New research highlights how AI is reshaping the patient experience, but the shift also raises concerns about privacy and accuracy.
From Public Chatbots to Closed-Loop Systems
A March 2026 poll found that almost one‑third of adults have typed their symptoms into generative AI platforms to seek health advice. Those tools often draw on public databases, meaning personal details can be exposed to entities without strict confidentiality safeguards.
Scientists at the Duke University School of Medicine warned that such chatbots may produce “hallucinations,” or misleading answers, because the models prioritize user satisfaction over clinical correctness. The risk of inaccurate guidance has prompted many health systems to develop closed‑system, HIPAA‑compliant AI tools.
These newer solutions range from simple agentic chatbots to sophisticated applications that tap into electronic health record (EHR) data for real‑time feedback. A small Deloitte study of senior healthcare executives revealed that just under a third are now focusing on “virtual‑first, AI‑orchestrated access” to streamline the patient journey.
“They’re looking at digital patient access as a real opportunity,” said Dr. Bill Fera, a principal at Deloitte. “In addition to direct care where appropriate, digital interactions give hospital systems a better understanding of what patients are concerned with. If there is a change in the pattern of visits or complaints, they can better ramp up or ramp down specific services and improve the system’s ability to react to and meet patient needs.”
Case Studies: Hartford HealthCare
Connecticut‑based Hartford HealthCare, a network with more than 500 locations and 48,000 employees, partnered with K Health to launch HHC PatientGPT earlier this year. The tool runs on Google Cloud and can access a patient’s records in the organization’s Epic EHR system.
Dr. Padmanabhan Premkumar, president of Hartford HealthCare Medical Group, said the primary driver was a commitment to expanding patient access and deepening digital engagement. “The healthcare journey can be incredibly fragmented and overwhelming for patients,” he explained. “When patients leave a phone message or portal inquiry, that waiting period creates an information vacuum, and many turn to open search engines for immediate answers.”
Related: Pacifica Hospital Files for Bankruptcy Protection
Patients can authorize HHC PatientGPT to retrieve their medical data, allowing the AI to answer general health queries, schedule appointments, or direct users to 24/7 virtual care. An additional AI layer reviews each interaction for quality and safety, escalating any concerns to a clinician when needed.
Hartford HealthCare aims to reach more than 1 million adult patients across Connecticut by the end of 2026, inviting them to try the service.
Both initiatives illustrate a broader trend: health systems are moving from open‑ended AI chat experiences toward secure, patient‑specific tools that tie directly into clinical workflows.
From a broader perspective, the push for AI integration in patient interfaces reflects an industry-wide effort to reduce friction points that often delay care. By linking questions directly to a patient’s record, these systems aim to cut down the “information vacuum” that can lead patients to seek answers from unreliable sources.
Still, the balance between convenience and confidentiality remains delicate.
As more hospitals adopt similar platforms, regulators and providers will need to monitor how these tools perform in real‑world settings, ensuring that the promise of a smoother AI patient experience does not come at the expense of safety.