Governing the attention dividend: AI-reclaimed clinician capacity as a health policy resource
TL;DR
The article specifies the payment and care model conditions that make deliberate reallocation more feasible, and proposes 3 priority uses—recognition of overlooked patients, continuity, and safety-netting and reassurance; and pairs 8 allocation questions with measurable indicators, concrete policy and operational mechanisms, and accountable actors.
Abstract
Abstract Ambient artificial intelligence (AI) is reducing documentation burden in primary care, with real but modest and variable effects: reclaimed minutes in some settings, reduced cognitive load in others, and no guarantee that either reaches patients. This article argues that the resulting capacity—the attention dividend—should be governed as a health policy resource. Without deliberate allocation, it defaults to throughput, administrative absorption, and already-advantaged patients. The article specifies the payment and care model conditions that make deliberate reallocation more feasible, including hybrid and value-based payment, continuity add-on payments, monthly per-patient care-management payments, and primary care spending floors; proposes 3 priority uses—recognition of overlooked patients, continuity, and safety-netting and reassurance; and pairs 8 allocation questions with measurable indicators, concrete policy and operational mechanisms, and accountable actors across health system leaders, payers, purchasers, primary care practices, AI vendors, regulators, and accreditors. Seven named capture mechanisms describe how the dividend fails to reach patients, each with a corrective governance response. The policy question is not whether ambient AI saves time, but whether health systems govern the capacity it returns—through mechanisms that can be named, measured, and assigned.