Healthcare providers are dramatically underestimating patient acceptance of artificial intelligence in clinical decision-making, according to new research from Gallup that could reshape how health systems deploy AI diagnostic and treatment tools.

The survey, which polled over 5,000 American adults in August, found that more than half of U.S. patients are comfortable with AI making clinical decisions about their diagnosis and treatment—but only when healthcare providers retain the ability to review and override those decisions. The catch: physicians don't realize patients feel this way.

The 20-Point Perception Gap

In a separate survey of over 1,000 practicing primary care physicians, Gallup discovered that only about one-third of providers believe their patients would be comfortable with AI-assisted clinical decisions, even with provider oversight built in. That represents a roughly 20-percentage-point gap between actual patient comfort levels and what clinicians assume.

The disconnect suggests that providers may be unnecessarily limiting AI deployment based on incorrect assumptions about patient resistance. Rather than presuming patients will reject AI-driven recommendations, the research indicates clinicians should directly ask patients about their preferences for AI involvement in care decisions.

Nearly half of adults surveyed identified provider review as the most critical factor increasing their comfort with AI-led decisions on diagnosis, treatment, and imaging review. One-third pointed to strong privacy protections as the next most important consideration. Patient comfort dropped sharply when providers were removed from AI decision-making entirely.

Implications for Deployment

The findings arrive as healthcare AI adoption accelerates. Utah approved new healthcare pilots last month that would eventually allow prescriptions to be issued and refilled without clinician review, representing a more autonomous approach than most patients currently prefer, according to the Gallup data.

For health systems implementing AI clinical decision support tools, the research suggests an opportunity to accelerate integration in diagnostic and treatment workflows, provided they maintain robust provider oversight mechanisms. Patients showed highest comfort with AI applications in diagnosis, treatment decisions, medication selection, and imaging review—all contingent on that human safety net.

The perception gap has immediate implications for AI vendors and healthcare organizations alike. For technology companies, the research indicates that emphasizing human-in-the-loop architectures and provider review capabilities should be central to product positioning and user interface design.

For healthcare organizations, the findings highlight a need for shared decision-making conversations between clinicians and patients about AI use preferences. As the industry moves toward more autonomous agentic AI systems capable of completing multi-step clinical tasks with minimal human intervention, understanding actual patient comfort thresholds—rather than assumed ones—becomes essential to both adoption success and trust maintenance.

The Gallup research suggests the barrier to AI clinical adoption may be less about patient resistance and more about provider assumptions that don't align with patient reality.

Sources

  1. Patient comfort with AI's clinical decisions tied to provider oversight: Gallup — Healthcare Dive