Abstract
-
The rapid proliferation of generative AI (genAI) tools has fundamentally changed the university environment, especially since the emergence of large language models such as ChatGPT. Unlike most studies, which focus on the perspective of students, this research examines the professional use of genAI by doctoral and postdoctoral students at Mittweida University of Applied Sciences, a university with a strong digital focus. It is based on qualitative focus group interviews within the NextGen programme and analyses how genAI is used as an ‘assistant’ in academic research, teaching and administration. NextGen is a qualification programme designed to support researchers on their academic career path to professorship at universities of applied sciences in Germany.
The results show that experienced researchers create individual workflows for writing, literature research, statistics, programming, translation and presentation design by combining various genAI tools such as ChatGPT, DeepL Write and Gemini. The AI-generated results are subject to rigorous validation by experts and iterative improvement, ensuring that academic responsibility remains with the user. AI is not seen as a substitute, but as a ‘personalised assistant’ whose results are critically questioned and reflectively integrated into scientific routines, revealing both its flexibility and its limitations.
Despite optimism about productivity potential, several challenges are emerging, including concerns about cognitive debt, risks to academic depth, unchecked efficiency expectations, and social impacts, including data transparency and misinformation. Of particular note, the academics surveyed see difficulties in teaching students to approach genAI as a tool of assistance only. They observed a tendency among students to accept AI-generated content without critical review, often lacking the necessary expertise to properly evaluate and improve AI outputs. Therefore, teaching fundamental subject knowledge remains crucial, with an emphasis on striking a didactic balance between technological skills and independent academic thinking.
Peer learning, particularly through institutionalised peer group sessions (PGS) and informal exchanges, has been identified as an important way to build practical genAI skills and support reflective evaluation of different tools. These collective formats facilitate familiarity with evolving digital technologies and promote the sustainable, responsible use of AI in science.
Both formal and informal peer-led training environments help researchers stay up to date, share expertise, and navigate rapid technological change. By institutionalising such formats, the NextGen programme at Mittweida University of Applied Sciences represents a model for knowledge exchange and the responsible integration of AI into higher education.