Centre for Teaching and Learning (ZLL)
On this page, you will find information and guidelines on how you and your students can cite and document the use of AI technologies in academic writing and research. We also offer a comprehensive FAQ on the topic of artificial intelligence, as well as further, carefully curated information and materials that are continuously updated.
The information on this page is intended as a suggestion and a guide. We support you in deciding how the use of AI in your modules—especially regarding text production—should be declared and documented.
Our recommendation: Seek an active dialogue with your students. Proactively communicate how AI may and may not be used in your courses.
On these pages, you will find practical tips, guidance, and ready-to-use text templates to assist you.
Since generative AI technologies have automated text production and are changing common writing and reading practices in the academic sphere, the question of how to disclose and document AI usage in scientific writing and research processes has become a topic of discussion.
While certain trends are emerging in terms of how to disclose and document AI usage in academic writing, there is still no universal agreement across disciplines on the best approach. Instead, there are currently only guidelines specific to individual fields.
The following information is not intended to be a definitive or binding set of rules. Instead, it aims to provide a starting point for discussion between teachers and students, help reduce uncertainty, and offer a framework for implementing AI disclosure and documentation in academic writing.
Here you will find a compilation of materials relating to the labeling and documentation of AI use academic writing.
The debate about disclosure, documentation and labeling options for AI use in academic writing is in full swing. We have selected interesting resources, which can help you and your students in dealing with the topic.
The possibilities for using generative AI (GenAI) are vast - yet students are often uncertain about its appropriate use. Orientation creates a sense of security - and as instructors, you can offer your students this framework:
Clearly explain how generative AI systems (GenAI) will be used and handled in your course—ideally during the first or second session of the semester.
If you provide a syllabus, include this explanation there as well.
You will be setting up an AULIS course or group for your class. Please include your guidelines on AI usage in your course space. You may use the example wording provided below.
GenAI has become a part of our everyday and professional lives. We therefore encourage you to actively and purposefully integrate AI into your teaching to help foster your students' digital literacy. The focus should be on the responsible and reflective use of AI. However, there will also be learning goals that can only be achieved through direct (i.e., non-AI-supported) work. In such cases, it may be appropriate to restrict the use of AI to ensure students can reach these goals independently.
Below, we present four approaches for integrating GenAI into your teaching. Choose the model that best aligns with the learning objectives of your course to support the development of students' skills in working competently with future-oriented technologies. Each approach is briefly explained and supplemented with suggested wording for your syllabus or AULIS course.
Here you will find various examples of how major professional associations, publishers, and research funding institutions formulate their guidelines for disclosure and documentation of AI use, and thereby define what is considered scientific integrity within their framework and what is not accepted.
IEEE is a worldwide professional association for professions and scientists in the field of electrical engineering and information technology.
APA is the American Psychological Association. The APA reference style is widely used across many disciplines. Current information and changes can be found in the APA Style Blog.
Springer - a large publishing house with 2,900 magazines and 300,000 books.
Statement by the Presidency of the German Research Foundation (DFG) on the impact of generative models for text and image generation on the sciences and the DFG's funding activities, as of September 2023.
In the discourse surrounding Artificial Intelligence (AI), many terms are in circulation; in the academic world, for example, the term Generative Artificial Intelligence (GenAI) has become established for language and image generation models. However, some of the established terms are misleading, yet they are familiar to all interested parties and stakeholders. On these pages, we therefore adopt a pragmatic approach: To ensure that our services and materials are easy to find and understand, we use the terminology commonly employed within the higher education sector. At the same time, we make a point of being as clear as possible in our communication to avoid any misunderstandings.
Even the concept of ‘intelligence’ in artificial intelligence is misleading, because behind the common language models (such as ChatGPT) and image generation models (such as DALL-E) lie models that perform mathematical calculations using vast amounts of computing power, on the basis of which they generate their output. They are able to mimic natural language based on the size of the training data. However, these processes are not comparable to human intelligence. The term GenAI emphasises the generative function of the models. Depending on the model, these can typically generate text, audio and/or visual content (images, videos). This term is therefore more precise than simply AI.
However, when naming models, we recommend thinking in terms of functions in even greater detail. This approach makes it clear which tasks the respective models can perform, thereby counteracting the ‘black box’ effect – that is, the lack of understanding of how the models work. Possible functions could include, for example, text generation, text-to-image, text-to-speech, speech-to-text, etc.. Modern chat interfaces, such as ChatGPT or Claude, typically combine text generation (text-to-text), image-to-text and speech-to-text functions. Identifying these specific functions can also help us when considering language models in academic work, as it enables us to decide more precisely which processes we wish to replace with models and which we wish to retain.