A recent analysis by the Blue Cross Blue Shield Association (BCBSA) reveals that the use of artificial intelligence (AI) tools by hospitals during insurance claim submissions contributed to an additional $942 million in healthcare spending over a span of two years. The study highlights a significant rise in patients being categorized with complex conditions, yet it points out a disconnect between the medical coding and the actual treatment delivered, indicating no substantial change in patient care despite increased costs.

This rise in expenses is drawing attention as another indication that AI may be driving up healthcare costs. Coverage from the New York Times notes that conflicts between insurers and hospitals over payments and treatments have long existed, but the introduction of AI on both sides appears to exacerbate these disputes. The deployment of AI technologies in administrative and billing processes may be contributing to higher claims and denials, intensifying tensions within the healthcare payment ecosystem.

Dr. Shiv Rao, founder of the AI health startup Abridge, acknowledges the potential risks of AI sparking a future fraught with adversarial automated systems — “bots fighting bots” and agents clashing — which could negatively impact healthcare cost dynamics. However, he also sees a possibility for AI to streamline processes, reduce conflict, and lower expenses if managed correctly. Meanwhile, BCBSA’s senior vice president Luke Chalker describes the current situation as heavily skewed against insurers, characterizing it as a “one-sided blood bath” with insurance companies bearing the brunt of rising costs.

The growing integration of AI into healthcare administration exposes challenges not only in cost management but also in aligning technology with actual clinical benefits. As AI tools become more prevalent in hospitals and insurance firms, ongoing scrutiny is needed to balance innovation with cost containment and ensure that patient care improvement remains the core focus amid changing billing and coding practices.