Related Experiment Videos
Enhancing Physician Resilience to Generative AI: Multilevel Framework for Shared Authority, Verification, and Skill
Hongxia Pan1, Jialin Liu2,3, Siru Liu4
1Rehabilitation Medicine Center and Institute of Rehabilitation Medicine, West China Hospital of Sichuan University, Chengdu, Sichuan, China.
Journal of Medical Internet Research
|June 24, 2026
Summary
Generative artificial intelligence (AI) in healthcare offers efficiency but risks physician deskilling. A new framework enhances physician resilience and safe AI integration by managing cognitive load, clinical authority, and organizational oversight.
Area of Science:
- Medical Informatics
- Artificial Intelligence in Healthcare
- Clinical Decision Support Systems
Background:
- Generative AI tools are increasingly used in clinical settings for diagnosis, triage, and treatment planning, promising efficiency gains.
- However, risks such as AI hallucinations, miscalibrated confidence, and automation bias can increase physician verification burden and deskilling.
- Current AI governance frameworks inadequately address physician-facing resilience, crucial for maintaining safe clinical judgment alongside AI.
Purpose of the Study:
- To propose a multilevel governance framework to enhance physician resilience when collaborating with generative AI.
- To outline mechanisms for safer integration of generative AI in clinical care, minimizing workflow disruption and preserving clinical judgment.
Main Methods:
- A viewpoint proposing a governance framework with three coordinated domains: cognitive workload shaping, clinical authority governance, and organizational safety governance.
- Includes mechanisms like risk-sensitive verification triggers, bounded delegation, and structured interprofessional review.
- Focuses on reducing verification burden and aligning institutional oversight with safe, context-sensitive AI use.
Main Results:
- The proposed framework aims to reduce physician verification burden and preserve decisional authority.
- Mechanisms are designed to support safe clinical integration while minimizing workflow disruption.
- Potential limitations include workflow friction, alert fatigue, and resource variability.
Conclusions:
- A structured governance framework is essential for the safe integration of generative AI into clinical practice.
- The framework addresses cognitive workload, clinical authority, and organizational accountability to support physician resilience.
- Ongoing monitoring and recalibration are necessary to ensure safeguards remain clinically useful and not burdensome.
