Guidelines for using generative AI in teaching at SDQ

Violation of the guidelines can be treated as an attempt to cheat and lead to failing an examination (5.0).  

The current guidelines of the Faculty of Computer Science apply to all theses written at SDQ: https://www.informatik.kit.edu/faq-wiki/doku.php?id=generative_ki

Bachelor and Master Theses

As stated in the faculty guidelines, the use of AI should be documented, for which we distinguish three cases:

1. Spelling, Grammar, Text Style and Translation
   • Usage of AI for checking or correcting the listed aspects of a text does not need to be documented
2. Text or Image Generation
   • Students must document the generation of content that is included in the thesis
   • Documentation should include the used system, language model and, if applicable, prompt engineering technique
   • Documentation of individual prompts and results is not necessary
   • The preferred place for the documentation is in a separate paragraph on the same page as the signed declaration (Selbständigkeitserklärung)
3. Part of the Research
   • AI can be used in various parts of the research, e.g., literature search, code generation or data analysis, as long as you adhere to the faculty guidelines
   • As with other tools, usage of AI tools needs to be documented to ensure transparency and reproducability
   • Documentation should include the used system, language model, prompt engineering technique and, in the appendix or repository, individual prompts and results
   • The documentation can, e.g., be part of a Used Tools or Evaluation Design section

As stated in the faculty guidelines:

In all cases, students remain responsible for their work. This also applies to the parts of their work that have been created using or influenced by AI.

Best Practices in using Generative AI in Scientific Writing

• When using generative AI for formulating text, review all test and especially any generated adjectives and check whether this is really what you want to say.
  • Make sure to remove all chatty, fluffy, subjective and/or unnecessary adjectives and adverbs. Some generative AIs tend to fill in words like "extensive", "intricate", "rich", "meticulously", "comprehensive", "pivotal", "significant", "valuable" that often are used in a subjective way (unsachlich) and unsubstantiated. Do not claim that e.g. your overview is "comprehensive" if you do not provide any evidence for it actually being comprehensive. Instead, leave the assessment to the reader.
  • Other words are commonly generated by AI and often sound inappropriate in scientific texts, such as "delve" or "elucidate". Use words that you yourself are comfortable with.

Guidelines for “Prüfungsleistungen anderer Art“, such as the final tasks of “Programmieren”

• For graded alternative assessments (Prüfungsleistungen anderer Art), guidelines will be defined at the level of the individual course. If the use of generative AI is allowed in a course, the use must be documented as described for final theses above.

• For the lecture Programming, AI-generated artifacts, such as code, are considered disallowed aids (unerlaubte Hilfsmittel). You can chat with generative AI assistants as a tutor to try to better understand a general topic about programming (e.g., ask it to explain Polymorphism) or questions about the Java language. For more guidelines, please also refer to the section “Verwendung generativer KI” in the slides “00.3_Organisatorisch”. All submitted artifacts must be written by the student or created by simple completion mechanisms of IDEs. If in doubt, ask in the ILIAS forum.