This paper argues that artificial intelligence now functions as a determinant of health, proposing a novel epidemiological framework to measure and study its population-level effects. It differentiates between ambient and personal AI exposure and discusses the implications for study design, health equity, and AI governance.