Documentation of the use of AI tools

In principle, as with other scientific sources, the responsibility for using AI-generated content (e.g., texts, images, programming code) lies with the respective user (student or teacher). It is their duty to independently assess and verify the quality of the content, as any potential errors cannot be attributed to the information source.

How do you document the use of AI tools when writing an assignment?
This website offers a decision-making guide with links to detailed information on various documentation tools.

documentation options

In addition to referencing AI-generated results as a source, documenting the use of AI-based tools is one way of making the AI-based support transparent and traceable. Since requirements and conditions may differ depending on the subject, course and learning objective, several options for documenting the use of AI-based applications are presented below, each with a different focus. The relevant requirements should always be clearly formulated, easily accessible and transparently communicated to students.

Keyword list

If only a limited amount of information needs to be documented, this can be done using simple lists of keywords. If, for example, only a single AI tool is approved for use in the context of a specific service provision, or if only the tool used for a non-exam-related task and the purpose for which it was used is to be indicated, a bulleted list may suffice.

Example 1: Bulleted list of the steps in which a particular AI chat system was used

  • to generate ideas,
  • to improve linguistic formulations,
  • to generate examples or for translation.

Example 2: List of AI-based tools including the purpose of use

  • elicit.com for literature research,
  • semanticscholar.org for research,
  • gamma.app for presentation creation or
  • stablediffusionweb.com for image generation.

Various documentation tools

In order to document a larger amount of information, tables can be used that are geared towards different requirements, such as tools, work phases or reflection.

In collaboration with universities in Lower Saxony, a document has been developed that contains descriptions and examples.

Der folgende Überblick soll als Entscheidungshilfe für die Auswahl dienen. Die entsprechenden Vorlagen für die konkreten Umsetzungen sind in der Tabelle direkt verlinkt. 

InstrumentsRequirements AdvantagesLimitations
List with only one approved tool List work stepsimple, quick to create and to overview
(students and lecturers)
Since only one tool is allowed: none
list with intended useUnsorted listing of all tools including intended useSimple, quick to create
(students)
Sequence not adaptable
(lecturers)
Holistic documentation Reflection on use in a paragraph Few specifications
(students and lecturers)
Requires dealing with indeterminacy (students
and lecturers)
Tool-oriented documentation (only in German)Tools are in the focusEasy to create (students); easy to read (lecturers)Gives little indication of
concrete use (teachers)
Work phase-oriented documentation (only in German)

Tool use in the work process

should be documented

Quick overview 
(Lecturers)
Requires intensive
examination of
the table and the assignments in advance of
completion (students)
Reflection-oriented documentation: Knowledge matrix for AI skills acquisition (only in German)
 
Documentation requirements are specifically defined and 
conveyed to students in the form of questions to students
Documentation purposes
and requirements are specifically adapted (lecturers)
Planning effort in advance (lecturers)
Reflection-oriented documentation: Handout, reflection diary and documentation (zip folder) (only in German)Tool use in the scientific work process is fully documented, reflected (diary) and prepared for furtherprepared for further use. Building AI literacy
(students) and metacognitive knowledge about one's own scientific action (lecturers)
must be guided
(lecturers);
time-consuming (students)