Despite the rapidly increasing cost of healthcare, productivity in the industry remains stagnant. In nearly every other sector of the economy automation drives efficiency and lowers costs. In healthcare, regulators are creating roadblocks by restricting AI tools that improve diagnostic accuracy and treatment efficiency. Instead of preemptively limiting these technologies, AI should be allowed to compete with human providers and legacy systems.
The United States faces a growing physician shortage. Making the existing workforce more productive is essential. That means allowing autonomous AI tools in clearly defined use cases when performance exceeds human standards, and robust real-world validation data support deployment.
AI technologies can already outperform humans in many healthcare tasks. In diagnostics they are often both faster and more accurate than physicians. They can predict medical complications, personalize drug dosing, conduct routine but time-consuming follow-up calls, and automate the collection and organization of patient records. These are exactly the types of tasks that should be automated in a system struggling with rising costs and workforce shortages.
Yet most states have enacted healthcare AI regulations that restrict its use. While caution is understandable with emerging technologies, many of these rules prevent AI from automating processes where it has demonstrated superior performance. Instead of streamlining care, regulators are adding redundant steps that require physicians to double-check work that could be safely automated. AI is not ready to handle every medical decision, but where it clearly exceeds human benchmarks, it should be given greater latitude.
Outdated clinical restrictions also preserve healthcare scarcity. Requiring mandatory human review for every AI-generated result ensures specialists remain bottlenecks, driving up costs and limiting access. This problem is especially acute in rural areas, where specialists are often unavailable. These are precisely the settings where validated autonomous AI could expand access to high quality diagnostic services.
States governments are not alone in limiting AI’s healthcare potential. The Food and Drug Administration (FDA) regulates AI diagnostic and treatment tools as medical devices. While oversight is appropriate, FDA approval timelines are often far slower than software development cycles. AI systems improve incrementally and continuously. Requiring lengthy re-approvals for updates delays deployment and can prevent improved versions from reaching patients in time.
AI does not fit neatly into a regulatory framework designed for static hardware devices. The FDA should create a streamlined approval pathway specifically for adaptive software, particularly for upgrades. Predetermined Change Control Plans (PCCPs), could be a path forward. While not yet standard practice, PCCPs allow companies to list anticipated updates and how those updates will be validated in advance. This aligns AI oversight with the evolving nature of software instead of treating each update as an entirely new device. This approach should become the norm.
Overregulation of new technology is a recurring pattern. Regulatory systems built for earlier eras rarely anticipate emerging innovations. Agencies are incentivized to avoid visible errors rather than maximizing long-term system-wide improvements. Incumbent providers also have strong incentives to resist technologies that threaten their economic position, often framing the argument as patient protection. But patients are best protected by a healthcare system that deploys the most advanced and effective tools available.
Regulation should be outcome-based. Once AI tools are evaluated on measurable performance metrics—accuracy, error rates, complication reductions—those that match or exceed human performance should be approved and subject to ongoing post-market monitoring. Autonomy should be permitted within clearly defined boundaries, with physician oversight available when clinically appropriate but not mandated as a universal requirement.
The stakes are huge. Healthcare costs will continue rising faster than inflation without productivity growth. AI is one of the few technologies capable of expanding capacity, reducing wait times, lowering administrative expenses, and improving healthcare outcomes simultaneously. Healthcare does not need more precautionary barriers. It needs more capacity. Properly deployed AI is the fastest path to delivering it.
Justin Leventhal is a senior policy analyst for the American Consumer Institute, a nonprofit education and research organization that advocates for consumers through evidence-based analysis and data. Visit www.TheAmericanConsumer.Org or follow us on X @ConsumerPal.
