The main aim was to examine patterns of AI use among university students in Croatia. Specific objectives included exploring AI use for learning and academic tasks, investigating whether AI substitutes interpersonal communication, and examining students’ perceptions of ethical and safety issues related to AI. A qualitative research design was employed using semi-structured interviews. Data were collected between March 11 and April 26, 2025. The final sample consisted of 24 students (average age 22.6) from various Croatian universities and faculties, all with prior experience using AI tools. Interviews were audio-recorded, transcribed, anonymized, and analyzed using thematic analysis. Two researchers independently coded 203 statements, with an inter-rater reliability of 75.37%, ensuring analytical rigor and reliability. All participants reported using ChatGPT, making it the most dominant AI tool, while smaller numbers used Perplexity or DeepSeek. Students primarily learned about AI through professors, peers, and media. Academically, AI was most often used for idea generation, summarizing literature, translation, studying support, and finding information, while it was viewed as less reliable for calculations. Students generally perceived AI as a time-saving and supportive learning tool rather than a replacement for genuine understanding. Although some concerns were raised regarding overreliance, reduced learning depth, data privacy, and ethics, most participants viewed AI as a helpful assistant rather than a substitute for human relationships or creativity. Overall, the findings suggest that Croatian students integrate AI pragmatically, valuing its efficiency while maintaining awareness of its limitations.

Authors:
Vanesa Varga, University of Zagreb, Croatia
Katarina Dadić, University of Zagreb, Croatia

About the Presenter(s)
Dr Vanesa Varga is a University Assistant Professor/Lecturer at Faculty of Croatian Studies, University of Zagreb in Croatia

See this presentation on the full schedule – Tuesday Schedule

Bookmark