Use of AI for mental health support is rising faster than the clinical evidence supporting it, yet individual-level predictors of acceptance remain poorly understood. The Technology Acceptance Model's core constructs do not capture domain-specific concerns about care quality, safety, equity, and the therapeutic relationship that influence how people evaluate AI in healthcare contexts. Using nationally representative data from the Pew Research Center's American Trends Panel (N = 4,874), this study examined nativity-based differences in willingness to use an AI mental health chatbot and the extent to which perceived impact of AI in healthcare accounts for these differences. Foreign-born adults were 37% more willing than US-born adults (OR = 1.37, p = .009) after adjusting for demographics, technology experience, and survey language. Five measures of perceived impact (care quality, medical mistakes, patient-provider relationships, racial and ethnic equity, and data security)were jointly associated with willingness after adjusting for other covariates (p < .001), and their inclusion reduced the nativity coefficient to marginal significance (OR = 1.24, p = .096). Karlson-Holm-Breen decomposition indicated that approximately 34% of the sample nativity difference was associated with group variation in perceived impact, with care quality and patient-provider relationship expectations contributing to the largest shares. Among foreign-born adults, those with 11–20 years of U.S. residence were the most willing (OR = 2.04, p = .007), while those with 21+ years were comparable to US-born adults. These findings suggest that immigrant populations' greater openness is partly associated with differences in how they evaluate AI's consequences for care quality, equity, and the therapeutic relationship.
Yoo et al. (Fri,) studied this question.