Development and psychometric evaluation of the BERTA Artificial Intelligence Addiction Scale (B-AIAS)

Abstract

Artificial intelligence (AI) technologies are increasingly integrated into daily life. While offering significant benefits, their excessive use may trigger problematic patterns resembling behavioral addictions. Despite these concerns, no standardized instrument currently exists to assess AI-related addictive tendencies. This study aimed to develop and validate the BERTA Artificial Intelligence Addiction Scale (B-AIAS) among adults in Turkey. Following the COSMIN framework, the scale development process included item generation, expert review, exploratory factor analysis (EFA), confirmatory factor analysis (CFA), assessments of convergent and discriminant validity, Item discrimination analysis, and test–retest reliability. Data were collected via online surveys distributed through social media platforms. The sample consisted of 308 participants for EFA and 306 for CFA, all reporting AI tool usage. The final scale comprised eight items rated on a five-point Likert scale (0 = Not at all, 4 = Completely) across two factors: Deprivation (4 items) and Sensitivity (4 items). The two-factor model explained 60.2% of the variance. Cronbach’s alpha was 0.88 and McDonald’s omega (ω) coefficient was 0.90. No items were reverse-coded. The B-AIAS demonstrated strong psychometric properties and provides a valid and reliable tool for assessing problematic AI use in the Turkish context. This instrument can support future research on technology-related behavioral addictions and enable cross-cultural investigations of AI use patterns.