Ethical Guidelines on the Use of AI
The use of AI is subject to the special ethical responsibility of the user.
Particularly in the context of studies and research, scientific integrity requires that AI systems are only used in a supportive manner and that the user's own scientific contribution remains clearly recognisable. The following principles apply:
Transparency
The use of AI tools must be declared in the manuscript. This serves to maintain trust and traceability for the readers and corresponds to the terms of use of the respective tools.
The chapter on the effects of generative AI on the design of the course and the assessment of examination results provides further information on the scope and structure of disclosure.
Personal responsibility
AI-generated content must not be adopted without scrutiny. It must always be critically reviewed, revised and categorized in the respective scientific context.
The authors are fully responsible for the content.
Authorship
AI tools cannot be named as authors.
Contributions of AI systems must be disclosed in the manuscript - e.g. in a separate section "Declaration on Generative AI" - and can be named as an aid in the references if necessary.
The following overview comprises typical use cases to show which forms of AI use are considered acceptable in academic work and which practices must be avoided:
| Use case | Acceptable use of AI | Non-reasonable use of AI |
|---|---|---|
| Text creation | AI can help to overcome writer's block or suggest definitions and wording alternatives. However, its use must always be accompanied by critical thinking and personal judgement in order to ensure the autonomy of the service provision. | The generation of new texts (entire paragraphs or chapters) by AI and their adoption in academic work is not permitted. Scientific texts must be written independently and clearly attributable to human authors. |
| Translation | AI can be used to support the translation of texts into another language and thus facilitate the understanding of foreign-language literature. This only applies if learning the language is not a competence objective of the course or part of an examination | The use of AI to translate a previously published work into English raises ethical concerns with regard to self-plagiarism. If the language competence is part of a competence objective or an examination achievement, the use of AI is not justifiable. |
| Linguistic revision | AI can be used to make spelling, grammar or minor stylistic improvements in order to increase comprehensibility and readability. | Revising entire paragraphs or chapters using AI without critical review weakens the academic output and undermines the author's voice and thoughts and can perpetuate biases from training data. |
| Image generation | The use of AI for image generation is permitted if the work itself deals with the generation of images as a research topic. | The use of AI to create scientific figures (e.g., diagrams, illustrations) is generally not permitted, as the results cannot be independently verified for accuracy and appropriateness. Furthermore, modifications or adjustments in the sense of an interactive process, as is possible with generated texts, are not feasible; figures can only be recreated from scratch. |
| Rephrasing | AI can help to reformulate individual sentences in order to improve clarity, conciseness or expression. This supports linguistic fine-tuning | The systematic reformulation of entire passages by AI without human post-editing carries the risk of losing one’s individual writing style and uncritically adopting biases present in the training data (see section on linguistic revision above). |
Based on CEUR-WS GenAI Policy (external link, opens in a new window).