How to Use AI to Study in College Without Hurting Your Marks
By Eduvy Editorial · 25 September 2026 · AI & Education
Use AI as a tutor that makes you think, not as a ghostwriter that thinks for you. That one habit keeps you inside your college's rules and protects your exam marks. Below: the Indian rules, a traffic-light list of uses, prompts to copy, how to disclose and check AI, and a weekly exam routine.
Why the way you use AI matters
A 2025 experiment published in PNAS gave nearly 1,000 high-school students in Turkey GPT-4 tutors for maths practice: one like standard ChatGPT, one that gave hints, not answers.
- With the standard chatbot, practice grades rose 48%. But when AI was taken away for the exam, those students scored 17% lower than students who never had AI.
- With the hint-only tutor, practice grades rose 127%, and the exam penalty largely disappeared.
If AI does the work, you get better homework and worse exam marks. Every prompt below makes AI behave like the second tutor.
What Indian rules actually say (September 2026)
We found no UGC, AICTE or Ministry of Education regulation on how undergraduates may use generative AI. If a website quotes "UGC AI guidelines", look for the actual notice on ugc.gov.in.
The UGC plagiarism regulations, 2018
The main national rule is the UGC (Promotion of Academic Integrity and Prevention of Plagiarism in Higher Educational Institutions) Regulations, 2018, notified on 23 July 2018. It predates ChatGPT and does not mention AI. Its penalties cover masters and research theses, and publications by faculty, researchers and staff. The "script" it covers expressly excludes "assignments / term papers / project reports / course work / essays and answer scripts".
So your B.E., B.Sc or BBA assignments fall under your university's and college's own rules. But every institution must teach academic integrity as a compulsory UG and PG module and publish its plagiarism policy on its website. Read yours.
| Level | Similarity | Masters or research thesis (student) | Publication (faculty, researcher, staff) |
|---|---|---|---|
| Level 0 | Up to 10% | No penalty | No penalty |
| Level 1 | Above 10% to 40% | Submit a revised script within 6 months at most | Withdraw the manuscript |
| Level 2 | Above 40% to 60% | Barred from submitting a revised script for one year | Withdraw the manuscript, lose one annual increment, no new Master's, M.Phil. or Ph.D. students to supervise for two years |
| Level 3 | Above 60% | Registration for the programme cancelled | Withdraw the manuscript, lose two successive annual increments, no new students to supervise for three years |
Repeat offences are punished one level higher. Attributed quotations, references, bibliography and standard terms, symbols and equations are excluded from the count. And a low similarity score is not a clean chit: AI-written text can score low and still be misconduct if your teacher asked for your own work.
What Karnataka's technical university has done
A VTU circular of January 2026, reported by Deccan Herald, caps computer-generated content in Ph.D. theses at 20%: a thesis over the cap is returned, and repeated failure means rejection. It covers research scholars, not B.E. students, but shows the direction: measuring and capping AI use, not banning it.
We found no published generative AI policy from Bangalore University. Where the university is silent, your teacher's instructions and your college's code of conduct apply.
A sensible default
IIT Delhi's guidelines (committee report dated 30 October 2024) are worth copying: disclose AI help, take full responsibility for accuracy, and never paste personal or confidential data into AI tools. Before any graded task, ask your teacher in writing: Is AI allowed here? For what? How do I disclose it?
Green, amber, red: which uses are safe
Green is fine for almost any course. Amber needs your teacher's permission and a disclosure. Red is out unless your teacher explicitly allows it.
| Use | Light | Why |
|---|---|---|
| Explaining a concept you are stuck on, in simpler words or with an example | Green | You still do the learning |
| Quizzing you on a unit or making practice questions | Green | You answer; AI only asks and checks |
| Turning your own notes into flashcards | Green | The content is yours; AI only reformats it |
| Checking your working and pointing to the first error | Green | You fix the mistake yourself |
| Grammar polish on a draft you wrote | Amber | It changes the text you submit |
| Brainstorming topics or an outline for an assignment | Amber | Some ideas in your work now come from AI |
| Debugging code for a graded lab or project | Amber | Fine for practice; ask first for graded code |
| Translating your own writing between Kannada and English | Amber | Disclose it and check technical terms |
| Drafting the assignment, lab record or report text you submit | Red | The work is marked as yours |
| Generating references you have not read | Red | AI invents citations (see below) |
| Any use during a test, exam, viva or lab exam | Red | Exam malpractice under university rules |
| Making up lab readings, data or survey responses | Red | Fabrication, with or without AI |
Study prompts you can copy
Replace the square brackets, and paste your own syllabus or notes so the AI works from your course.
1. Quiz me on a unit
I am a [second-year B.E. Electronics] student. Here is Module [3] of my syllabus: [paste]. Quiz me one question at a time, from easy to hard. Wait for my answer before the next question. Tell me if I am right and explain any gap in two or three lines. At the end, list the topics I got wrong.
2. Walk me through a derivation, but make me try each step
Explain the derivation of [the EMF equation of a DC generator] one step at a time. After each step, stop and ask me what comes next. Do not show the next step until I have tried. If I am wrong, give a hint, not the answer. Use the notation from [textbook name].
3. Make VTU-style questions
Under VTU's 2022-scheme regulations, a three-hour theory SEE paper has ten 20-mark questions, two per module, each with up to three sub-questions; you answer one per module.
Here is Module [2] of my syllabus: [paste]. Here is a past question paper for this course: [paste]. Write two full questions in the same style, 20 marks each, with no more than three sub-questions and marks shown, mixing topics from this module. Include at least one numerical or application question. No answers until I ask.
4. Turn my notes into flashcards
Turn these notes into 20 flashcards in the format Q: ... A: .... Use only what is in my notes. If anything in them looks wrong or incomplete, list it separately at the end instead of correcting it silently.
5. Check my working without giving the answer
Here is my solution: [paste or upload a photo]. Do not solve it for me. Tell me the first line where my working goes wrong and whether the mistake is in the concept, formula, algebra or units. Give one hint, then let me try again.
6. Mark my answer like an examiner
Here is the question [paste] and my answer [paste], worth [10] marks. Mark it strictly, as a university examiner would, and explain where I lost marks. Then tell me the two changes that would gain the most marks. Do not rewrite it for me.
7. Find the repeat topics in past papers
Here are [four] past question papers for [course]: [paste]. List the topics that appeared more than once, grouped by module, with how often each appeared. Do not add topics that are not in these papers.
How to disclose AI use
Use your teacher's format if there is one. If not, add a short statement at the end of your submission:
AI use: I used [ChatGPT / Gemini / Claude] on [date] to [explain the concept of X / check grammar in sections 2 and 3]. I did not use it to write [the analysis, code or conclusions]. I checked its output against [textbook or lecture notes]. My prompts are available on request.
If you quote AI output in a report, cite it:
- APA (company, year, tool and version, description, link): OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat
- MLA (prompt, tool, model, company, date, shared link): "[Your prompt, shortened]" prompt. ChatGPT, model [name], OpenAI, [date], [shared chat link].
Keep your chat history until results are out. Never cite AI as the source of a fact: find the original source, read it, and cite that.
How to check what AI tells you
- References. A 2023 study in Scientific Reports found that 55% of citations produced by GPT-3.5, and 18% of those by GPT-4, were fabricated. Among the real ones, 43% and 24% had substantive errors in authors, titles, dates or pages. Look up every reference in Google Scholar or your library and read at least the abstract before citing it.
- Formulas and numericals. Check each formula and sign convention against your textbook, check units on both sides, plug in a case where you know the answer, and redo the arithmetic. In IIT Delhi's survey, students reported that AI tools are weak at mathematical and logical reasoning.
- Rules and dates. Exam patterns, fees and deadlines change. Check the current official notice.
- Health and pharmacy facts. Never learn a drug dose, interaction or clinical value from AI without checking your prescribed textbook or the current National Formulary of India.
A weekly AI study routine for semester exams
For 2 to 3 hours of self-study a day, one module a week. You think first; AI comes in after.
| Day | You do first (no AI) | Then AI does | Prompt |
|---|---|---|---|
| Monday | Read the first half of the module; write your own notes | Explains what you did not understand | 2 |
| Tuesday | Solve 4 to 6 problems on paper | Points to the first mistake in each | 5 |
| Wednesday | Make flashcards from Monday's notes; self-test | Turns the rest of your notes into flashcards | 4 |
| Thursday | One past-paper question, timed, closed book | Marks it like an examiner | 6 |
| Friday | Read the second half of the module; make notes | Quizzes you on the whole module | 1 |
| Saturday | Mock: two full questions in 72 minutes, phone away | Marks both, lists weak topics | 3, 6 |
| Sunday | Redo flashcards you got wrong; rest | Quizzes you on the week's weak topics | 1 |
Why 72 minutes: three hours for five questions is 36 minutes each. In the last two weeks, stop asking AI for new explanations. Write full past papers in three hours, closed book, and use AI only to mark them. Prompt 7 helps you choose which topics to revise first.
Related reading: AI for College Teachers: 10 Tasks to Hand Over, 5 to Keep · Which Careers Will AI Change? Skills to Add During Your Degree
Frequently asked questions
Is using ChatGPT for college assignments allowed in India?
There is no single national rule for undergraduates. It depends on your university, college and teacher. Ask for each assessment whether AI is allowed and how to disclose it, and keep the answer in writing.
Do the UGC plagiarism levels apply to my B.Tech or B.Sc assignments?
No. The 2018 regulations cover masters and research theses and dissertations, and publications by faculty, researchers and staff. They expressly exclude assignments, project reports, course work and answer scripts, which fall under your university's rules.
Can my college detect AI-written work?
Some colleges use AI-detection tools, but they make mistakes, and Turnitin itself says its AI score should not be the sole basis for action against a student. Your best protection: write your own work, keep your drafts and version history, and be ready to explain your answers in a viva.
Is it safe to upload my notes to an AI tool?
Your own notes, yes. Not other people's personal details or confidential material: Google's Gemini privacy page says some chats are read by human reviewers and asks users not to enter confidential information.
Still choosing where to study? Compare VTU-affiliated colleges in Bangalore and Bangalore University colleges, and see our entrance exams guide. Eduvy's AI counsellor, Edubro, answers questions on courses, fees and hostels; check its answers as you would any AI's.
Sources
- UGC Promotion of Academic Integrity and Prevention of Plagiarism Regulations, 2018, Gazette text (ugc.gov.in)
- PNAS 2025, "Generative AI without guardrails can harm learning: Evidence from high school mathematics" (pmc.ncbi.nlm.nih.gov)
- Deccan Herald, 9 Jan 2026, VTU circular on AI content in Ph.D. theses (deccanherald.com)
- VTU Ph.D. circular dated 07-01-2026 (vtu.ac.in)
- VTU B.E./B.Tech Regulations 2022, SEE question paper pattern, clause 22OB A6.2.1 (vtu.ac.in)
- IIT Delhi, On the use of Generative AI at IIT Delhi, committee report dated 30 Oct 2024 (academics.iitd.ac.in)
- ThePrint, 28 May 2025, on the absence of a national AI policy for universities (theprint.in)
- Scientific Reports 2023, "Fabrication and errors in the bibliographic citations generated by ChatGPT" (pmc.ncbi.nlm.nih.gov)
- APA Style, how to cite ChatGPT (apastyle.apa.org)
- MLA Style Center, citing generative AI (style.mla.org)
- Turnitin guidance on its AI writing detection model (guides.turnitin.com)
- Gemini Apps Privacy Hub (support.google.com)
- Indian Pharmacopoeia Commission, National Formulary of India (ipc.gov.in)