Peer perceptions of clinicians using generative AI in medical decision-making

Area of Science:

• Medical Decision-Making
• Artificial Intelligence in Healthcare
• Clinical Peer Review

Background:

• Generative AI (GenAI) is increasingly integrated into healthcare, impacting clinical workflows.
• Understanding clinician perceptions of GenAI use is crucial for its responsible adoption.

Purpose of the Study:

• To evaluate how peer clinicians perceive physicians using GenAI in medical decision-making.
• To compare perceptions when GenAI is used as a primary tool versus a verification aid.

Main Methods:

• A randomized experiment involving 276 practicing clinicians.
• Evaluation of three vignettes: no GenAI, GenAI-primary, and GenAI-verify.
• Assessment of perceived clinical skill, healthcare experience, and competence.

Main Results:

• Physicians using GenAI-primary were rated significantly lower in clinical skill (3.79) compared to control (5.93).
• Using GenAI as verification (4.99) partially mitigated negative perceptions.
• Clinicians recognized GenAI's accuracy benefits and favored institutionally customized versions.

Conclusions:

• While clinicians acknowledge GenAI's utility, its primary use in decision-making can negatively affect peer evaluations.
• Framing GenAI as a verification tool can lessen but not entirely remove these negative impacts.
• Customized GenAI tools may be perceived more favorably by clinicians.