Physician versus patient use of AI for diabetes prevention: public perceptions and comfort levels

Type 2 diabetes is a major public health challenge that can often be prevented through lifestyle interventions. Artificial intelligence (AI) is increasingly used for risk prediction, behavioral coaching, and individualized prevention, offering scalability and low-intensity interventions. However, AI also raises ethical and regulatory concerns, especially for patient-facing tools. Limited evidence compares public comfort with physician use versus patient use for diabetes prevention.

We analyzed data from a 2025 national survey conducted through the NORC AmeriSpeak Panel, a probability-based sample of 1,939 respondents. Participants evaluated two hypothetical AI use cases for diabetes prevention: Physician use of AI and patient use of AI-chatbot. Paired t-tests compared comfort across use cases. Weighted univariable and multivariable logistic regression models identified predictors of comfort.

Participants reported significantly greater comfort with physician use of AI than with patient use of an AI-chatbot for diabetes prevention (p < 0.001). Comfort across both cases was strongly associated with belief that AI benefits population health (patient-AI: OR = 3.67, p < 0.001; physician-AI: OR = 3.86, p < 0.001), trust in health system AI use (patient-AI: OR = 1.47, p < 0.001, physician-AI: OR = 1.72, p < 0.001), and physician confidence in AI reliability (patient-AI: OR = 1.31, P = 0.003; physician-AI: OR = 1.30, p < 0.001). Comfort with physician use of AI was lower among Black/African American participants than White participants (OR = 0.51, p < 0.001), while women reported lower comfort with patient use of AI compared than men (OR = 0.71, p = 0.032).

Public comfort with AI for diabetes prevention appears higher when integrated with professional oversight. Trust in clinicians, health systems, and AI reliability may be central to acceptance. Differences across demographic groups highlight the importance of equity-focused, physician-led implementation, transparent communication, and inclusive trust-building strategies for ethical AI adoption.