Guidelines for the thesis preparation using generative AI

The following recommendations should be followed in preparing dissertations using generative artificial intelligence (GSI):

General principles

  1. The submitted dissertation must allow verification of the competencies acquired by the student, in particular, the ability to analyse and infer independently and should include collaboration with the student over the academic year, specifically, regular progress presentations in the thesis preparation.

Acceptability of the use in the thesis

  1. Before using content-generating systems within the thesis, the student should consult the objectives, modalities and scope of their use with the supervisor.
  2. The reasons, methods and scope of use of content-generating systems shall be verified by the academic supervisor at the thesis acceptance stage, considering the criteria under Article 76 of the Act of 20 July 2018. Law on Higher Education and Science, particularly the independence criterion in its preparation and the student's ability to analyse and infer independently.
  3. Permitted forms of use, with the prior approval of the academic supervisor, may include, in particular, editing or proofreading a text written independently (i.e. without adding new content); searching for information on a given topic and conversation to enhance understanding of a given issue; generating code snippets that are part of a lengthier script.
  4. Unauthorised forms of use include, in particular, copying and pasting content generated by artificial intelligence systems without appropriately marking the source of the content; using GSI systems in a way that makes it unable to assess the competencies acquired by the student and the independence of the submitted work; using GSI systems incompatible with the arrangements with the academic supervisor and/or the rules adopted for the subject.

Labelling the use of GSI systems

  1. The thesis author should describe in the introduction to the thesis the content-generating systems used and 1) the reasons for their use (e.g. enhanced understanding of the topic, translation from a foreign language, time-saving, breaking creative blockage), 2) how they were used (e.g. idea generation, language editing, generation of abstract/bibliography/abbreviation list, preparation of illustrations, generation of code snippets), and 3) the scopes of such use (e.g. which uses covered the whole work and which covered only parts of it).
  2. The paper's author should mark all fragments of the paper (i.e. sentences, paragraphs, images, tables, etc.) created by using content-generating systems, mainly by indicating the tools used, the commands used ("prompts") and the output obtained (e.g. in footnotes, an appendix to the paper or other equivalent form).
  3. In case of doubt in explaining and/or indicating the use of SI systems, the author should consult the supervisor.

Handling suspected undisclosed use of content-generating systems in the thesis

  1. In the case of suspected use of content-generating systems in a manner inconsistent with these guidelines, the academic supervisor should request, specifically, the student to clarify the purposes, means and scope of use of such systems in the thesis.

Guidelines for Thesis Reviewers

  1. Reviewers should be aware that the use of generative AI in diploma theses is allowed following Order No. 426/2023/2024 of the Rector of the Adam Mickiewicz University, Poznań from 20 March 2024 on the principles of using artificial intelligence systems in the process of student education and should not be perceived as unfavourable, specifically should not affect the final grade of the thesis, provided that it has been used after prior agreement with the supervisor, and has been correctly applied and appropriately marked.
  2. Reviewers should carefully assess the extent to which the thesis reflects the author's work and analysis and not just the quality of the content generated by GSI. Individual attention should be paid to sections of the thesis marked as being supported by generative AI to ensure an appropriate degree of independence has been demonstrated by the author, notably in terms of analysis and inference.
  3. Reviewers should verify that the author has adhered to these guidelines, particularly in terms of (1) adequately explaining the reasons, methods and extent of the use of the GSI in the introduction to the paper; (2) appropriately labelling all content generated using the SI.