Teachers’ self-reported AI literacy across educational levels in Greece: Awareness without conceptual depth?

Artificial intelligence (AI) literacy is increasingly recognized as an essential teacher competence, yet evidence concerning its dimensional structure and variation across educational sectors remains limited. This study investigated self-reported AI literacy among 1,173 in-service teachers in primary, secondary, and higher education in Greece, examining its empirical structure, overall levels, and associations with prior AI-related training and teaching level, while accounting for academic qualifications. Data were collected through a researcher-developed questionnaire and analyzed using exploratory factor analysis, descriptive statistics, and general linear models. Two dimensions emerged: Conceptual–Technical AI Literacy and Critical–Ethical AI Literacy, while Practical AI Use was retained as a separate indicator. Overall self-reported AI literacy was relatively low, with substantially higher scores on critical–ethical indicators than on conceptual–technical indicators, suggesting a profile of perceived awareness without corresponding conceptual depth. Prior AI-related training was positively associated with all AI-literacy outcomes across educational sectors. However, significant teaching-level × training interactions revealed particularly strong associations in higher education, especially for Conceptual–Technical AI Literacy. These findings highlight the multidimensional and context-sensitive nature of teachers’ self-reported AI literacy and the importance of distinguishing perceived confidence from demonstrated competence. They underscore the need for differentiated, pedagogically grounded professional development that integrates conceptual understanding, practical AI applications, critical evaluation, and ethical judgment across educational sectors.