Generative AI Chatbots in Higher Education : ATAM-Based Analysis of Discipline-Specific Adoption Patterns for Students at the University of Borås

Sammanfattning

The rapid adoption of generative AI chatbots like ChatGPT and Microsoft Copilot is reshaping academic practices in higher education, offering both learning opportunities and challenges to academic integrity. However, gaps remain in understanding how students across disciplines actually use these tools and universities' impact on usage patterns. Therefore, this thesis explores chatbot usage among students at the University of Borås in Sweden. More detailed, it analyses the frequency and purposes of use, perceived usefulness and ease of use, and the influence of social and institutional factors. The study applies the Technology Acceptance Model (TAM), extended with constructs of utility, social influence, and concerns. A mixed-methods approach was used, combining a survey (n = 157) and semi-structured interviews (n = 2). Results show high adoption, especially in technology-related fields. Chatbots are commonly used for brainstorming and feedback on writing. However, the discipline-specific usage varies. Students also find chatbots most useful for improving efficiency and understanding. Despite the provision of Microsoft Copilot by the University of Borås, just 9.30% of students use it compared to 80.50% using ChatGPT. Moreover, only 3.40% of students stated that this was a decisive factor in their choice. This highlights the large gap between the use of chatbots and awareness of AI policies, as 75.70% of students are only partially aware or not aware at all. The findings highlight the need for clearer policies and targeted communication to support ethical and effective chatbot use in higher education.