KI in der Testentwicklung: Zur Nützlichkeit von LLMs bei der Erstellung von Prüfungstexten
AI in test development: On the usefulness of LLMs for the creation of exam texts
Journal article › Research › Peer reviewed
Publication data
| By | Anja Peters, Anastasia Drackert, Andrea Horbach |
| Original language | German |
| Published in | Informationen Deutsch als Fremdsprache, 53(1) |
| Pages | 47-59 |
| DOI/Link | https://doi.org/10.1515/infodaf-2026-0005 |
| Publication status | Published – 02.2026 |
| Keywords | academic language skills, readability assessment, reading comprehension, TestDaF exam, preparatory materials |
In order to determine the potential of generative AI for the development of input texts for use in a test for German as a foreign language, we evaluated reading passages created by trained writers against GPT-generated texts. The results of the analysis show that the AI-generated texts provide a valuable starting point for the creation of exam texts, but that adjustments are necessary. Finally, we discuss possible applications of the research results to language teaching.
