How to use artificial intelligence to study: practical guide 2026 (15 prompts + 5 workflows)

TL;DR — Updated 22 June 2026: yes, AI can help you study in 4 concrete ways: (1) summarise your material (PDFs, notes, slides), (2) generate multiple-choice quizzes for active practice, (3) simulate the oral exam with follow-up, (4) build concept maps and flashcards for spaced repetition. This guide gives you 15 copy-paste prompts by subject (3 prompts × 5 subjects: medicine, law, economics, engineering, languages), 5 real workflows (week before exam, train revision, 24h prep, semester catch-up, thesis) and 7 mistakes to avoid. The golden rule: always start from your real material, use prompts with 4 fixed ingredients, and measure results in weeks, not minutes.

The 4 concrete ways AI helps university study

AI in 2026 is not a magic wand: it is an accelerator for 4 specific activities. Understanding which they are and when to use them makes the difference between a student who wastes hours in vague chats and one who prepares an exam in half the time.

1. Summarising uploaded material. You upload a 200-page PDF, slides, notes. AI returns an operational summary, a schema, a concept map. It is the most immediate use and the one with the highest ROI for university study, because the time saved in the synthesis phase is reinvested in active practice. The tools that in 2026 do this work best are NotebookLM free (Google) for synthesis from sources, ChatGPT free and Gemini free for quick summaries, and AiLearn360 free for a structured summary in university-schema style.

2. Generating multiple-choice quizzes. AI generates relevant quizzes from your material: 10-20 questions per chapter, with the correct answer, common trap, page reference. It is active practice, not passive reading. The effect: in 2 weeks, the topic coverage triples compared to just re-reading. The best tools for quizzes from uploaded material are AiLearn360 free (integrated workflow), Knowt free (automatic from slides), Quizlet free (community library). The limit of automatically generated quizzes is variable quality: AI can generate questions that are too easy or misleading, they must always be double-checked.

3. Simulating the oral exam. AI asks the first question like a professor, you answer (by voice or chat), AI corrects, asks the next question, maintains the examiner's personality. It is the best ally for those with oral-exam anxiety, because practising in a safe environment breaks the fear. The tools that do this well in 2026 are AiLearn360 free (tutor with multiple personalities), StudierAI free (vertical on oral), and secondarily ChatGPT voice with a specific prompt. The limit: free simulation has less nuanced personalities, and the follow-up on wrong answers is less structured.

4. Building concept maps and flashcards. AI turns a chapter into a tree map (3 levels, 5 nodes per level, links) or into 20-30 flashcards with a sharp question on the front and a synthetic answer on the back. These artefacts are perfect for spaced repetition: you use them on Anki, on Quizlet, or on AiLearn360, and review them at intervals of days to move knowledge from short-term to long-term memory. The limit: automatically generated maps must always be reviewed, because AI can flatten the real complexity of the material.

The practical rule: one AI per activity, one activity at a time. Do not start from everything at once. Pick a single chapter of your course, try one activity (e.g. quiz), measure the result, expand it.

15 copy-paste prompts by subject

The prompts below are tested on ChatGPT free, Gemini free, AiLearn360 free, NotebookLM free. They work best if adapted to your case: replace the parts in square brackets with your real data.

Medicine (3 prompts)

Prompt 1 — Clinical chapter summary

"I uploaded the PDF [title] of chapter [X] of [clinical subject]. Give me a clinical summary in 8 bullets, each with: pathology or key concept, pathogenetic mechanism in 1 line, main clinical manifestation, first-choice diagnostic exam, first-line therapy. Max 4 lines per bullet. Highlight pathology names in bold. Format: ready to paste into notes."

Prompt 2 — Progressive clinical quiz

"Generate 12 exam questions on chapter [X] of [subject]. Format: 4 answer options. Difficulty: 4 basic questions (definitions, classifications), 5 clinical questions (short patient case + question), 3 difficult questions (mechanisms, differential diagnosis). For each question indicate the correct answer, a common trap students fall into, and the reference to the page of the uploaded PDF."

Prompt 3 — Clinical oral simulation

"You are a university professor of [clinical subject] examining me for the oral exam. Start with a short clinical case (3 lines: patient, symptoms, exams). Ask me the first question on diagnostic reasoning. Wait for my answer, correct me in 2 lines, ask the next question. Maintain a strict but constructive tone. Ask 5 questions total. At the end give me a grade from 18 to 30 and a summary of weak points."

Law (3 prompts)

Prompt 1 — Legal institution schema

"Explain to me the institution [name: e.g. 'termination for breach of contract'] of the [code: e.g. 'civil code']. Format: definition in 1 sentence, 3 operational prerequisites, 2 legal effects, 1 practical case in 5 lines, 3 relevant Supreme Court cases (references only, no invented numbers). At the end ask me 2 verification questions."

Prompt 2 — Constitutional / civil / criminal law quiz

"Generate 10 exam questions on [branch of law: e.g. 'constitutional law'], chapter [X]. Format: 4 answer options. Mix: 4 theoretical questions (definitions, principles), 4 applied questions (short practical case + question), 2 case-law questions (Supreme Court orientation). For each question indicate the correct answer, the trap, and the normative or jurisprudential reference."

Prompt 3 — Legal oral simulation

"You are a university professor of [branch of law] examining me for the oral exam. Start with a practical case of 4 lines. Ask me the first question on legal qualification. Wait for my answer, evaluate the legal reasoning in 2 lines, ask the next question. Formal tone, attentive to technical vocabulary. Ask 5 questions. At the end give me a grade from 18 to 30."

Economics (3 prompts)

Prompt 1 — Economic model synthesis

"Explain to me the model [name: e.g. 'IS-LM model'] as if I were 20 years old. Start from an analogy of everyday life (3 lines), then formalise in 3 points with the essential equations, then ask 2 verification questions. Max 250 total words. Format: ready to paste into a note."

Prompt 2 — Microeconomics / macroeconomics quiz

"Generate 12 exam questions on chapter [X] of [subject: e.g. 'microeconomics']. Format: 4 answer options. Mix: 5 calculation questions (with simple numbers and guided solution), 4 theoretical questions (definitions, theorems), 3 economic policy questions (practical implications). For each question indicate the correct answer, the trap, the chapter reference."

Prompt 3 — Solved economics exercise

"I have an exercise in [subject: e.g. 'economic policy'] on [topic: e.g. 'monopoly market equilibrium']. Solve it step by step: (1) identify the problem data, (2) set up the equations, (3) carry out the steps showing every formula, (4) verify consistency of units of measure, (5) comment on the result in 2 lines. Format: structured document ready to study."

Engineering (3 prompts)

Prompt 1 — Theoretical explanation + application

"Explain to me [concept: e.g. 'Laplace transform'] starting from the physical intuition (2 lines), then formalise in 3 points with the mathematical definitions, then show a worked example of medium difficulty. At the end ask me 2 verification questions. Max 300 total words."

Prompt 2 — Technical quiz with calculations

"Generate 10 exam questions on chapter [X] of [subject: e.g. 'structural mechanics']. Format: 4 answer options. 5 questions require explicit calculations (with simple numbers and guided 4-5 step solution), 3 are theoretical, 2 are applied (real engineering case). For each question indicate the answer, the trap, the reference."

Prompt 3 — Technical oral simulation

"You are a university professor of [engineering subject] examining me for the oral exam. Start with a technical problem (5 lines: data, schema, constraints). Ask me to set up the solution. Wait for my answer, correct any conceptual errors in 2 lines, ask the next deepening question. Technical tone, attentive to formalism. Ask 5 questions. At the end give me a grade."

Languages (3 prompts)

Prompt 1 — Comprehension exercise

"I have a text in [language: e.g. 'English'] of [N] words. Ask me 5 comprehension questions: 2 on explicit content, 2 on inferences and vocabulary, 1 on reformulation in my language in max 50 words. For each question indicate the exact answer and the reference to the paragraph of the text. Format: ready-to-use sheet for a B2 student."

Prompt 2 — Grammar / vocabulary quiz

"Generate 15 questions in [language] on the topic [e.g. 'conditionals in English']. Mix: 5 grammar questions (choose the correct form), 5 vocabulary questions (synonyms, collocations, false friends), 3 translation questions (sentence native language → target language), 2 guided production questions (complete the sentence with a given word). For each question indicate the answer and an explanation in 1 line."

Prompt 3 — Simulated conversation in language

"I want to practise [language]. You are a [interlocutor: e.g. 'receptionist of a Madrid hotel / shop assistant in a Berlin bookshop / waiter in a Paris restaurant']. Start the conversation with the greeting and the first question. Wait for my answers, correct grammatical errors in 1 line, go on. Maintain a natural tone, suited to the context. Make 10 total exchanges. At the end give me overall feedback on the 3 most recurring mistakes."

Practical tip: the difference between a prompt that works and one that disappoints is almost always in the 4 fixed ingredients: expected output format, course context, material reference, difficulty level. The more the prompt knows what you are preparing, the more the output will be usable for your specific exam.

5 real workflows for studying with AI

A workflow is a repeatable sequence of actions. Here are 5 tested workflows for real situations of the university student.

Workflow 1 — Week before the exam (focus: synthesis + active practice)

When to use: 7-10 days before the exam, when you have already read the material at least once.

Sequence:

1. Upload all course PDFs to NotebookLM free or AiLearn360 free.
2. Generate an operational summary for each chapter (format: 7 bullets, 1 line each).
3. Create 15 multiple-choice quizzes per chapter with AiLearn360 free or Knowt free.
4. Do all the quizzes in spaced-repetition mode: question → answer → if wrong, review the bullet.
5. Identify the 5 most difficult concepts, run an oral simulation on them with AiLearn360 free or ChatGPT voice.
6. Last day: only flashcard review, no new material.

Estimated time: 3-4 hours per day for 7 days, total 21-28 hours.

Expected outcome: 90% topic coverage, ability to take the written exam and basis for the oral.

Workflow 2 — Train revision (focus: flashcards + audio overview)

When to use: during commutes, 20-60 minutes per day, for 2-4 weeks.

Sequence:

1. Create 30-50 flashcards from course material (Anki free, Quizlet free, or AiLearn360 free).
2. Download NotebookLM free audio overviews (5-15 minute podcast-style summaries).
3. Listen to audio overviews on the train or while walking.
4. Do 10-15 flashcards per day on Anki or Quizlet (spaced-repetition mode).
5. When a flashcard is wrong 3 times in a row, review the original chapter.

Estimated time: 20-60 minutes per day, total 8-20 hours over the period.

Expected outcome: long-term memory consolidation, topic refresh in dead time.

Workflow 3 — 24h preparation (focus: intelligent cramming)

When to use: in the 24 hours before the exam, when time is very short.

Sequence:

1. Upload the 3-5 most important PDFs to NotebookLM free or ChatGPT free.
2. Ask for an ultra-compressed summary: 5 bullets per chapter, max 3 lines each.
3. Ask for 20 likely exam questions (format: question + answer in 1 line).
4. Do a quick oral simulation round with ChatGPT voice (3 questions in 10 minutes).
5. Re-read only the wrong quiz answers.
6. Sleep at least 6 hours: memory consolidates during sleep.

Estimated time: 4-6 effective hours, plus sleep.

Expected outcome: minimum coping for a light exam, not suitable for complex exams.

Workflow 4 — Semester catch-up (focus: rebuilding from scratch)

When to use: mid-semester, when you realise you are behind.

Sequence:

1. Take inventory: how many chapters/exams are missing, how much time you realistically have.
2. Identify the 20% of topics worth 80% of the grade (ask a classmate or AI: "out of 10 chapters of [subject], which 2 are the most important for the exam?").
3. Start from those 2 chapters, apply workflow 1 to each (summary + quiz + oral).
4. Only after mastering them, expand to the other chapters.
5. Review the 2 key chapters every week until the exam.

Estimated time: 5-8 hours per week for 4-6 weeks.

Expected outcome: semester catch-up with the minimum necessary effort.

Workflow 5 — Thesis (focus: support to writing, not replacement)

When to use: during the thesis drafting, month by month.

Sequence:

1. Upload the bibliography (article PDFs, book chapters) to NotebookLM free or AiLearn360 free.
2. Use AI to schematise each article (format: thesis, method, results, limits, 1 citation sentence).
3. Ask AI to identify recurring themes across articles (format: concept map).
4. Write the thesis chapters yourself: AI can help reformulate paragraphs, but the final text must be yours.
5. Use AI to simulate the discussion: have 10 likely commission questions generated for you.
6. In the thesis declaration or in a note, declare the use of AI as a work tool.

Estimated time: variable, roughly 2-4 hours per week for 4-6 months.

Expected outcome: thesis written in a more structured way, discussion prepared, intellectual originality preserved.

Important note: in Italy, using AI to produce text to present as your own in a thesis is considered a form of plagiarism, punishable. The correct approach is to declare the use of AI as support (review, schematisation, simulation), not as author.

Mistakes to avoid: 7 things AI gets wrong if you do not guide it well

AI is a powerful tool, but it is not magic. Here are the 7 most common mistakes that university students make when they start using it.

1. Vague prompts without context. Asking "explain philosophy to me" instead of "explain to me the difference between natural law and legal positivism in 3 points with examples from chapter 4 of [title]". Without context, AI responds generically and is hardly usable.

2. Not uploading real material. AI knows Wikipedia, not your textbook. Without an uploaded PDF, AI responds with clichés. With an uploaded PDF, AI responds with your material.

3. Expecting perfect answers on the first try. First prompts generate mediocre output. The practice of rewriting the prompt 2-3 times with more context is the norm, not the exception.

4. Using AI as a shortcut to not read. AI summarises, but if you do not read the summary you do not understand. The summary is an accelerator, not a substitute for active reading.

5. Relying on output without verification. AI makes mistakes, especially on specific data, citations, recent case law, complex formulas. Always verify with the original source.

6. Not repeating active practice. Doing a quiz once and not redoing it is useless. Active practice works through spaced repetition, not a single session.

7. Mixing too many tools without mastering one. Using ChatGPT + Gemini + Copilot + Perplexity + AiLearn360 + NotebookLM all at once is a dispersion. Pick 2-3 tools, learn them well, expand gradually.

When NOT to use AI to study

AI is not suitable for everything. There are contexts in which using it is counterproductive, and it is fair to declare them honestly.

When AI does NOT help you:

• First reading of a complex text. AI summarises, but deep comprehension requires your active reading, your marginal notes, your doubts. Asking AI for a summary before you have read is a shortcut that does not work.

• Solving unseen exercises. AI can help you understand the method, but not replace practice on exercises. Doing 20 exercises by yourself is worth more than 100 exercises solved by AI.

• Writing the final thesis. AI can help reformulate, but the final writing must be yours. In Italy, presenting AI-generated text as your own is considered plagiarism. The correct approach is to declare the use of AI as support, not as author.

• Decisions on study path or career. AI can inform, but decisions on what to study, where to do Erasmus, which thesis to write are personal, contextual and value-laden decisions. AI cannot make them for you.

• Severe performance anxiety. AI is a support, not a psychologist. If exam anxiety is disabling, a professional should be consulted (psychologist, psychotherapist, university counselling service).

• Verification of sensitive information (health, law, finance). AI can confuse outdated or invented information. For important decisions on health, regulations, investments, always rely on an updated human professional.

The golden rule: AI is an accelerator for active practice, not a substitute for real studying. Use it for summaries, quizzes, oral simulation, maps. Do not use it to read for you, write for you, or think for you.

Related subject hubs

If your exam concerns a specific subject, these vertical guides start from your real case:

• AI Tutor for studying: complete guide to the vertical AI tutor for university study
• Quiz generator from PDF: if your priority is turning notes into multiple-choice quizzes
• Medicine oral exam simulator: specific focus for clinical medicine oral exams
• Private law oral exam simulator: for those preparing the civil law oral
• Economics oral exam simulator: for the micro and macro oral
• TOLC AI simulator: if you are preparing for a university admission test
• Best AI for students 2026: reasoned map of the 7 most useful platforms in 2026

Who wrote this guide

This guide was written by the AiLearn360 editorial team on 22 June 2026. The editorial team includes educators, instructional design engineers and specialists in AI applied to university learning. The qualitative assessments reflect internal tests, comparative benchmarks and feedback from the user base. For reports or contributions: [email protected].

Editorial disclaimer — Practical guide + workflow version (22 June 2026): this practical guide was prepared by the AiLearn360 team for informational and promotional purposes. The qualitative assessments and workflows described derive from internal tests conducted in June 2026, public-feature analysis of the cited tools and comparative benchmarks. For the general definition of artificial intelligence, see Wikipedia — Artificial intelligence. For the methodological framework on spaced repetition, see Wikipedia — Spaced repetition. For the European regulatory framework on AI, see EU AI Act — European regulation on artificial intelligence. For the international education framework, see OECD Library — Education at a Glance. Features, prices and availability of the cited tools may change over time: before subscribing, verify on the provider's official site.