Healthcare exceptionalism in patient perspectives on AI: implications for policy and practice

Evolving healthcare policy and debates about patient protections require empirical evidence on public perspectives to promote trust and sustainability. If the public does not see AI in healthcare as fundamentally distinct from AI in other domains, major policy intervention or fundamental changes to AI governance may not be necessary. However, healthcare exceptionalism in public perception may indicate that current practice is insufficient to safeguard patient trust.

The main outcomes were comfort with healthcare AI and comfort with 12 AI applications in other domains, measured on a 4-point Likert scale. Two factors of AI comfort were retained in exploratory factor analysis. Composite scores of AI comfort were calculated for each factor and compared with respondents' comfort with AI in healthcare using the Wilcoxon signed rank test. Weighted multivariable logistic regressions were used to analyze predictors of comfort with AI.

Comfort with healthcare AI was more closely aligned with perceptions of newer and riskier AI (eg, self-driving cars) than other AI types with which the public may be more familiar or more likely to perceive potential personal benefit (eg, fraud alerts).

Investment in transparency and patient interests will be important to promote sustainability and trust in healthcare AI.