Artificial intelligence
Binding Guidelines on the Use of AI for Students in Study Programmes Guaranteed by the Department of Algebra
(translated from the Czech original by Claude Opus 5)
Generative artificial intelligence is developing very quickly, so we think it is important to set out basic rules for working with these tools in teaching. The rules are binding for students and for the teaching staff who work with them.
1. The general framework is set by the recommendations of Charles University: https://ai.cuni.cz/AIEN-42.html
2. We consider the individual judgement and conscience of each of us, student and teacher alike, to be essential, since probably nobody knows how things will develop even in the near future. These rules are not meant to restrict any meaningful use of AI tools in advance.
Rules for theses (bachelor’s, master’s, doctoral):
3. Whether and to what extent the student will use artificial intelligence is for the student and the supervisor to agree on. The student must agree on these rules with the supervisor before starting to use AI (this also applies to the uses listed in point 5 below). It is advisable to include brief rules on the use or non-use of AI in the official thesis assignment.
4. Any use of artificial intelligence in preparing the thesis must be explicitly declared. If AI was not used at all, the thesis should also say so explicitly. One suitable way to do this is the university tool available here: https://ai.cuni.cz/AIEN-39.html
5. Unless the student and supervisor agree otherwise, we consider the following uses of AI appropriate by default: language editing of the text, searching the research literature, and consulting on mathematical concepts and results that appear in the literature and are not the direct subject of the thesis. AI may also be used, to a reasonable extent, to check the text from a mathematical point of view.
6. Without the supervisor’s explicit consent, AI must not be used to generate mathematical proofs or text. Where it is used this way, the supervisor will decide how it should be presented (*). When presenting mathematical results obtained with AI, we consider it appropriate to describe the methodology in detail, for example by publishing the entire conversation with the AI as an appendix to the thesis.
7. As before, the assessment of a thesis will include an evaluation of the student’s own contribution. To take an extreme example, even a solution to a famous open problem obtained through a short conversation with an AI does not make a defensible thesis.
(*) Examples:
- short passages as direct quotations in quotation marks;
- a longer machine-generated text, even one that was substantially edited, with an explicit description, e.g. “Claude Opus 5 wrote the first draft of Section 3.2, which the author then lightly rephrased; this mainly involved revising and completing the proof of Lemma 3.4.” or “Claude Opus 5 pointed out an error in Section 4 and suggested a fix, which became the basis for the proof of Theorem 10.”
Rules for other homework:
8. The lecturer or tutorial instructor will set out explicitly the rules for using AI in a given course and assignment, preferably directly in SIS or wherever else the course completion requirements are described. The rules in point 5 can serve as a starting point. Within these rules, the use of AI does not need to be declared further.
Remarks:
- We recommend that students and staff follow the same rules when writing research papers or teaching materials. In particular, we consider it essential to declare explicitly how, and to what extent, AI was used.
- Please be aware of “watermarking” of AI-generated text, meaning cryptographic marking that makes it possible to identify the text’s origin even after it has been disguised by light editing: https://scottaaronson.blog/?p=10032
Useful resources on the use of AI in mathematics
Charles University information and rules: https://ai.cuni.cz/AIEN-42.html
(and other pages on the same site)
Declarations:
- Statement by 25 Fields Medal laureates: https://mathandai.org/
- The Leiden Declaration: https://leidendeclaration.ai/
Interesting blogs on AI (and more):
