AI and algorithmic literacy among health workers: a scoping review through a digital health literacy lens

Abstract

Introduction: Artificial intelligence (AI) and algorithmic systems influence health workers' access, interpretation, and action on clinical and public health information, positioning them as intermediaries between algorithmically mediated outputs and patients, communities, and decision makers. This study examines how AI and algorithmic literacy are conceptualized and measured among health workers through a digital health literacy (DHL) lens. Methods: Using Arksey and O'Malley's scoping review framework, we searched Ovid MEDLINE, Ovid Embase, Scopus, IEEE Xplore, ACM Digital Library, Europe PMC, and arXiv for English language sources published between January 2020 and May 2025. Two reviewers screened records and extracted data using a theory informed charting framework grounded in Nutbeam's model (functional: basic understanding and use; critical: evaluation and ethics; communicative: interacting with AI systems and explaining AI-mediated information). We synthesized findings using descriptive statistics and a narrative synthesis. Results: Twelve studies published between 2021 and 2025 met inclusion criteria. Evidence was concentrated in health professions education (10/12), primarily among medical (6/12) and nursing students (2/12), with no studies exploring public health practice. Explicit, theory-grounded definitions of AI literacy were uncommon, and links to DHL were only implied. AI literacy was frequently operationalized through self-reported instruments, commonly the Artificial Intelligence Literacy Scale (AILS; 3 studies), Meta Artificial Intelligence Literacy Scale (MAILS; 2 studies) and the Scale for the Assessment of Non-Experts' AI Literacy (SNAIL), alongside self-developed tools. Only one study explicitly defined and measured algorithmic literacy as a distinct construct; in other studies, algorithmic considerations appeared indirectly through recognizing AI presence in systems or evaluating AI generated content. Across studies, competencies aligned mainly with functional and critical dimensions of DHL, particularly awareness, use, evaluation, and ethics, while communicative literacies were infrequently assessed. Discussion: AI and algorithmic literacy among health workers is underdeveloped, weakly integrated with digital health literacy, and inconsistently measured. Research prioritizes AI literacy using non-health-specific self-report tools and largely overlooks communicative competencies essential to clinical and public health practice. These findings point to the need for clearer conceptual alignment, health-specific measurement, and systems-based approaches to workforce readiness as AI-enabled tools expand across healthcare and public health.

Ähnliche Arbeiten