- Healthcare costs are climbing faster in sectors where AI adoption is highest, contradicting years of promises that automation would make medicine cheaper.
- Hospitals and insurers are spending billions on AI infrastructure — and passing those costs to patients before any efficiency gains materialize.
- A growing body of evidence suggests AI is adding layers of complexity to healthcare billing and diagnostics rather than simplifying them.
The healthcare industry sold AI to patients and policymakers with a simple pitch: automation would cut costs, reduce errors, and make care more accessible. The reality, according to a growing body of evidence, is moving in the opposite direction. Healthcare costs are surging in the sectors where AI adoption is highest, and the bill is landing squarely on patients.
The problem isn’t that AI doesn’t work — in many clinical applications, it performs well. The problem is that implementing AI in healthcare is expensive in ways that nobody accounted for during the sales pitch. Hospitals are spending billions on AI infrastructure: specialized hardware, data pipelines, integration with legacy systems, training for staff, compliance frameworks, and ongoing model maintenance. Those costs show up on balance sheets long before any efficiency gains do.
Insurers are doing the same thing. AI-driven claims processing, fraud detection, and prior authorization systems require massive upfront investment. The vendors selling these systems charge premium prices because they can — healthcare is a captive market where switching costs are astronomical and “AI-powered” is the magic phrase that justifies any line item.
Why AI Is Adding Costs to Healthcare Instead of Cutting Them
There’s a secondary effect that’s harder to measure but potentially more significant. AI is creating new categories of healthcare spending that didn’t exist before. AI-assisted diagnostics generate more tests, not fewer — when a model flags something as potentially abnormal, doctors order follow-ups to confirm. The false positive rate of AI screening tools in radiology, dermatology, and pathology means more imaging, more biopsies, more lab work. Each additional test is revenue for someone and cost for the patient.
Administrative AI is adding complexity too. Automated prior authorization systems, designed to reduce paperwork, have created a new layer of algorithmic gatekeeping that patients and providers spend hours navigating. When an AI denies a claim, appealing it requires human intervention that’s more expensive than the manual process it replaced.
The pattern mirrors what happened with electronic health records in the 2010s. EHRs were supposed to save time and money. Instead, they created a documentation burden that doctors estimate consumes two hours of every workday, contributed to physician burnout, and spawned a multi-billion-dollar industry of consultants and compliance officers. AI is following the same trajectory — with higher stakes.
None of this means AI can’t eventually reduce healthcare costs. But “eventually” is doing a lot of work in that sentence. The transition period, which could last a decade or more, is characterized by simultaneous spending on AI systems and the human infrastructure required to operate them. Patients are paying for both.