Development of a Multidimensional Psychometric Scale for Assessing Artificial Intelligence Dependency in Higher Education Task Completion

Putri, Wicha Amalia
Aunurrahman, Aunurrahman
Karolina, Venny
Halida, Halida
Afandi, Afandi
Tazijan, Farina

Jurnal Pendidikan Progresif
Vol 16, No 2 (2026): Jurnal Pendidikan Progresif

Recent large-scale surveys indicate that the use of generative Artificial Intelligence (AI) in higher education has become nearly ubiquitous. A survey in the United Kingdom reported that 92% of university students use generative AI tools in their studies. In contrast, a national Indonesian survey involving 1,501 respondents found that 86.21% use AI to assist with academic tasks at least once a month. Preliminary institutional data further revealed that 68.6% of students reported using AI in nearly every assignment. Such high prevalence suggests that AI use has shifted from occasional assistance to habitual reliance, raising concerns about potential dependency and reduced independent cognitive engagement. Although existing instruments, such as the Cognitive AI Dependence and Interaction Scale (CAIDS), assess attitudes and general interaction patterns toward AI, they do not specifically measure functional and cognitive dependency in academic task completion. Therefore, this study aimed to develop and validate an AI task completion dependency scale for university students. A psychometric scale development design was employed involving 500 students from several universities in Indonesia who reported using AI to complete academic assignments. Data were analyzed using exploratory factor analysis (EFA), confirmatory factor analysis (CFA), Rasch modeling, and differential item functioning (DIF) analysis based on gender. The findings revealed a stable three-factor structure comprising functional dependency on AI, reflective attitude, and independent use, and regulation and critical evaluation of AI, yielding a final 10-item scale. The results indicated preliminary structural support for the three-factor model, acceptable reliability, strong item functioning based on Rasch analysis, and acceptable gender invariance. Overall, the developed instrument provides a valid and reliable measure of students’ dependency on AI in academic task completion and offers a practical tool for evaluating and managing responsible AI use in higher education contexts. Keywords: academic task completion , AI dependency , indonesian students , rasch model , scale development.