Related Experiment Videos
Artificial Intelligence Chatbot Dependence, Anxiety, and Depressive Symptoms in University Students: Initial
Iván Barrios1,2, Julio Torales2,3,4, Sandra Blanco1
1Cátedra de Metodología de la Investigación II, Filial Santa Rosa del Aguaray, Facultad de Ciencias Médicas, Universidad Nacional de Asunción, 111421 Santa Rosa del Aguaray, Paraguay.
Background:
Artificial intelligence (AI) chatbots are rapidly becoming integrated into higher education, supporting human-computer interaction, knowledge explanation, rapid access to information, access to online learning resources, and personalized learning pathways. However, little is known about potential patterns of dependence on these technologies and their relationship with anxiety and depressive symptoms. This study aimed to conduct an initial psychometric evaluation of the Spanish version of the Artificial Intelligence Chatbot Dependence Scale (AICDS) and to explore its association with anxiety and depressive symptoms among university students.
Methods:
A cross-sectional study was conducted among 200 university students from Santa Rosa del Aguaray, Paraguay. Participants completed the AICDS together with screening measures of anxiety and depressive symptoms. Data were analyzed using R version 4.5.3. Confirmatory factor analysis using diagonally weighted least squares estimation with robust standard errors was performed to examine the scale's factor structure, and multiple linear regression models were used to evaluate associations between AICDS scores and anxiety and depressive symptoms.
Results:
The findings supported a unidimensional factor structure with acceptable-to-good model fit and high internal consistency (Cronbach's α = 0.833; McDonald's ω = 0.903). Fully standardized factor loadings ranged from 0.538 to 0.873. In bivariate analyses, students who screened positive for anxiety symptoms had higher AICDS scores than those who screened negative, although the effect size was small. However, this association was not maintained after adjustment for sociodemographic and academic covariates in multiple regression models. No significant association was observed between AICDS scores and depressive symptoms under two-tailed testing.
Conclusions:
These findings provide preliminary evidence supporting the internal consistency and factorial validity of the Spanish version of the AICDS in this sample. However, associations between AI chatbot dependence and anxiety or depressive symptoms were weak and not robust after adjustment. Further longitudinal and psychometric research is needed before broader conclusions can be drawn about the mental health implications of AI chatbot dependence among university students.
