

At AcademyHealth, we believe deeply in delivering evidence to the people who can use it. With AI, evidence can now be summarized and disseminated faster than ever before, and the need for widely communicated science is at an all-time high. AI can also be used to spread misinformation and fan the flames of anti-science rhetoric.

AI platforms are emerging and improving rapidly—including some that are specifically designed to assist with elements of health services research such as conducting literature reviews, creating data visualizations, and writing manuscripts. Generative AI can be utilized as an extra member of your Communications team, tailoring research to better suit different audiences (e.g., policymakers, health system leaders, or patients and caregivers) and drafting content for different outlets (e.g., email, newsletters, blogs).

AI can also produce false information, create nonexistent citations that seem credible at first glance, and propagate misinformation on social media. A 2026 systematic review found that misinformation largely focused on vaccines, infectious diseases, and mental health topics spread rapidly from AI’s ability to increase the “volume, speed, and perceived credibility of health disinformation production.” This problem cannot be attributed to just those who create false information—but also those who control the spread.

Policymakers can access many trusted sources when they need information, such as the Congressional Research Service, think tanks, or professional associations. However, their constituents, and, yes, some policymakers themselves, can be influenced by AI-generated information circulating social media. For example, deepfakes, or “a photograph or video of a real person, or a sound recording of their voice, that is altered to make the person appear to do or say something they haven’t done or said” falsely depicted medical professionals from The Baker Heart and Diabetes Institute in Australia promoting a diabetes supplement. The fact that misinformation is able to shape policy, influence political will, and affect health care access and delivery is incredibly frustrating, especially for those trying to get good evidence into practice. Instead of using time and resources to further improve the public’s health and wellbeing, researchers must now advocate for well-established science.

The same properties that make AI a powerful spreader of misinformation—speed, scale, fluency, and accessibility— also make it a promising tool to support good research communication. For example, local health departments are using generative AI to draft plain language summaries, translate outreach messages into multiple languages, and create social media posts to communicate about heat events and respiratory illness upticks. For a field that has long struggled to move evidence from journals into practice in a timely fashion, it is a tool worth exploring.

Researchers and health policy organizations have an opportunity to use AI tools strategically: synthesizing findings for non-technical audiences, translating dense evidence into formats for policymakers who are short on time, and reaching viewers (or readers) that traditional dissemination strategies never did. The question isn’t whether AI will shape how health information spreads — it already does. The question is whether the research community will help shape how it’s used— and those decisions are quickly being made.

To contribute to the uptake, we must think about communication strategy as part of a research plan from the beginning, not as an afterthought. Key steps, such as 1) defining clear communication goals; 2) identifying key audiences; and 3) developing messaging strategies that reach decision-makers and inspire action, can help researchers’ interactions accomplish their communication goals.

The HSR field must also build AI literacy — understand how these tools work, where the gaps are, and how to evaluate AI-generated health content critically. Organizations working at the intersection of research and policy have a particular role to play here, both in modeling responsible AI use and in equipping members and partners to do the same. This all hinges on the continued production of rigorous evidence. Given our current environment that moves faster than peer review, the field’s ability to communicate clearly, quickly, and through the right channels may matter as much as the research quality itself. To best adapt, researchers must improve AI literacy, practice dissemination methods using AI, and consume health information with a critical eye.

Producing good evidence remains the foundation but ensuring that science reaches the right people, through the right channels, at the right moment is a key tenet of HSR, too. As a field, we must figure out how to effectively communicate good science to inform decisions that will impact public health and health care delivery for years to come.