FRAMEWORK
The instructor must specify the expectations regarding the use of AI in the syllabus at the beginning of the semester. For the sake of clarity and to help students navigate the system, instructors must define and communicate their decisions regarding the use of generative AI according to the categories in the framework below, for which templates and additional aids are available in this website.
Category | Meaning | When to choose? |
|---|---|---|
🟢 GREEN | Supported / Integrated | If the use of AI is part of the curriculum and the goal is to develop critical tool use. |
🟡 YELLOW | Restricted / Conditional | If AI is only allowed for auxiliary processes (drafting, grammar), but not for content writing. |
🔴 RED | Prohibited | If the goal is to measure basic competencies, where the use of AI would circumvent the learning objectives. |
We recommend using or adapting the following template messages as needed in the syllabus (see Appendix 1):
Green (Supported / Integrated Use)
The use of generative artificial intelligence (AI) is permitted and encouraged in this course. You may use tools (e.g., ChatGPT, Copilot, Gemini, DALL-E) for brainstorming, drafting, content generation, or coding. Students bear sole responsibility for the accuracy of the generated content. Hallucination is not an exonerating circumstance.
Yellow (Restricted / Conditional Use)
The use of generative AI is permitted to a limited extent within the course. The tools may only be used in the preparatory stages of the work (e.g., brainstorming, structure planning, language editing). It is PROHIBITED to use the tool to write the substantive text content of the assignment or to completely replace coding.
Red (Prohibited use)
The use of generative AI tools (text, code, or image generation) is NOT permitted within the framework of the course. The goal of the course is to measure basic competencies that require independent work. Submitting AI-generated content is considered academic misconduct. (Exception: The prohibition DOES NOT apply to simple, non-generative spell checkers (e.g., Word's built-in tool, Grammarly) and reference managers (e.g., Zotero, Mendeley).)
APPLICABLE DOCUMENTATION REQUIREMENTS
When defining the requirements for a course, instructors may compile the mandatory documentation requirements for students for each task related to the course from the elements below (see Appendix 2).
The purpose of the documentation is to support the cognitive effort essential to the learning process and to allow the instructor to focus on the product, the learning process, and, where appropriate, the evaluation of AI use. You can choose from the following options and combine them as you wish. The options can be freely adapted to the learning outcomes of the course, and new options can also be incorporated.
1. Documentation of prompts (instructions) (select the appropriate level of expectation for your course):
Minimum: Only copy the initial prompt (the first instruction) verbatim. (Goal: To check the topic).
Selective: The initial prompt + a brief summary of additional instructions that mark a turning point. (E.g., "Here I asked you to switch to academic style"). (Goal: To follow the logical arc).
Complete: Link to the entire conversation or exported file as an attachment. (Purpose: Complete archiving and verifiability).
- Note 1: Although links are practical, they are not recommended for archiving purposes, as the conversation may be deleted later.
- Note 2: In order to manage the workload of teachers, it may be worth introducing a system of checking only a few uploads at random. This way, everyone is still required to provide documentation, but the teacher does not have to review everything item by item.
2. Process documentation (the supported and restricted MI usage requirements can be supplemented by selecting from the following elements):
Reflective assessment: A short reflection in which the student explains how they worked with AI, what they accepted and rejected from the content generated by AI, and why. It is advisable to support reflection with guidelines and a set of criteria in line with the course objectives. (Objective: Assessment of critical thinking).
Track changes: Ask for the submission of the assignment on a platform suitable for tracking (e.g. a Word document shared on OneDrive), where the editing process can be seen based on the version history. Where appropriate, the student may be asked to distinguish between AI and human content by colour coding or annotation. If the nature of the task allows (e.g. correction or further development of AI-generated content), you may request a version of the assignment with track changes, which clearly shows what the student has changed in the material compared to the original starting point.
Before-and-after snapshot: A variation on track changes. Ask for the attachment of two screenshots/code snippets showing the raw, generated material and the final version refined by the student. (Goal: To make the learning process visible.)
Fact-checking log: Depending on the nature of the task, you can ask for a table in which the student must list 3-5 statements generated by the AI and add the real sources found by the student that confirm or refute the given statement. (Objective: To strengthen accountability and critical thinking).
3. Verification requirements (the following elements can be used to reinforce/ensure the integrity and quality of learning):
Oral exam, report: If the focus of the assessment is on the process, an oral exam/report provides an opportunity to actually measure and demonstrate learning outcomes (if the course framework, e.g. number of students, allows this).
Spot checks/assessments: If the framework of the course does not allow for a comprehensive oral examination, it is possible to include spot checks. Students must be prepared to defend the content of their assignments orally on a spot-check basis. It is important to draw students' attention to this possibility at the beginning of the semester, as specified in the syllabus. Serious inconsistencies between the documentation and the oral response are considered conclusive evidence in accordance with the dean's instructions.
Some specific examples
Example 1: Preparing a lesson plan and worksheet for a differentiated group of students. 🟢 Supported/Integrated use
"The course is designed to develop the digital competences of future teachers, therefore the use of generative AI (e.g. ChatGPT, MagicSchool AI) is permitted and encouraged for the planning of lesson plans and differentiated tasks. Students are required to use AI for idea generation (e.g. icebreaker games, project ideas), but they must critically review and revise the generated material in line with pedagogical objectives and age-specific characteristics."
Required documentation:
Documentation of prompts (selective): Please copy the key instructions you used to request differentiation (e.g., "create a task for SNI students") from the tool.
Process documentation (reflective evaluation): In addition to the lesson plan to be submitted, please provide a one-page reflection answering the following questions: Which details were generated by the AI? Why did you accept/reject this particular version? How did you modify the material to make it pedagogically authentic?
Example 2: Writing an analytical essay or literature review. 🟡 Restricted/conditional use
"The use of generative AI tools is permitted to a limited extent when writing the seminar paper. The tools may be used to structure the literature, assist in translating foreign-language sources, or proofread the text (spelling, style). However, it is strictly PROHIBITED to use these tools to write the substantive part of the analysis, the professional conclusions, or to generate non-existent references. In order to preserve professional ethics and academic integrity, the text must reflect the student's own train of thought."
Required documentation:
Process documentation:
- Fact-checking log: Since AI often "hallucinates" literature, at the end of the submission, the student must verify the authenticity of the three most important statements or references suggested by the AI and incorporated into the thesis in a table (by providing a doi link or bibliographic data).
- Change tracking: Please prepare your paper in a cloud-based document (Google Docs/OneDrive), where the version history proves the writing process. It is important that the paper is prepared on this interface and that you do not copy the written text from your own text editor.
Example 3: Literature research and systematic processing Tool focus: Research-supporting, source-based AI (e.g. Elicit, Scite, Consensus) 🟢 Supported/integrated use
"In order to effectively explore the literature, the course explicitly supports the use of research-supporting AI tools (e.g. Elicit, Consensus, Scite) that work with real sources. These tools (as opposed to general language models (e.g. ChatGPT)) minimise hallucination as they work from existing databases (Semantic Scholar, PubMed). During the semester, these tools can be used to refine research questions, quickly find and filter relevant studies, and generate comparative analyses (e.g., methodology, sample size, comparison of results). Although the tools filter well, reading and interpreting the full text of the selected key studies (approx. 10-15 basic works) is unavoidable. The 1-paragraph summary generated by the tool cannot be cited in the thesis without the student's analytical work."
Required documentation:
Process documentation (exported matrix): Please attach the comparison table generated by Elicit (or another tool) (.csv or .pdf export) as an appendix to your submission, which contains the list of articles found and the filtering criteria. This verifies your extensive research.
Verification log (Interpretation check): The student must select 3 studies recommended by Elicit and compare the AI summary with reality in a short table (e.g. AI statement: "The study found a significant relationship between X and Y." | Correction after careful reading: "The relationship is indeed significant, BUT the authors emphasise that this is only true for the female sub-sample, which the MI summary did not mention." (Goal: To demonstrate that the student understands the difference between a "superficial" MI abstract and scientific deep drilling. The goal of verification here is to develop source criticism and attention to detail in line with the learning outcome).
Example 4: Compiling a scenario and communication plan for a fictional sporting event. 🟢 Supported/Integrated Use
"In line with labour market expectations, the use of generative AI tools (e.g. text writing and image generation applications) is supported during the course to increase efficiency in the preparation of project plans and marketing materials. The aim is for students to be able to use AI as an assistant in developing the creative concept for the event."
Required documentation:
Process documentation (before and after snapshot): Students must attach the "raw" marketing text or event script generated by AI and add their own final version optimised for the Hungarian sports market (e.g. legislation, local conditions).
Verification (random defence): During the practical class, the student must be able to justify the key figures of the budget or schedule verbally, regardless of whether they were calculated with the help of AI.
Example 5: Case study solution, plan for a simulated consulting conversation, or writing a self-reflection. 🔴 Prohibited use
"In this course, the use of generative AI tools for text generation, case analysis, or reflective journaling is NOT permitted. Counselling work is based on personal presence, empathy, and situation-sensitive, independent thinking, which AI cannot replace. The aim of the course is to measure these human competencies."
Required documentation / Assurance:
Verification (Oral exam/report): The student acknowledges that the written case analysis/reflection will be followed by an oral report. If the student is unable to provide an analysis or self-reflection in the oral situation that is commensurate with the standard of the submitted material, this may result in the submission being invalidated. (Exception: Use of spell checkers is permitted).
Example 6: Writing an end-of-semester analytical paper or essay. 🟡 Restricted/conditional use
Goal: To make students' work visible, to support group learning and to validate independent work in a "trust-based" environment. In the case of large groups, where it is not possible to conduct oral examinations for all students.
"The use of generative AI tools (e.g. ChatGPT) is permitted to a limited extent in this course. The tools can be used for brainstorming, drafting and proofreading the text, but students must develop the substantive ideas in their papers independently.
At the end of the semester, during the last class of the term, we will hold a joint knowledge-sharing session. On this occasion, we will randomly select 5-10 students from among the authors of the submitted papers, who will briefly (in 3-5 minutes) present their topics and most important conclusions to the group in a moderated discussion. The purpose of the draw is to share the diverse topics covered in the course. At the same time, we would like to point out that if the selected student is unable to make a meaningful statement about the content of their paper during this professional discussion, this will demonstrate a lack of independent work and result in the rejection of the paper.
Required documentation:
Verification requirement (random presentation): Personal attendance at the final knowledge-sharing session is a prerequisite for completing the course. Students must be prepared to summarise their work orally and answer questions from their peers or the instructor in the event of a draw. This method ensures that verification is also a useful learning event for the community.
Process documentation (change tracking): For security and subsequent verifiability, please prepare your thesis on a platform that supports version tracking (e.g. OneDrive, Google Docs). This background material will only be checked in case of dispute (e.g. if the oral presentation is unsuccessful).