Lower trust in artificial intelligence than physicians coexists with high willingness to engage in artificial intelligence‐assisted prostate cancer management among active surveillance patients
Eric S. Chai, Kaitlynn Abraham, Riley Hanes, Adam Williams, Howard Wolinsky, Oleksandr N. Kryvenko, Radka Stoyanova, Frank J. Penedo, Archan Khandekar, Bruno Nahar, Mark L. Gonzalgo, Dipen J. Parekh, Sanoj PunnenAbstract
Objectives
This study aimed to examine the sentiment for artificial intelligence (AI) among prostate cancer (PCa) patients on active surveillance (AS) and to evaluate how that sentiment relates to patient willingness to engage with AI.
Materials and Methods
We conducted a 23‐question online survey (November–December 2025) distributed through an AS platform. Analyses were restricted to 110 respondents with PCa on AS. Outcomes included trust ratings (10‐point Likert scale) for physicians versus AI, responses to hypothetical AI management recommendations and perceived concerns and facilitators of trust. Statistical analyses included the Wilcoxon signed‐rank tests, the McNemar tests and Fisher's exact tests.
Results
Median trust in urologists (8 [IQR 7–9]) exceeded that for AI (6 [IQR 5–7]), with smaller differences for radiologists and pathologists compared with diagnostic AI (all p ≤ 0.001). Despite lower trust in AI, most respondents anticipated improved PCa outcomes with AI integration. High willingness to consider AI recommendations was also observed. Only 4.6% ( n = 5) of respondents outright rejected AI recommendation for treatment and only 3.7% ( n = 4) rejected AI recommendation for AS, with all others willing to consider AI recommendations. No respondents fully accepted AI recommendation for treatment without additional verification, whereas 2.8% ( n = 3) fully accepted AI recommendations for continued surveillance. Primary concerns included AI errors (57.3%) and lack of human oversight (48.2%).
Conclusions
Among AS patients, lower trust in AI compared to physicians coexists with a high optimism for AI and willingness to incorporate AI recommendations into decision‐making. This discordance suggests that behavioural adoption of AI‐assisted care may precede full patient confidence.