Did Your Doctor Use AI in Med School? Effects of Disclosed AI Use During Training on Trust in Experts
Prior work asks how patients react to AI used at the point of care. We ask a different question: whether learning that an expert’s competence was developed with AI assistance changes how much that expert is trusted. Participants recruited on Prolific read a vignette about a doctor who recently finished training, with the training history manipulated between subjects, and then rated their comfort and confidence along with the doctor’s ability, benevolence, integrity, and autonomy of judgment: Mayer, Davis, and Schoorman’s model of trust, extended with a fourth dimension for independence of thought.
- Study 1
- Used AI / did not use AI / no information, cardiologist scenario. N = 216.
- Study 2
- Adds a peer-tutor arm, to separate a penalty for AI specifically from a penalty for any outside help. N = 335.
- Method
- Preregistered. One-way ANOVA with corrected pairwise contrasts, then parallel mediation (PROCESS model 4, 10,000 bootstrap samples).
Read the poster ↗
Zoomable, and open for comments.
