Federal health leaders are pushing artificial intelligence as a solution to rural America's healthcare crisis, but the patients it's meant to serve remain deeply skeptical - and the evidence backing these tools is thin.
Health secretary Robert F. Kennedy Jr. recently told senators that AI nurses could provide "concierge care" to rural patients. Mehmet Oz, who leads the Centers for Medicare & Medicaid Services, has suggested "AI-based avatars" could connect rural patients to mental health services. States are spending money from the $50 billion federal Rural Health Transformation Program on AI projects. The five-year program was created last summer as part of the One Big Beautiful Bill Act.
But in Hot Springs, South Dakota, a town of about 3,400 residents at the southern end of the Black Hills, residents who spoke with KFF Health News were reluctant to trust their health to software.
"I get artificial intelligence for certain things, but for personal healthcare - no," said Tara Haffner, standing outside the American Legion. She said she worries about AI making mistakes and wants healthcare to stay between her and her doctor.
Evidence lags behind adoption
The healthcare industry is adopting AI rapidly despite the tools being "poorly evaluated," according to a recent report from ARISE, a Stanford- and Harvard-led research group. While some AI works in controlled settings, there's less evidence it can perform in the real world. Few studies track patient outcomes.
The lack of evidence is worse in rural areas. A recent academic paper found only 26 peer-reviewed studies about AI in rural healthcare were published from 2010 through April 2025. Few analyzed implementation or outcomes.
"We do not have a lot of resources to waste on tools that don't work in rural areas," said Qian Huang, an assistant professor at the Center for Rural Health and Research at East Tennessee State University.
Huang said the technology is usually tested at large academic hospitals and trained on data from urban patients, who may not face the same obstacles - such as lack of transportation - as rural patients. The tech needs to work for distinct populations, she said.
A tool to help, not replace
Not everyone is against the idea. Phillip Mues, who oversees technology at Cherry County Hospital and Clinic in rural Valentine, Nebraska, said AI is already helping clinicians save time and reduce burnout.
"I think it will help reduce burden on actual staffing," Mues said. "It won't replace people, but I think it will help in rural communities."
The clinic uses AI scribes that record appointments and generate notes, letting providers keep "eye contact on the patient, not the computer," Mues said. Surveys by him show the scribes have helped reduce burnout.
But Mues cautioned that AI can't fix every challenge. Rural hospitals at risk of closing likely can't use AI to save enough money to prevent those outcomes, he said.
Stephanie Keller, another Hot Springs resident, wears a smartwatch to track her fitness but has no interest in an AI chatbot encouraging her to reach health goals.
"I don't have the time to chat with AI every day. I mean, are you kidding me? I don't want to spend my time on a cellphone," she said.
States move ahead anyway
A KFF Health News review of state plans for the rural health program found widespread interest in AI for back-office tasks like medical charting, coding, referrals, and prior authorization. Several states also want AI to suggest diagnoses and recommend treatment options.
Mississippi plans to use predictive AI to "guide" emergency medics with "triage, tracking, and treatment decisions." North Dakota wants AI to "detect early signs of chronic disease and behavioral health conditions." Utah is funding AI-powered fetal-monitoring devices, and Kentucky is exploring AI chatbots to "deliver personalized support" through health coaching and rewards.
Whether states will share results is unclear. CMS spokesperson Timothy Foster said the agency doesn't have AI-specific reporting requirements but is working on a form for states to report progress and outcomes.
Some states track only adoption metrics. Vermont, for example, requires organizations using AI scribes to report how many clinicians and patients are served, but not how much time they save. Connecticut will track AI-powered patient monitoring alerts. Texas will track cost savings.
Why this matters for healthcare professionals
If you work in healthcare - whatever your specialty - these tools are coming your way, and your experience will matter. The rural health program will put AI in small hospitals and clinics that could add steps rather than eliminate them. Even if you support it, remember: Mues calls it "a tool" for helping staff, not replacing them.
Rural health facilities face real barriers to making AI work. They may not have the hardware, IT staff, or internet speed needed to support the technology, Huang said. Clinicians already juggle multiple responsibilities and may not have time for training.
"In rural communities," trust and personal relationship is essential," Huang said.
Patients have clearly said they will resist care mediated entirely by machines. One resident echoed that when he told KFF: "I don't trust AI in healthcare, or anywhere else. I'm old-fashioned. I don't believe in it."
For rural healthcare professionals, the path is clear: use AI for the work it's good at - the paperwork, the charting, the administrative overhead - and keep the exam room human-facing. That's how you buy time without losing trust.
Your membership also unlocks:
