Ethical cognition, AI anxiety, and attitudes toward artificial intelligence: validation of the GAAIS, AT-EAI, and AIAS in an Albanian higher-education context.

backgroundCross-cultural psychometric evidence on instruments measuring attitudes toward artificial intelligence (AI) remains scarce in transitional higher-education settings. This study provides the first Albanian-language validation of three widely used instruments - the General Attitudes toward Artificial Intelligence Scale (GAAIS), the Attitudes toward Ethical Artificial Intelligence Scale (AT-EAI), and the AI Anxiety Scale (AIAS) - and, as a secondary exploratory aim, examines how ethical cognition and AI-related anxiety are associated with positive and negative orientations toward AI.

methodsA cross-sectional survey was conducted with 705 students and staff at one Albanian public university (University of Shkodra). Instruments were translated and back-translated, pilot-tested, and evaluated using confirmatory factor analysis (CFA). Hierarchical regression models examined exploratory associations between the AT-EAI subscales (fairness, transparency, non-maleficence, privacy, responsibility), the AIAS subscales (cognitive, affective), and the two GAAIS dimensions, controlling for demographics.

resultsAll three instruments showed good factorial validity and reliability in Albanian (CFI ≥ 0.987; α range 0.74-0.91). In the exploratory regression models, non-maleficence was the only ethical dimension positively associated with positive GAAIS (β = 0.17), while privacy concerns were associated with stronger negative GAAIS (β = 0.10); these ethical effects were small in magnitude. Cognitive and affective AI anxiety showed the largest associations, both with reduced positive attitudes and increased negative attitudes, with affective anxiety strongest in the negative-attitudes model (β = 0.32).

conclusionsThe study's primary contribution is the provision of the first validated Albanian versions of the GAAIS, AT-EAI, and AIAS, which are made available in the Supplementary Appendix for use by other researchers. Within this single-institution validation sample, the associational findings further suggest, in exploratory terms, that emotional responses to AI - more than ethical evaluations - are linked to students' general attitudes. Given the cross-sectional, single-institution design and the modest magnitude of several effects, generalisation to broader Albanian higher education or to AI policy contexts should be treated with caution.