ChatGPT theses. Identifying distinctive markers in AI-generated versus human-created texts: A multimodal analysis in university education

Abstract

This study examines the distinguishing markers between AI-generated and human-created texts within university education, specifically in the context of student assignments. Using a multimodal qualitative approach, we analysed selected Bachelor’s and Master’s theses, identifying recurring characteristics of AI-generated content. Key markers include among others: fragmented argumentation flow, specific use of language, formulaic structures, and excessive objectivity, which contrast with the more nuanced and cohesive style typical of human writing. The research also includes survey responses from students across two universities (Europe & Asia), revealing varied attitudes towards AI use in academic work. While some students view AI as a beneficial tool for text enhancement, others express ethical concerns regarding its impact on authorship and originality. The findings suggest that both students and educators must adapt to the growing presence of generative AI in academia, balancing its use with critical thinking and ethical guidelines. This research provides a foundation for developing AI literacy and detection methods, highlighting the need for systematic strategies to integrate AI responsibly within educational frameworks, thus supporting academic integrity and fostering digital literacy.

Ähnliche Arbeiten