Handout on the Use of Generative AI in Academic Studies and Teaching
Heidelberg University of Jewish Studies (HfJS) - Guidance on the Use of Generative AI in Academic Studies and Teaching
Guidance on the Use of Generative AI in Academic Studies and Teaching
Heidelberg University of Jewish Studies (HfJS) — Version 2026
Generative artificial intelligence (AI) opens up new possibilities for teaching, learning, and research. It supports academic work, can provide individualized learning support, and can make knowledge more accessible. At the same time, its use raises questions about academic integrity, transparency, data protection, copyright, and the quality of academic work.
The University of Jewish Studies Heidelberg (HfJS) views generative AI as a tool to support academic work. This guide aims to enable students and teaching staff to use generative AI competently, reflectively, and responsibly. Its central aim is to promote intellectual autonomy. This encompasses the ability to address scholarly questions independently, evaluate information critically, form well-reasoned judgments, and use AI systems deliberately as supporting tools. Intellectual autonomy also requires a willingness to address scholarly questions independently, even when generative AI appears to offer simple solutions. Academic education requires the ability to tolerate uncertainty, form one’s own judgments, and critically assess and reflect on scholarly findings. Generative AI is therefore intended to provide deliberate support for academic learning processes, rather than to shorten them.
This guide offers a framework for the responsible use of generative AI at the HfJS. It explains the legal, organizational, and pedagogical context, provides practical recommendations for teaching, learning, and examinations, and encourages a transparent and reflective approach to AI. As technical capabilities and legal requirements continue to evolve, the guide will be reviewed regularly and updated as needed.
Contents
0. Six Basic Principles for the Use of AI at the HfJS. 2
1. General Principles and Responsibilities. 2
2. Data Protection, Law, and Sustainability. 3
3. Use of Generative AI in Jewish Studies. 4
4. Guidelines for Assessments (for Students). 5
5. Assessing Student Work and Suspected Cheating (for Teaching Staff). 7
6. Practical Prompt Examples for Academic Work (for Students). 7
0. Six Basic Principles for the Use of AI at the HfJS
1. Generative AI supports academic work but does not replace independent thinking or scholarly judgment.
2. Users bear full responsibility for all content.
3. AI-generated content must be critically checked and supported by verifiable evidence.
4. AI use must be documented transparently in accordance with the requirements of this guide.
5. Data protection and confidentiality take precedence.
6. Core academic competencies are also practiced without AI.
1. General Principles and Responsibilities
Is the use of generative AI permitted at the HfJS?
Permissibility: Generative AI use is generally permitted, provided it is consistent with the course, the examination regulations, and the assignment requirements. Users must comply with applicable legal requirements, including the EU AI Act, copyright law, and the General Data Protection Regulation (GDPR).
Use Guided by Learning Objectives: The fact that generative AI is generally permitted does not mean that every academic task is best delegated to AI systems. The learning objectives and academic competencies to be acquired determine which steps must be completed independently and at which stages AI can provide support. Generative AI is intended to support academic work, not to replace independent thinking, scholarly judgment, or responsibility for the academic content. Reflective use of generative AI expands the possibilities for academic work, but requires core academic competencies to be acquired independently.
Who is liable for errors and AI-generated content?
Personal Responsibility and Duty to Verify: Users retain full responsibility for the content and legal compliance of AI-generated material. AI models can produce information that sounds plausible but is inaccurate (“hallucinations”), systematic distortions (“biases”), and fabricated source references or bibliographic citations. All outputs must therefore be checked against reliable sources before use.
No Authorship: AI systems have no legal personality and cannot claim authorship or co-authorship of scholarly work.
2. Data Protection, Law, and Sustainability
What data protection and copyright requirements apply?
Data Protection and Provider Selection: Preference is to be given to services that comply with data protection requirements (e.g., YoKI). For other providers, users must first check whether data transfers, storage locations, and model training comply with the GDPR and university requirements.
Copyright: AI-generated content may contain protected material. Incorporating such content without acknowledging it may therefore result in unintentional plagiarism.
Data Restrictions: Data requiring protection (e.g., personal or sensitive third-party data, protected primary sources, student work, or unpublished third-party texts) must not be uploaded to external AI systems. Users may enter their own draft texts into external AI systems only if it is ensured that the content will not be used to train AI models.
What requirements apply to inclusion and digital participation?
Inclusion and Accessibility: AI-assisted teaching and learning opportunities, as well as automatically generated content (e.g., summaries or transcripts), should be as accessible as possible and usable with assistive technologies.
What is meant by environmental sustainability?
Resource-Efficient Use: Generative AI consumes substantial energy and computing resources. Use AI deliberately and only where it adds academic or pedagogical value. For simple tasks, more resource-efficient alternatives, such as search engines or local spelling checkers, are sufficient.
3. Use of Generative AI in Jewish Studies
How can generative AI support scholarly work?
To counteract a loss of skills (de-skilling), core academic skills should not be delegated to AI systems without reflection. The University of Jewish Studies Heidelberg instead promotes the deliberate expansion of academic competencies (new-skilling) through reflective use of generative AI as a partner in learning and thinking.
Generative AI can support academic work, from literature searches and revisions to the wording of one’s own texts to examination preparation. It can offer particular pedagogical value in structuring complex topics, comparing scholarly positions, and developing precise research questions. As a supporting tool, however, it replaces neither independent thinking nor scholarly judgment.
Acquiring academic competencies requires certain learning and work processes to be undertaken deliberately without AI support. In particular, users should regularly practice developing their own research questions, analyzing and interpreting sources, constructing academic arguments, and writing key passages independently. Only on this basis can AI-generated results be placed in their disciplinary context, evaluated critically, and used responsibly.
This includes learning phases without AI, in which core academic competencies are regularly practiced without support from generative AI. These phases provide the foundation for reflective and responsible AI use.
What are the specific limitations in Jewish Studies?
Gaps in the Data: Generative AI relies on large amounts of readily available digital data. The insights it can offer are often limited when dealing with rare, specialized, or non-digitized sources of Jewish literature and history.
High Risk of Hallucinations in Source Languages: Source texts in Hebrew, Aramaic, or Yiddish, especially texts without vowel pointing or rabbinic texts, present an increased risk of incorrect translations and fabricated source references and bibliographic citations.
The Need for Scholarly Source Criticism: Interpreting religious, historical, or philological sources requires scholarly scrutiny grounded in subject expertise. AI-generated information must therefore never replace independent analysis of primary and secondary sources.
4. Guidelines for Assessments (for Students)
Is AI use permitted to the same extent in all forms of assessment?
Assessments serve to demonstrate academic competencies and students’ own scholarly contributions. The permitted extent of AI use therefore depends on the intended learning outcomes of the assessment. What matters is not AI use or non-use as such, but whether students themselves perform the work and demonstrate the competencies being assessed.
Subject to legal requirements and the HfJS AI Guidelines, the permitted extent of AI use depends on the form of assessment.
Unsupervised written assessments (e.g., term papers, theses): Generative AI use is generally permitted, subject to the requirements set out in this chapter. Assessment is based on demonstrated academic competencies, particularly analysis, synthesis, critical reflection, and scholarly judgment, rather than the mere reproduction of knowledge.
Supervised examinations (e.g., written examinations, oral examinations): Generative AI may be used only if this is expressly provided for in the assessment instructions and access is technically enabled during the examination. Where the main purpose is to test knowledge, generative AI use is generally prohibited.
How must I disclose AI use in my work?
The following requirements apply to AI use in assessed work:
Declaration of Independent Work: Students must confirm in writing whether they used AI and, if so, within what permitted scope, using the HfJS Declaration of Academic Integrity (Antiplagiatserklärung).
Documentation: A table in the appendix must record the AI system used, the model version, the date of access, and the purpose and nature of its use. A complete record of all prompts is not required. A clear record of the tasks for which AI was used is sufficient (see the record of the use of AI-based electronic tools required by the Declaration of Academic Integrity).
Citing AI-Generated Text: AI-generated passages incorporated verbatim or in paraphrase must also be acknowledged at the relevant point in the work by naming the system used, as with a citation.
What is prohibited when using AI?
The following forms of AI use are incompatible with academic integrity:
Complete Outsourcing: Using AI to produce assessed work without any independent academic contribution (e.g., generating an entire term paper for immediate submission) is prohibited and constitutes an attempt to cheat.
Use Without Verification: AI-generated content must not be incorporated without the student’s own subject-specific checks and academic scrutiny.
Undisclosed AI Use: Using generative AI in breach of the applicable documentation or disclosure requirements is not permitted and may be treated as an attempt to cheat under the examination regulations.
Does using AI, or choosing not to use it, affect my grade?
Voluntary Use: Generative AI use is voluntary. Where AI use is intended or necessary for an assignment, instructors offer equivalent alternatives for students who do not wish to use AI for ethical, data protection, or financial reasons.
Equal Treatment: Neither permitted and documented AI use nor a deliberate decision not to use AI may affect the grade.
5. Assessing Student Work and Suspected Cheating (for Teaching Staff)
Instructors design learning activities and assessments to promote academic independence, critical reflection, and subject-specific judgment. They also act as academic mentors, helping students to reflect on academic work processes, evaluate generative AI use critically, and gradually develop academic independence. Personal feedback and academic dialogue are gaining further importance compared with assessment based solely on the finished work.
What principles apply to assessment and suspected cheating?
Clear Requirements for AI Use: Instructors should state transparently, no later than the start of the course (e.g., in the syllabus or module description), whether and to what extent generative AI may be used. Documentation and disclosure requirements must be clearly stated. Supervised and unsupervised assessments must be distinguished.
Limited Use of AI Detectors: Software tools for identifying AI-generated content are technically unreliable and therefore cannot be used as the sole evidence of impermissible AI use.
Clarification Through Dialogue: Where there are reasonable grounds for suspicion, the facts should be clarified in a structured discussion.
Further Procedure: If the matter cannot be fully clarified in that discussion, any further proceedings must follow the applicable examination regulations.
Human Responsibility for the Final Decision: Responsibility for assessing student work always lies with the teaching staff. AI may assist with grading, but does not replace either the academic review or the final decision. Critical scrutiny remains necessary when AI assists with grading in order to counteract automation bias.
6. Practical Prompt Examples for Academic Work (for Students)
The quality of AI-generated or AI-assisted content depends largely on the precision of the prompt and the user’s subsequent critical review. Such content does not, on its own, replace independent academic analysis or verification against reliable sources.
The following examples offer guidance and must be adapted to the research question, discipline, and context of the work.
Stage / Area of Use | Prompt Template |
| 1. Topic Selection and Research Question | “I am interested in [topic]. Suggest five research questions of scholarly relevance. For each question, explain why it is of scholarly interest and which gaps in the research it could address.” |
| 2. Literature Brainstorming | “What scholarly debates are there about [topic]? Map out the main research positions and name key authors. Explicitly flag any uncertain information.” |
| 3. Research Design | “I would like to investigate the following research question: [research question]. Which research methods would be suitable? Explain the advantages and disadvantages of each method.” |
| 4. Support with Reading Notes | “Explain the main points of this text. Distinguish between facts, interpretations, and arguments. Which claims seem particularly convincing? What assumptions or weaknesses do you identify?” |
| 5. Outline Review | “Critically review the following outline. What logical inconsistencies, gaps in content, or missing aspects do you identify in relation to my central claim? [outline]” |
| 6. Critical Reviewer | “Take on the role of a reviewer for an academic journal. Assess the structure, argument, clarity of reasoning, and academic quality of my text. Also pose critical follow-up questions.” |
| 7. Argument Review | “Review my argument for logical weaknesses, unsupported assumptions, or possible counterarguments. Suggest improvements without changing my position.” |
| 8. Fact-Checking and Source Verification | “Check the following statement against the sources provided: [sources]. Identify any contradictions, gaps in evidence, or problems of interpretation.” |
| 9. Academic Language | “Revise the wording of the following section to match the style of academic writing. Do not alter the argument or the quotations. Explain your changes in bullet points.” |
| 10. Writing an Abstract | “Write an abstract of no more than 200 words based on my text. Use only information from my text and do not add any new content.” |
| 11. Examination Preparation | “Create ten examination questions of varying difficulty for this chapter. Provide the answers only when asked.” |
| 12. Learning Partner | “Ask me questions one at a time about [topic]. After each answer, give me feedback on the subject matter and gradually explore the topic in greater depth.” |
| 13. Translating Academic Texts | “Translate the following academic text into German/English. Keep the specialist terminology and source references unchanged, and flag any uncertain translations.” |
| 14. Time and Work Planning | “I need to complete a term paper of [number] pages by [date]. Create a realistic work plan covering literature searches, outlining, writing, revision, and proofreading.” |
| 15. Reflecting on AI Use | “Which steps in my project should I carry out myself, and where can AI provide useful support? Justify your assessment with reference to good academic practice.” |
7. Advice and Training
The university sees itself as a social space for learning and discussion. Scholarly knowledge develops not only through individual writing, but also, crucially, through academic exchange, discussions, colloquia, seminars, and dialogue between teaching staff and students. Generative AI can support this scholarly exchange, but cannot replace it.
Legal Requirement for AI Literacy: The ability to develop a critical understanding of AI systems, use them responsibly, and assess their capabilities and limitations (AI literacy) is a key competency for the future of higher education. In accordance with Article 4 of Regulation (EU) 2024/1689 (EU AI Act), the University of Jewish Studies Heidelberg supports the acquisition of the necessary AI literacy.
AI literacy encompasses not only the appropriate use of generative AI systems, but also a basic understanding of how they work, critical evaluation of their outputs, a reflective approach to uncertainty, hallucinations, and biases, and the ability to integrate AI responsibly and transparently into academic work and learning. The aim is to promote academic practice in which generative AI supports independent inquiry without replacing scholarly judgment or critical thinking.
Training Opportunities: The heiSKILLS programs at Heidelberg University support students and teaching staff in continuously developing their AI competencies and using generative AI reflectively in teaching and learning.
Support Structures: The Office of the Dean of Studies and the Examinations Office at the HfJS provide advice on assessment design, permitted uses of generative AI, and how to handle suspected cheating.
Reflective use of generative AI enriches contemporary academic practice and opens up additional possibilities for teaching and learning at the University of Jewish Studies Heidelberg. Responsible use adds value by combining technical support with academic integrity, critical reflection, and independent judgment.
The University of Jewish Studies Heidelberg views responsible generative AI use as an ongoing process. It regularly reviews this guide in light of technical, legal, and pedagogical developments and continues to develop its support and training opportunities. The aim is to build lasting AI competencies among students and teaching staff and to promote a culture of academic integrity and intellectual autonomy.
- ^ AI systems that can autonomously generate new content (such as text, translations, programming code, or graphics) based on training data.
- ^ Heidelberg University’s AI platform, which provides various language models via the university’s own infrastructure.
- ^ https://ki-campus.org/lernangebote/kurse/generative-ki-fuer-abschlussarbeiten. erative-ki-fuer-abschlussarbeiten.