Abstract

The first steps of integrating AI into teaching environments were automated essay grading systems. These mainly struggled with converting sentences into vectors which arguably is one of the most foundational steps in Machine Learning and the probably the most relevant function for the intended purpose (Awidi, 2024). Further, they also struggled with evaluating overall completeness and coherence of submissions, one equally relevant step. These Intelligent Tutoring Systems tried to reduce teachers' workload and offer more personalized learning for students (Chaudhry & Kazim, 2021).

But Technology has come a long way since then and the new options namely Generative AI may still harbor unused or undiscovered potential. Generative AI, meaning AI models capable of generating new content in basically any form of media (Feuerriegel et al., 2024) the applications in teaching are extensive. But since substantial testing requires educational contexts (Wang et al., 2023; Colby, 2017) truly relevant evaluations of AI in education are more of a rarity than the norm. Therefore, it is the purpose of this thesis to explore and evaluate the capabilities of an ITS based on ChatGPT in the context of an university course in the field of computer science at the University of Duisburg-Essen.