AI Insight
This commentary introduces the concept of the "recursive care law," an evolution of Hart's 1971 inverse care law, arguing that AI implementation in healthcare is unevenly distributed and risks amplifying existing health inequities. A 2023-24 analysis of 3,560 US hospitals found that AI adoption was geographically clustered, with hospitals in higher-need regions being less likely to use AI systems. This pattern suggests that AI, rather than improving equitable access to care, may create self-reinforcing feedback loops that deepen disparities over time.
Why it matters
Unequal AI adoption in healthcare could systematically disadvantage already underserved populations, making it critical for policymakers and health systems to ensure equitable distribution of AI tools as part of broader health equity strategies.
In The Lancet in 1971, Julian Tudor Hart gave medicine one of its most important insights: the availability of medical care tends to vary inversely with need—the inverse care law.1 Half a century later, artificial intelligence (AI) risks turning this observation into something more dynamic and dangerous. Evidence suggests that AI implementation is already unevenly distributed. In a 2023–24 analysis of 3560 US hospitals, implementation of AI models was geographically clustered, and regions with greater health-care need were less likely to have hospitals using such systems.