Generative AI in Academia: A Survey of Students and Staff at the University of Tübingen

Abstract

Generative AI tools, such as ChatGPT, are significantly impacting various facets of daily and academic life, including academic practices at the University of Tübingen. To empirically understand how generative AI is reshaping academic activities, we conducted a survey among students and staff at the University of Tübingen, focusing on their use of generative AI-based tools within study, research, and teaching contexts. The survey was released on March 3, 2024, and remained online for one month. It aimed to understand the frequency and depth of generative AI usage, assess user knowledge and trust in these tools, and identify the specific tools and tasks for which they are used. The primary objective was to determine usage patterns and how these vary across academic disciplines, academic positions, and educational levels. Over 500 students and staff participated in the survey. The findings reveal that nearly half of the respondents regularly use AI-based tools for their academic work. The analysis highlights that these tools are predominantly used for answering comprehension questions, explaining subject-specific concepts, translation, and programming. Some interesting findings indi-cate that with more prior knowledge of generative AI technology, trust in these technologies increases. Furthermore, the higher the academic status at the university, the less researchers trust in generative AI. We could also see that younger participants (18-20 years) reported a higher incidence of no prior knowledge and less frequent use of generative AI.