Stefan Münzer, Stefanie Go und Benjamin Paaßen for the Research Network Artificial Intelligence and Digital Autonomy in Research and Education (AIDARE).
This is the inofficial English translation of the paper’s summary. For this translation (not for the actual text) a local Gemma 4-26b model was used. One of the authors then checked the translation carefully. The full text of the paper can be found (in German) on the website of the Higher Education Forum Digitization (Hochschulforum Digitalisierung): Link to the paper
Generative AI facilitates many tasks for students. But does it support self-determined learning—or does it jeopardize the acquisition of fundamental competencies? The new HFD discussion paper by Stefan Münzer, Stefanie Go, and Benjamin Paaßen examines the risk of deskilling and calls professional societies, instructors, and students to action.
The largely free decision to use generative AI in studies does not automatically express digital autonomy. This is one central thesis of the new discussion paper. The authors understand digital autonomy as the ability to use digital tools in a self-determined manner to pursue authentic educational goals—such as acquiring professional competencies, personal development, and integration into a scientific community and professional culture.
Generative AI has long been part of everyday university life. Students use it, among other things, to make scientific texts more accessible to them, write term papers, prepare for exams, or organize their studies. The primary motivations are often time savings, easier access to complex content, and the outsourcing of cognitive steps. In contrast, deep understanding and intensive engagement with the content are less frequently cited as reasons for use.
The discussion paper therefore warns against underestimating the risk of deskilling—meaning the loss or insufficient development of professional competencies. While quantitative surveys can capture attitudes and self-assessments, they provide little insight into how the use of AI actually affects professional skills. However, observations from concrete teaching and learning situations indicate that fundamental competencies in reading, writing, and independent problem-solving may be impaired. Although many students recognize this risk, they do not necessarily act accordingly (in a self-regulating fashion).
As a consequence, traditional forms of examination are under pressure. Students doubt whether classic term papers are still suitable as reliable evidence of competence under conditions of extensive AI use. At the same time, in the face of inconsistent or missing guidelines, they often operate in a gray area: What kind of AI support is permitted, and at what point does AI support violate the required independence?
The authors therefore take a deliberately sharpened stance: In certain phases of study, stronger regulation of AI use may be necessary to ensure the acquisition of fundamental competencies. Such a restriction of short-term freedom of action would not necessarily contradict digital autonomy. Rather, it could contribute to students being able to judge, act, and professionally evaluate AI results independently in the long term. This does not imply a blanket ban. Rather, the authors call for subject-specific decisions regarding which competencies are indispensable, how they are learned, and in which examination formats their acquisition can be reliably demonstrated.
The paper describes three levels of action: Academic subjects (e.g. via professional societies and faculties) should review their competency goals, curricula, and examination formats. Instructors should make learning objectives and rules for AI use transparent, and create learning opportunities that promote independent competency development. Students, in turn, are called to weigh short-term advantages, such as time savings, against the long-term goals of their studies. The publication is aimed particularly at instructors, faculties, and program coordinators, as well as students, student councils, and professional societies.
Initial discussions regarding digital autonomy in higher education took place during a workshop at the Center for Interdisciplinary Research, Bielefeld. This exchange led to the formation of the AIDARE Education group. Its members helped initiate the development of the issues addressed in the discussion paper.
The discussion paper „Does generative AI jeopardize digital autonomy in studies?“ has been published as Discussion Paper No. 41 of the Hochschulforum Digitalisierung. The authors are Stefan Münzer, Professor of Educational Psychology at the University of Mannheim; Stefanie Go, research associate at Bielefeld University and the Social Services Competence Center; and Benjamin Paaßen, Junior Professor for Knowledge Representation and Machine Learning at Bielefeld University.