When Your AI Recommends a Doctor You Can't Afford

In this episode of AI Ethics with Fexingo, Lucas and Luna explore the hidden bias in AI-powered healthcare referral systems. They focus on a 2025 study by the University of California, San Francisco, which found that three major hospital networks' AI referral algorithms recommended expensive specialists 40% more often to patients with private insurance than to those with Medicaid, even when clinical need was identical. The hosts discuss how training data from historical referral patterns encoded financial incentives, the lack of transparency in vendor algorithms, and why fixing this isn't just about adding a fairness constraint — it requires rethinking what data goes into the model. Lucas shares a concrete case: a woman in Phoenix whose AI triage system suggested a $2,000 imaging study for a condition that later turned out to be a simple vitamin deficiency. They also consider the regulatory landscape, including a recent CMS proposal to audit referral algorithms. The episode ends with a reflective question about whether efficiency metrics are masking systemic inequity. #AIEthics #HealthcareAI #AlgorithmicBias #Medicaid #InsuranceBias #ReferralAlgorithms #UCSF #ClinicalDecisionSupport #HealthEquity #AIAccountability #MedicalAI #BiasInHealthcare #FexingoBusiness #BusinessPodcast #Technology #Podcast #DataEthics #Regulation Keep every episode free: buymeacoffee.com/fexingo