AI at university by field

Every field has its own uses for AI and its own traps. A medical student needs clinical cases to practise on, a law student a case analysis and an architecture student a concept. Here you'll find how universities handle AI, prompts for every field and a page for each field. General studying (required reading, flashcards, final exams, thesis defence) is in the guide Students. As of 26 October 2026.

🎓 Fields

Each page has prompts for learning, practice and exams, and at the top a safety net: what AI must not do in that field and where to check things.

NewClinical cases · pharmacology · patients

Medicine

Clinical cases to practise diagnosis, pharmacology from your own notes, practising conversations with patients and mock final exams.

Open →
NewCases · species · owners

Veterinary medicine

A clinical case of an animal, comparing species, talking to owners, zoonoses and food hygiene. Drug doses never from AI.

Open →
NewCases · exams · judgments

Law

Analysing a case step by step, a mock oral exam, comparing legal concepts and reading judgments. Statutes always in their current version.

Open →
NewConcept · research · crits

Architecture

Concept and moodboard, visualisation from a sketch, research into reference buildings and a mock crit. Codes and regulations checked.

Open →
NewCode · maths · testing

Engineering and IT

Code you have to be able to explain, finding a bug with hints, checking your working and verifying a result with a script.

Open →
NewCase studies · data · models

Economics and management

Case studies, a business plan, economic models, Excel and regression. Numbers only from sources, not from AI.

Open →
NewLab · statistics · papers

Natural sciences

A lab report, choosing a statistical test, a graph from a paper and designing an experiment. Lab procedures never from AI.

Open →
NewTexts · sources · research

Humanities and social sciences

Analysing a source, an opponent for your argument, theory made clear, a questionnaire and coding an interview. Quotes only from the source.

Open →
NewLessons · assessment · placements

Education

Planning a lesson, a task for pupils of different abilities, assessment criteria, reflecting on a placement and talking to a parent.

Open →

⚖️ Before you start: university rules

Universities mostly don't ban AI. They want you to use it openly and for the work to stay yours. Rules differ, though, so it's always your school's rules that count. The examples here come from Czech universities.

📜 Your school sets the rules

  • Rules differ by university, faculty and even course. What counts is the assignment from your teacher and your supervisor. If you can't find the rules anywhere, ask before you use AI.
  • Charles University's recommendation for theses has three levels. Just mention it in your declaration: ideas, outline, grammar, basic research. Discuss with your supervisor: text generated from your notes, rewording your own text, analysing literature, code. Don't: the whole thesis, the introduction and conclusion or a literature review written by AI.
  • AI has no place in exams, tests and closed-book written exams unless the teacher explicitly allows it. That's how it works at the Czech Technical University and the Faculty of Law at Charles University, for example.

✍️ Declare your use of AI

  • A thesis includes a declaration: which tool, what for and to what extent. Faculties have templates for it, for example Charles University's Faculty of Humanities since 15 April 2026. Charles University has a declaration generator at ai.cuni.cz.
  • Undeclared use of AI that replaces your own work can be a disciplinary offence. According to the Faculty of Law at Charles University, though, permitted use of AI on its own is no reason for a lower grade.
  • The results and conclusions of your work must be yours. The Czech Technical University adds that you must be able to reach the same result without AI.

🔎 You're responsible for the mistakes

  • AI makes up citations, authors, statutes and numbers, and sounds confident doing it. You're responsible for everything in your work, including mistakes made by AI. Charles University and the Czech Technical University both say so.
  • Look up and read every source AI recommends. Don't copy citations from AI; the Czech Technical University explicitly advises against it.
  • Your school can ask how you worked with AI and even for the materials you uploaded to it. The Faculty of Law at Charles University does, for example.

🔒 What doesn't belong in AI

  • Personal and sensitive data: patients, clients, animal owners, people in your research, classmates.
  • Internal documents from placements and internships, unpublished research data and data from contract research.
  • If your school gives you AI through a school account, use that one. Charles University offers Microsoft Copilot to all students and allows protected data only in approved tools, such as Copilot signed in with a university account.

Based on Charles University's recommendations on AI in theses and on safe tools (pages updated in June 2026), the recommendation of the Faculty of Law at Charles University (7 October 2024), the framework rules of the Czech Technical University (in force since 19 February 2024) and Masaryk University's statement. Your school may have different rules, and its rules are the ones that apply.

📋 Prompts for every field

They work in ChatGPT, Claude, Gemini, Grok and in Copilot with a school account. Replace the yellow parts in square brackets with your own. When you work with your own course notes, Gemini Notebook (formerly NotebookLM) is a good fit too: it answers only from the materials you upload and cites them.

A semester plan from the syllabus ChatGPTClaudeCopilot
<syllabus>
[paste the course syllabus or a list of topics and dates]
</syllabus>

I'm studying [field, year]. Turn this syllabus into a week-by-week plan for the semester.
- For each week: the topic, what to read or review, and one question I should be able to answer afterwards.
- Put assessments, deadlines and exams in the plan as fixed points. Plan a review in the week before each of them.
- I have [6] hours a week for self-study.
Work only from the syllabus. If something is missing from it (dates, scope, reading), tell me what to ask the teacher.
An academic paper in a foreign language ClaudeGemini NotebookGemini

Helps you understand the paper, not read it for you.

<paper>
[paste the text of the paper or upload the PDF]
</paper>

I'm studying [field] and I have to read this paper for a seminar. Help me understand it:
1. What the paper is about, in 3 sentences in [my language].
2. A glossary: 10–15 technical terms from the paper, each with its equivalent in my language and a short explanation.
3. The authors' main claims and what they support them with. For each one, say where it is in the text (section or page).
4. Five questions to check my understanding. Show me the answers only when I ask for them.
Answer only from the pasted text. Don't fill in what isn't in the paper; write "not in the text".
A mock oral exam from your own materials ClaudeChatGPTGemini Notebook

With voice mode in ChatGPT, Gemini or Grok you can even take the exam out loud.

<materials>
[paste your notes, course notes or the list of exam topics]
</materials>

You're the examiner at an oral exam in [course]. Examine me on the topic [topic].
- Ask one question at a time and wait for my answer.
- Follow up on each answer the way an examiner does who wants to find out whether I understand it or just know it by heart.
- After five questions, grade me: what I knew, what was missing and what to review. For each gap, say where it is in the materials.
Ask only about what's in the materials. If you're not sure about something, say so and don't grade me on it.
An AI use declaration for your thesis ClaudeChatGPTCopilot
Help me write a declaration on the use of AI for my [bachelor's thesis / master's thesis / seminar paper].

My faculty's template or instructions:
[paste the template text, if you have it]

What I used AI for:
- tool: [ChatGPT / Claude / Gemini / Grok / Copilot / Gemini Notebook], what for: [outline, grammar check, research, translating the abstract, code], where in the work: [chapter]

Write the declaration factually and truthfully, only from what I've listed. Don't add anything and don't play anything down.
If the template asks for something I haven't given, write [to fill in] there.
At the end, list the uses I should discuss with my supervisor beforehand because they affect the content of the work (such as rewording text or analysing literature).
Checking a reference list ClaudeChatGPTGemini
<references>
[paste your reference list]
</references>

Check this reference list against the citation style [ISO 690 / APA 7 / my faculty's guidelines].
- For each entry, say whether it has all the required details and whether the format is right.
- Don't make up missing details (year, pages, DOI, publisher). Just say what's missing and I'll look it up.
- Flag entries that look suspicious (an unknown journal, a year that doesn't match the volume) so I can check that they exist.
Put the result in a table: entry number, what's wrong, how to fix it.