Clinical Report: Balancing Rapid Access and Clinician Oversight in Patient-Facing AI

Overview

In a 2024 cross-sectional survey of 1,185 US adults, preferences between rapid AI-generated portal replies and slower clinician-reviewed replies differed by message type and respondent. Respondents more often chose immediate AI for logistical messages and clinician review for medical messages, while mental health support preferences varied across therapist and chatbot options.

Background

Patient-facing AI may provide timely information and support, while clinician review requires time and can delay responses. Benda and colleagues examined how respondents weighed response speed against professional involvement across portal-message categories and mental health support options. Related survey evidence reports that positive views of AI can coexist with a preference for physicians to retain final diagnostic authority. The source also notes that the survey bundled response time, authorship, and review, so it cannot isolate which attribute drove each choice.

Data Highlights

Survey: 1,185 US adults recruited using quota sampling; portal-message choices contrasted an AI response within 5 minutes with a health-professional-reviewed response within 48 hours.

Key Findings

• For logistical messages, 57.4% selected the immediate AI response; more than 40% nevertheless preferred waiting for professional review.
• For general messages, choices were nearly evenly divided: 50.6% selected immediate AI.
• For medical messages, 63.1% selected professional review, while more than one-third selected immediate AI.
• Likely use of mental health support options was reported by 63.9% for a licensed therapist available in 3 months, 50.2% for a therapist-monitored chatbot available in 1 month, and 44.3% for an immediately available stand-alone chatbot.
• The survey did not independently vary response time, authorship, and review, and did not define the nature of clinician oversight; therefore, it cannot determine which of these features drove respondents’ choices.
• Related evidence cited in the source found that 57.6% of 13,806 hospital patients across multiple countries viewed AI in health care positively, while 72.9% wanted physicians to retain final diagnostic authority.

Clinical Implications

The source states that safety should establish minimum oversight, while patient preference may prompt additional review. It describes direct AI responses as potentially suitable, following local validation, for standardized tasks with verifiable outputs, limited consequences of error, and dependable escalation pathways; new symptoms, treatment decisions, ambiguous requests, and emotionally consequential communications may warrant clinician assessment before transmission. A clinician-reviewed label alone does not specify the safeguard: the source identifies adequate reviewer context and time, authority to modify responses, and a practical means to intervene as important features of review.

Conclusion

Survey respondents’ preferences varied by task and individual, and rapid access did not uniformly outweigh human involvement. The findings support distinguishing the response-time, authorship, and oversight components when interpreting patient preferences.

Related Resources & Content

1. Benda et al, JAMA Network Open, 2026 -- Public Views on Response Speed and Human Oversight in AI-Assisted Care
2. MDSpire News, year not specified -- Oversight May Be Critical in Patient Acceptance of AI in Radiology
3. ASCO AI in Oncology, 2026 -- Panelists Map Which Clinical Tasks AI Should Own, and Which It Should Not
4. aace endocrine ai — Effective medical AI oversight requires practical safeguards
5. Public Views on Response Speed and Human Oversight in AI-Assisted Care
6. Oversight May Be Critical in Patient Acceptance of AI in Radiology
7. Panelists Map Which Clinical Tasks AI Should Own, and Which It Should Not
8. Artificial intelligence-related health research: ethics review and oversight
9. Public Perceptions of Timeliness for AI Use and Human Oversight | Digital Health | JAMA Network Open | JAMA Network