Arclight Insights

AI in healthcare does not fail because of capability alone. It fails when systems are optimized around the wrong institutional incentives before affected stakeholders have helped define what the system should actually do. Using Medicare's WISeR model and private-payer denial systems as case studies, this paper applies objective-function governance to public-sector AI: define the objective, constraints, edge cases, evidence standards, and accountability structures before procurement and deployment.