AI is a good partner for discussion and for checking an argument. But it makes up quotes, authors and dates, and the interpretation has to be yours. Prompts for texts and sources, theory, research and writing. As of 26 October 2026.
How universities handle AI and what doesn't belong in it in general is on the university page. Here's what applies on top of that in the humanities and social sciences.
AI asks you first and then fills in, pointing to the place in the text.
<text> [paste the text or source] </text> You're a seminar tutor in [history / philosophy / literature]. Analyse this text with me, but ask me first: 1. Who wrote the text, for whom and with what aim? 2. What's the main thesis and how does the author support it? 3. What does the text leave out or keep quiet about? After each of my answers, add what I overlooked and point to the specific place in the text. Work only from the pasted text. Mark any historical context you add yourself as "check".
I need the context for [a source from 1848 / a work by Karel Čapek]. Make me a timeline of the main events and connections in the years [1840–1850]. For every date and name, say how sure you are. If you're not sure about something, write "check". Don't include quotes.
AI as an opponent for your essay and a patient explainer of theories. You read the original.
PD2V Web <essay>
[paste your essay or argument]
</essay>
You're the opponent at a seminar. Find the weak points in my argument: logical fallacies, unsupported claims, hidden assumptions and counterarguments I don't deal with. For each one, say where it is in the text.
Don't rewrite the text. Assess the arguments, not the style.
Explain the concept of [discourse in Foucault / habitus in Bourdieu / performativity in Butler] to me:
1. Simply, with an everyday example.
2. How it differs from the everyday meaning of the word.
3. In which works the author develops it. Only works you know for sure, without page numbers.
4. What the main criticisms are.
Don't give verbatim quotes, I'll find them in the original.
<notes> [paste your notes or text from the textbook] </notes> Compare [Durkheim and Weber / structuralism and post-structuralism] in a table: main questions, key concepts, method and criticism. Under the table, add 3 questions examiners use to test the differences. Work from the notes. Mark anything you add yourself as [added by AI].
A questionnaire and the first coding of an interview. You make the research decisions.
<questionnaire> [paste your draft questions] </questionnaire> Check my questionnaire for research on [topic] among [target group]. For each question, say whether it's leading, double-barrelled (asks about two things at once), unclear or sensitive, and suggest better wording. Then say whether the questions cover my research questions: [paste your research questions].
Only with an anonymised transcript, and only when the respondents' consent and your faculty's rules allow it.
<transcript> [paste an anonymised interview transcript: no names, places or details that could identify the respondent] </transcript> I'm doing qualitative research on [topic]. Suggest initial codes for this transcript. For each code, give the verbatim passage from the transcript it's based on. Only suggest codes, I'll decide on them. Don't add anything that isn't in the transcript.
Paraphrasing without plagiarism and a mock oral exam.
<original> [paste the original text with its source details] </original> <paraphrase> [paste your paraphrase] </paraphrase> Compare my paraphrase with the original: Has it kept the meaning? Is it too close to the original (the same words, the same sentence structure), so it should really be a quote? What's missing from the reference to the source? Don't rewrite the paraphrase for me, just say what to change.
<materials> [paste your notes or the exam topics] </materials> Quiz me on [the history of sociology] like an oral exam. Ask a question and after my answer follow up on the connections: who builds on whom and who argues with whom. After 6 questions, sum up what I'm missing. Ask only about what's in the materials.