GenAI in Health Research

We are interested in how generative AI can be used to support health research, including evidence synthesis, data extraction, harmonization, and other tasks that are often labor-intensive and difficult to scale. At the same time, these tools raise important questions about validity, reproducibility, and interpretability. Our goal is not simply to use AI, but to understand when it is useful, when it fails, and how it can be integrated into research workflows without lowering scientific standards.

We are also interested in the broader conceptual questions AI raises for science. Rather than relying on reassurance or speculation, we think these systems should be studied empirically: what they can do, where they fail, what kinds of errors they introduce, and how they may change scientific reasoning and practice. For us, the important question is not whether AI is “human,” but how it affects the production, evaluation, and communication of knowledge.

AI is already reshaping how knowledge is produced and evaluated in both academia and public health. In that sense, it is not just a tool but an epistemic challenge: it can alter how evidence is generated, interpreted, communicated, and trusted. If academic researchers or public health institutions respond by ignoring these systems or treating them as beneath serious study, they risk becoming less relevant to the technologies and decision-making environments that are already influencing science, policy, and practice. We believe it is important to engage these methods directly, understand their limitations, learn how to use them well, and monitor their effects on research quality and public trust.