Most students use AI – but their attitudes toward it are ambivalent. On the one hand, they see it as helpful for working more efficiently, broadening their knowledge, and overcoming writing difficulties. On the other, they worry about becoming overly dependent on it and losing important skills, while also recognizing the value of writing without AI support. These are among the findings of a recent survey conducted by Goethe University’s Schreibzentrum with 4,048 participants. The key question now is how universities should respond.

The wish commonly voiced by students: Responsible limits on AI use
According to the survey, 89 percent of students use AI tools such as ChatGPT or DeepL in their academic writing – primarily to begin working on a topic, brainstorm ideas, or revise language and style. AI is used less frequently for analyzing research texts or drafting and revising substantive content: only around one-third of respondents reported using it for these purposes. “The majority of students try to limit their AI use to tasks they consider justifiable,” explains study author Nora Hoffmann. “Many are highly aware that AI systems are prone to errors, and they are concerned about losing skills through excessive reliance on them.” Slightly more than half of respondents said they felt they accomplished less independently when using AI, while 45.5 percent worried about losing their ability to think critically. At the same time, 92.2 percent said they felt responsible for the texts they submitted, and 78 percent described writing as an important way to develop ideas and learn.

The reality: Personal boundaries are often crossed
The study found that students’ nuanced – and often critical – attitudes toward AI also shape how they use it. Ten percent of respondents reject AI use entirely. Another 18 percent manage to use AI responsibly, for example as a “sparring partner” for exchanging ideas – including critical engagement with the AI-delivered result –, or as a “tutor” offering methodological support. Yet despite these reservations, 72 percent are tempted – particularly while reading – to let AI take over entire tasks in a “ghost” role, raising questions about individual authorship and contribution. The reasons are varied: time pressure, anxiety about grades, motivational struggles, difficulties with academic writing, and a desire for dialogue, information, and inspiration all intersect with the design of AI chatbots, which encourage increasingly intensive use. One student summarized this tension in a free-text response: “Sometimes I have to actively motivate myself to think independently instead of handing the task over to AI.”

The takeaway: Universities need to act now
The current situation still offers a favorable starting point, Nora Hoffmann says, adding that “many students learned to write before AI became widespread and still have a sense of what is lost when too many tasks and responsibilities are delegated to artificial intelligence. But that is likely to change.” The study therefore recommends that universities:

• reflect at the disciplinary level on the role of writing, reading, and AI within specific academic cultures, including by defining which literacy skills remain indispensable in the age of AI, which may reasonably be delegated, and which AI-related competencies students should develop;
• expand AI literacy initiatives to include emotional and motivational dimensions, rather than focusing primarily on technical aspects and applications;
• establish subject- and discipline-specific guidelines for AI use in academic writing; and
• strengthen support for academic reading and writing skills. In addition to the cross-disciplinary resources offered by university writing centers, this topic should be more deeply integrated into subject-specific teaching in order to help students experience the writing process as a learning process and to strengthen their social integration. Until now, the focus has been predominantly on the final written product as a form of assessment.

The study makes clear that AI has already become part of academic writing, but that its use remains complex and contested. Universities now face the challenge of helping students use AI in ways that support – rather than undermine – their own intellectual contribution. “The goal is not to ban AI,” Nora Hoffmann concludes, “but to enable students to use it critically and thoughtfully.”


Publication:
Hoffmann, N./Göbel, H./Rütten, D./Schmidt, S. (2026): “Manchmal muss ich mich aktiv zum selbst denken motivieren” [Sometimes I have to actively motivate myself to think independently]. Report on the 2025 Student Survey on the Use of AI for Academic Writing. Schreibzentrum der Goethe-Universität Frankfurt a. M. https://doi.org/10.21248/gups.97289