A harm-reduction framework for responsible AI in public health research

Abstract

Abstract Artificial intelligence is reshaping population medicine and public health, yet the same systems that generate insight can amplify misleading content and blur where evidence ends and AI fabrication begins. This perspective offers a four-level harm-reduction framework, comprising responsible population selection, data governance, public engagement, and transparent dissemination, illustrated through six case strategies. The aim is operationalizing responsible AI in population health research without slowing the innovation the field needs.