This paper presents an educational use-case for creating assessment materials with Chat-GPT 3.5 turbo in the domain of software engineering. We attempt to (a) alter specific aspects of a generic domain model (UML class diagram) and (b) generate requirement texts (problem scenarios) that creatively describe the previously altered model. The challenge that comes with these two approaches is (1) to change only the variables contained in the generic domain model in such a way, that the resulting domain model makes sense and (2) to generate natural language requirement texts (scenarios) that creatively and correctly describe the underlying domain model. Results: The first approach (a) yielded satisfactory results. After providing Chat-GPT with examples, it was able to alter all variables in a coherent way according to the topic specified in the prompt. This result indicates potential for other educational use cases and ongoing software engineering research with LLMs. Chat-GPT’s generated requirements texts were useful but required caution from the lecturers before handing them out to students. We reflected on the contribution of this paper (and papers similar to this) and argued, that the provided value lies not only in the quality of the yielded results but much rather in describing educational use cases that are worth following up on with future versions of large language models.