Making Cognitive Outsourcing in Education Visible

Educational institutions are increasingly confronted with the use of generative AI to write essays, papers and theses, what disrupts established forms of performance evaluation and support. Beyond structural challenges, cognitive outsourcing also has consequences for individual development. For that reason, some argue that AI should be banned from educational contexts. Many who share this perspective think that a controllable space such as a classroom is enough to implement this approach. But even if you have a controllable space and there is no AI use within it, what happens outside of it will still shape its internal processes. When students use AI at home, the influence of this usage on their thought structures does not stop at the entrance door of the university or school, it enters the "AI-free" zone and shapes cognitive processes there. I think that banning AI systems from educational institutions does not work like banning cars from city centers, because their influence continues to flow invisibly across human minds. Hence, what is banned is not AI, but the discourse about it. Other people argue that AI should be allowed, but that in written outputs AI-generated sentences or text blocks should be highlighted. There are even specific tools for educators to verify this. However, such an approach ignores that a manually written text does not mean that the outlined thoughts also originated from the writer. A person might interact with an AI system, discuss a topic with it over weeks, and later write an essay completely manually. In that case, the person would not highlight the involvement of AI, even though it may have had a huge influence on the thinking of the person. On the other hand, a person might write down a few notes and then have an AI turn them into a polished text. In that case, the ideas that mattered were developed entirely by the person themselve. The AI system simply put them into text. In the highlighting logic, the person would have to label this as AI-generated, even though every idea that mattered came from human thinking. For that reason, I would argue that focusing on the product of thinking, treated as if it exists independently of external influences, is no longer a sufficient approach. Instead, we might need to look at the process of how the interaction with AI actually unfolds and who contributed what and to which extent. Process phases of research-based learning, such as conceptualization, methodology, analysis, interpretation, visualization and writing, should therefore be made visible for reflection, transparency, and evaluation. From the perspective of seeing cognition as distributed rather than as something that happens in the brain of one person, it can be argued that full authorship is not, and never was, a reality. Many current discussions start from the assumption that before the emergence of generative AI, a person was the only author of their intellectual output. I think we have never thought alone and rather within a structurally coupled network with others. Co-authorship, in this sense, is nothing new. What generative AI is disrupting is not just performance evaluation, but the myth of full authorship.

Sep 25, 2026