Patient attitudes toward artificial intelligence-driven dermatology

Artificial intelligence (AI) systems are increasingly being developed for dermatology, but the conditions under which patients trust these tools remain poorly characterized. This Preferred Reporting Items for Systematic Reviews and Meta-Analyses-guided systematic narrative review synthesizes peer-reviewed empirical studies, published from January 1, 2015 through July 7, 2026, that directly assessed patient or public attitudes toward AI in dermatologic diagnosis, skin-cancer screening, image sharing, teledermatology image-quality support, and related patient-facing workflows. Eligible evidence included qualitative interviews and focus groups, surveys, preference/conjoint or vignette experiments, and early clinical deployments. Twenty studies met the inclusion criteria using the review framework. Across study designs, patients expressed conditional support for dermatology AI but consistently favored dermatologist-supervised AI over autonomous AI, especially for high-stakes diagnostic decisions. Trust was strengthened by visible clinician accountability, clear explanations of AI outputs and uncertainty, privacy safeguards, transparent data use, and evidence that performance is reliable across skin tone and diagnosis. Trust declined when AI was framed as replacing clinicians, when image governance was unclear, or when unequal performance was perceived. These findings support a model of augmented intelligence: dermatology AI should serve the clinical relationship; be subject to human oversight; and be evaluated and reported stratified by skin tone, diagnosis, and care context.