Abstract
In order to examine how people learn in the age of AI, concepts of Human-AI Interaction are of central relevance. The present study investigates AI-supported learning processes within the context of writing in German as a Foreign Language. Rather than evaluating the final product in terms of correctness, structure, or argumentative depth, this research focuses on the impact of Human-AI Interaction on the writing process under the premise of sustainable learning.
Successful learning processes require that content is actively processed and understood. While the resulting cognitive load (Sweller et al. 2011) can be reduced through targeted cognitive offloading (Risko 2016) to free up capacity for deeper cognitive processes, delegating tasks to Artificial Intelligence also poses significant risks. Learning, as an active endeavor, requires intensive cognitive engagement with the subject matter (Chi & Wylie 2014). If parts of the process that are constitutive for language acquisition are outsourced to AI, it may negatively affect the internalization and memorization of linguistic forms as well as critical thinking (Yiğit 2025; Tian & Zhang 2025; Sparrow et al. 2011). Regarding the writing process, this implies that while texts may appear superficially coherent, an independent development of arguments may be lacking, and learners may be unable to critically evaluate the results (Eaton 2023).
From a didactical and methodological perspective, the promotion of co-construction is therefore essential, positioning the AI as a critical partner (Kress & Kimmerle 2023). By designing tasks based on Socratic impulses or iterative procedures, cognitive sovereignty is intended to remain with the human agent.
Within the framework of a university-level German course, students are instructed on how to design prompts that allow them to remain active agents in the writing process. Working in learning tandems with a partner and the AI, they complete tasks while the process is documented via screen recording and audio monitoring.
The analysis of these interaction dynamics provides insights into the effectiveness of specific scaffolding approaches in securing writer’s agency and critical thinking (Moorhouse et al. 2025). Furthermore, the study examines which interaction patterns ensure that AI suggestions are not merely adopted but are instead utilized as cognitive stimuli for individual language acquisition.
References
- Chi, M. T. H., & Wylie, R. (2014). The ICAP Framework: Linking Cognitive Engagement to Learning Outcomes. Educational Psychologist, 49(4), 219–243.
- Eaton, S. E. (2023). Postplagiarism: Transdisciplinary Ethics and Integrity in the Age of Artificial Intelligence. International Journal for Educational Integrity.
- Kress, S., & Kimmerle, J. (2023). Co-construction of knowledge in human-AI interaction. Trends in Cognitive Sciences.
- Mayer, R. E. (2012). Multimedia Learning (2nd ed.). Cambridge University Press.
- Moorhouse, Benjamin Luke; Wan, Yuwei; Wu, Chenze; Wu, Meixin; Ho, Tsz Ying (2025): Generative AI tools and empowerment in L2 academic writing. System 133. https://doi.org/10.1016/j.system.2025.103779
- Risko, E. F., & Gilbert, S. J. (2016). Cognitive Offloading. Trends in Cognitive Sciences, 20(9), 676–688.
- Sweller, J., Ayres, P., & Kalyuga, S. (2011). Cognitive Load Theory. Springer Science & Business Media.
- Tian, M., & Zhang, L. (2025). The Erosion of Critical Thinking? A Longitudinal Study on AI-assisted Academic Writing in Higher Education. Journal of Educational Computing Research.
- Yiğit, S. (2024/2025). Cognitive Offloading or Cognitive Erasure? The Impact of Generative AI on Language Acquisition. Modern Language Journal.
