In the modern world, generative artificial intelligence (AI) tools are becoming increasingly adopted, fundamentally transforming higher education learning environments. University students are adapting closely to this situation, increasingly using AI chatbots not only for academic assistance such as information access or homework support, but also for emotional reassurance and stress management (Beale, 2025). In this context, AI chatbots have become important tools for students in terms of both academic and emotional functions.

University students commonly experience academic anxiety. In addition, the perception of loss of control over one's academic responsibilities is closely related to impaired self-regulation and dependence on avoidance-focused coping strategies (Cassady, 2010). Furthermore, individuals experiencing high levels of academic anxiety seek external support to address perceived deficiencies in coping and academic control (Li and Peng, 2021). When external supports, particularly AI tools, are perceived as accessible, supportive, and non-judgmental, anxious individuals become increasingly dependent on them for academic and emotional comfort (Bhatti et al., 2025).

The nature of students' interactions with AI chatbots significantly influences their attitudes toward the tools. If they perceive AI systems as human-like, empathetic, or socially responsive, they may form stronger emotional bonds with these tools (Park et al., 2023).  This phenomenon, known as anthropomorphic interaction, reflects users' tendency to attribute human characteristics and social presence to non-human elements (Wan & Chen, 2021). From an educational perspective, anthropomorphic  interaction may increase engagement and perceived benefits; however, for individuals with attachment anxiety, anthropomorphic  conversational AI may enhance emotional bonding and affect self-regulation by raising the risk of over-reliance (Heng & Zhang, 2025).

Students' beliefs about their ability to effectively use artificial intelligence tools can influence both their coping behaviors and learning strategies from a social-cognitive perspective. When they perceive AI chatbots as socially supportive, especially under conditions of academic anxiety, they may increasingly prefer AI-mediated interaction over independent problem solving or adaptive coping (Kim et al., 2025). Thus, AI chatbots function not only as learning tools but also as emotionally meaningful partners in the academic process.

This study aims to examine the extent to which students’ academic anxiety affects their levels of AI chatbot dependence and how anthropomorphic interaction mediates this relationship. In order to address this aim, the following research question guides the current study:
1. Do anthropomorphic interaction variable have a mediator role in the relationship between university students' academic anxiety and ai chatbot dependence?

Methodology A cross-sectional correlational research design was used to examine the relationships among academic anxiety, anthropomorphic interaction with ai chatbots, and ai chatbot dependence. Participants The sample comprised 456 undergraduate students from various academic disciplines (education, health, social sciences and technical sciences) at multiple universities. Convenience sampling was used to select participants of whom 66% were women. Participants’ ages ranged from 18 to 32 years. The majority (74%) perceived their socioeconomic status as middle-class. Measurement Tools Students’ academic anxiety levels were measured using the Academic Anxiety Scale, which assesses the cognitive, emotional, and somatic dimensions of anxiety in academic contexts (Bülbül & Odacı, 2025). The scale is a 5-point Likert scale (1=Not at all appropriate, 5=Completely appropriate); items in the cognitive dimension are reverse-scored, and higher scores indicate increased academic anxiety. The internal consistency coefficients of the scale were reported as total Cronbach Alpha = .84 (AFA) / .82 (DFA) for the undergraduate sample and total Cronbach Alpha = .85 for the graduate sample. Students’ anthropomorphic interaction levels were measured using the AI Self-Efficacy Scale's (Aca et al., 2025) anthropomorphic interaction subscale, which assesses students' perceptions of AI chatbots as human-like, empathetic, and socially interactive. The Turkish adaptation study consists of 19 items and 4 dimensions: assistance, anthropomorphic interaction, comfort with AI, technological skills; it is a 7-point Likert type scale with no reverse items. The total internal consistency in the adaptation study was 0.94. Students’ AI chatbot dependence levels were measured using the Artificial Intelligence Chatbot Dependence Scale, which measures excessive use, emotional dependence, and difficulty disconnecting from AI chatbots (Durak & Balıkçı, 2025). The scale consists of a single factor and 8 items. The scale is a 7-point Likert type (1=Strongly disagree, 7=Strongly agree) and there are no reverse items. The Cronbach's Alpha for the Turkish adaptation was reported as .90. Data Analysis Descriptive statistics were analyzed through SPSS 26. Structural equation model (SEM) was tested by AMOS 22 as it elaborates the relationships among constructs, and assesses both the structural model and the measurement model (Gefen et al., 2000; Hair et al., 2010). The reliability of all measurement scales was evaluated using Cronbach's alpha coefficient.

According to the study results, there is a positive relationship between academic anxiety and dependence on AI chatbots. As academic anxiety levels increase, students tend to turn more to AI chatbots for academic and emotional support. Students often use AI tools as short-term coping mechanisms to alleviate academic stress (Gao, 2024; Bhatti et al., 2025). According to the research results, there is a positive relationship between anthropomorphic interaction and dependence on AI chatbots. Students who perceive AI chatbots as human-like or emotionally sensitive tend to develop a greater sense of dependence. In the mediating role test, anthropomorphic interaction was found to mediate the relationship between academic anxiety and dependence on AI chatbots. Based on the findings of this study, structured and pedagogically guided AI use may facilitate learning and reduce anxiety. On the other hand, emotional and anthropomorphic interactions with AI may increase students' risk of dependency on these bots. This dependency can weaken students' self-regulation and learning skills (Benchallal, 2025; Klimova & Pikhart, 2025).

Aca, Z., Solmaz, U., & Koçak, O. (2025). Turkish Adaptation of the AI Self-Efficacy Scale: A Psychometric Evaluation. International Journal of Human–Computer Interaction, 1-17. Beale, R. (2025). Adapting University Policies for Generative AI:Opportunities, Challenges, and Policy Solutions in Higher Education. Bhatti, J. A., Ilyas, F., Naseer, H., & Farooq, M. (2025). Exploring the Impact of AI-Powered Assessment Tool on Test Anxiety and Academic Self–Efficacy. Review of Applied Management and Social Sciences, 8(2), 1007-1021. Benchallal, A. (2025). Influence of artificial intelligence on students' learning and psychological well-being: a systematic literature review. Journal of Science and Knowledge Horizons, 5(1), 32-48.‎ Bülbül, K., & Odacı, H. (2025). Akademik Kaygı Ölçeği: Bir Ölçek Geliştirme Çalışması. Dokuz Eylül Üniversitesi Buca Eğitim Fakültesi Dergisi, (63), 477-499. Cassady, J. C. (2010). Anxiety in schools: The causes, consequences, and solutions for academic anxieties. Peter Lang. Hair, J., Black, W., Babin, B., & Anderson, R. (2010). Multivariate data analysis: A global perspective. Pearson. Heng, S., & Zhang, Z. (2025). Attachment anxiety and problematic use of conversational artificial intelligence: Mediation of emotional attachment and moderation of anthropomorphic tendencies. Psychology research and behavior management, 1775-1785. Kim, M., Lee, S., Kim, S., Heo, J. I., Lee, S., Shin, Y. B., Cho, C. H., & Jung, D. (2025). Therapeutic Potential of Social Chatbots in Alleviating Loneliness and Social Anxiety: Quasi-Experimental Mixed Methods Study. Journal of medical Internet research, 27, e65589. Klimova, B., & Pikhart, M. (2025). Exploring the effects of artificial intelligence on student and academic well-being in higher education: A mini-review. Frontiers in Psychology, 16, 1498132. Li, Y., & Peng, J. (2021). Does social support matter? The mediating links with coping strategy and anxiety among Chinese college students in a cross-sectional study of COVID-19 pandemic. BMC public health, 21(1), 1298. Park, G., Lee, S., & Chung, J. (2023). Do Anthropomorphic Chatbots Increase Counseling Satisfaction and Reuse Intention? The Moderated Mediation of Social Rapport and Social Anxiety. Cyberpsychology, behavior and social networking, 26(5), 357–365. Wan, E. W., & Chen, R. P. (2021). Anthropomorphism and object attachment. Current opinion in psychology, 39, 88–93. Yıldız Durak, H., & Balıkçı, H. C. (2025). Yapay Zekâ Sohbet Botu Bağımlılık Ölçeğinin Türkçeye uyarlanması. In All Sciences Academy (Ed.), Proceeding book of 3rd International Conference on Trends in Advanced Research ICTAR 2025 (pp. 1–6). All Sciences Academy.