# How to Use AI for Studying Without Skipping the Learning

A five-step way to use AI for studying with your own course materials: hints first, explain it back, try it unaided, then check. Copyable prompts included.

Published by BoredProf. 2026-11-13.

You have a stack of lecture slides, a problem set due Friday, and an AI chat window. Do you ask it for the answer? You can, but you will recognize the solution on the page and still freeze on the exam. The better way to use AI for studying is to make it ask you things, give you hints before answers, and work from your own course materials, then to attempt a similar problem on your own and check the result against your notes.

This guide gives you that workflow in five stages, with a prompt you can copy for each one, a worked example that explains why every step follows, and a quick check at the end.

BoredProf

app.boredprof.com

Study from your own course materials

Upload your lecture slides or problem, then use voice chat and whiteboard explanations to work through the ideas and check your understanding.

[Start free](https://app.boredprof.com/?start=1)

## TL;DR

- Use AI to strengthen your thinking, not replace it: [EPFL’s student guidelines](https://www.epfl.ch/education/teaching/index-html/ai-teaching/guidelines-for-using-ai-in-studying/) say the same.

- Work in five stages: ask for a hint, explain the idea back, request a different diagram, attempt a similar problem unaided, then verify against your course notes.

- Specific prompts beat vague ones. Name the course, paste the slide or problem, and say what you have already tried.

- Always check AI output for errors, especially in math, and follow your course’s AI-use rules.

- BoredProf teaches from your uploaded slides by voice and whiteboard. You can try the demo or start the free plan, no card needed.

## Why “just give me the answer” backfires

Clemson University’s guide describes the trap well: [passive learning happens when you ask “How do I solve this problem?” and only read the response](https://media.clemson.edu/asc/using-ai.pdf). Active learning means thinking about the content, trying it yourself, and then reflecting on your thought process. A fluent explanation feels like understanding. It is not the same as being able to produce the steps from a blank page.

The University of Pennsylvania’s Weingarten Center adds that [learning involves uncertainty and productive struggle](https://weingartencenter.universitylife.upenn.edu/studying-with-ai/): growth comes from grappling with hard concepts, making mistakes and revising. An AI that removes all the struggle removes the part that builds the skill.

There is encouraging evidence on the other side. A guide by Paul Francis at the Australian National University cites a study of first-year Physics students at Harvard in which [students who used AI as a learning buddy did about 15% better](https://www.mso.anu.edu.au/~pfrancis/How_to_Use_AI.pdf) than those taught in normal ways. The point of that result is the word “correctly”: the AI was used to guide learning, not to hand over solutions.

The same logic explains why retrieval practice matters. EPFL’s guidelines note that AI can act as an interactive partner for retrieval practice, [one of the most powerful study methods](https://www.epfl.ch/education/teaching/index-html/ai-teaching/guidelines-for-using-ai-in-studying/), because it makes you pull information out of memory instead of rereading it.

## Start with your own materials

A generic chatbot answers from everything it has ever seen. Your exam is written from what your professor said and showed. So the first move is to give the AI your lecture slides, your notes or the exact problem, and ask it to stay inside them.

Duke’s Academic Resource Center suggests [previewing class materials to find key terms](https://arc.duke.edu/prepare-class-ai/) and asking AI whether you captured the most important concepts. Its toolkit also tells students to [compare AI output with their course materials](https://arc.duke.edu/ai-toolkit/) to stay actively engaged. If you want more on what a slide-based tool should do, our guide to a [lecture slide tutor](/blog/lecture-slide-tutor) covers it.

BoredProf is built around this idea. You upload lecture slides, PDFs, Word or PowerPoint files, or a photo of homework, and it explains the material out loud while drawing steps and figures on a whiteboard.

## The five-stage workflow

Each stage has one job. Skip none of them, and do not move on until the current one is done.

### Stage 1: Ask for a hint, not the solution

The first prompt sets the rules of the session. Many tools now have study or learning modes, and [Clemson notes that these give guidance, questions and hints rather than complete answers](https://media.clemson.edu/asc/using-ai.pdf). Without such a mode, you can ask for the same behavior yourself.

Copy this prompt:

I’m studying [course] and I’m stuck on this problem: [paste problem]. Here is what I tried so far: [your attempt]. Don’t solve it. Give me one hint that points to the next step, then wait for me to respond.

Why it works: you commit to an attempt first, so the hint lands on a real gap rather than on a blank page.

### Stage 2: Explain the idea back

Teaching the concept in your own words exposes the fuzzy parts faster than rereading. Duke’s toolkit advises students to [share their understanding and then ask for feedback](https://arc.duke.edu/ai-toolkit/), which keeps you in the lead.

Copy this prompt:

Here is my explanation of [concept] in my own words: [your explanation]. Compare it with my lecture notes below. Tell me what I got right, what is missing, and what is wrong. Don’t rewrite it for me.

### Stage 3: Request a different diagram or angle

If the first explanation did not click, do not ask for the same thing louder. Ask for a different representation: a picture, a smaller example, an analogy from something you already know. Our piece on [visual learning strategies](/blog/visual-learning-strategies) shows how diagrams help you study, without sorting anyone into a fixed learner type. Everyone benefits from seeing an idea more than one way.

Copy this prompt:

I still don’t get why [step or idea]. Explain it a different way: draw or describe a diagram, use a simpler numerical example, and tell me which line of my slide it connects to.

### Stage 4: Attempt a similar problem unaided

This is the stage most people skip, and it is the one that tests whether you learned anything. The University of Melbourne’s guide offers prompts such as asking for [test questions without the answers until you request them](https://students.unimelb.edu.au/academic-skills/study-skills/learning-with-genai/studying-with-genai). Johns Hopkins gives a similar prompt that asks for [practice problems without answers first, then waits for you to try them](https://guides.library.jhu.edu/c.php?g=1531299&p=11476047).

The UNC Learning Center states that [self-testing is one of the best ways to prepare for an exam](https://learningcenter.unc.edu/tips-and-tools/generative-ai-for-academic-study/). Close the chat, put the notes away, and solve it on paper.

Copy this prompt:

Give me two problems similar to the one I just worked on, with different numbers. Don’t show answers. After I send my working, tell me which step first went wrong, if any.

### Stage 5: Verify against your course notes

AI can be wrong, and it can sound confident while it is. Oxford’s AI Competency Centre reported that students found [accuracy issues particularly in mathematics](https://oerc.ox.ac.uk/ai-centre/ai-in-education-case-studies/how-students-used-genai-during-their-astrophoria-foundation-year) and concluded that AI works best as a complement to independent thinking. BBC Bitesize says the same in plain terms: [fact-check AI-generated information against reliable sources](https://www.bbc.co.uk/bitesize/articles/zf7xkhv) before relying on it.

Copy this prompt:

Point to the exact slide, page or formula in my notes that supports each step of your explanation. If a step isn’t in my materials, flag it so I can check it with my textbook or instructor.

Then do the actual check: find that slide, confirm the formula, and re-derive one step by hand. If you want a broader method for judging AI tools, this [AI testing methodology](https://blog.aimaestrohub.com/blog/ai-testing-methodology) article walks through how to evaluate them properly.

## A worked example: the chain rule

Say a calculus slide shows the chain rule and you are asked to differentiate sin(3x²). Here is how the five stages play out, and why each step follows.

Your attempt (Stage 1). You write d/dx sin(3x²) = cos(3x²) and stop, unsure what to do with the inside. That is a useful mistake. The hint you want is: “What happens to the inside function when you differentiate?”

The reasoning.

- Identify the outer and inner functions. The outer function is sin(u) and the inner is u = 3x². This comes first because the chain rule applies to a function nested inside another, so you must name the layers before you can differentiate them.

- Differentiate the outer function, leaving the inner alone. The derivative of sin(u) is cos(u), so you get cos(3x²). You leave the inside untouched because the rule differentiates one layer at a time.

- Multiply by the derivative of the inner function. The derivative of 3x² is 6x. You multiply because a small change in x changes u, and that change in u then changes the sine. The two rates of change compound.

- Combine. d/dx sin(3x²) = 6x cos(3x²). Your first attempt was missing exactly the factor 6x, which is the “inside” part.

Explain it back (Stage 2). “Differentiate the outside, keep the inside, then multiply by the derivative of the inside.” If you can say why you multiply, you have the idea. If you can only say what to do, ask for a different angle.

Different angle (Stage 3). Ask for a picture: a gear train where turning one gear turns another. The overall speed is the product of the two ratios, which is what the multiplication in step 3 represents.

Unaided attempt (Stage 4). Now try d/dx cos(5x³) on paper, with no help.

Verify (Stage 5). Compare your result with the chain rule slide in your notes, and re-derive the inner derivative yourself.

### Quick check of understanding

Differentiate e^(4x) and write down which function is outer, which is inner, and why you multiply at the end. The answer is 4e^(4x): the outer function is e^u, the inner is 4x, and you multiply by 4 because the inner function changes at that rate.

## Turn a vague question into a good one

Most weak AI answers come from weak questions. Oregon State’s Academic Success Center recommends [showing the AI real examples of the questions you are working on](https://success.oregonstate.edu/learning-corner/studying/using-ai-without-losing-learning) so its follow-ups line up with your course. Here is the difference.

Before:

Explain titration.

After:

I’m in a first-year chemistry course. My slide 14 shows a strong acid and strong base titration curve, and I don’t understand why the pH jumps so sharply near the equivalence point. Explain it using the slide, ask me one question to check my understanding, then give me a similar problem with different concentrations. Don’t show the solution until I try it.

The second version names your level, points to your material, states the exact confusion, and sets the order of events. It also builds in a check and a practice problem, so the session produces learning rather than just text.

## Using voice and whiteboard to study

If you have wondered how an AI tutor with voice works and whether it helps, the idea is simple: you talk, it answers aloud, and you can interrupt with a follow-up the way you would in office hours. Oxford’s Astrophoria case study reports students using [voice mode to revise without typing](https://oerc.ox.ac.uk/ai-centre/ai-in-education-case-studies/how-students-used-genai-during-their-astrophoria-foundation-year), along with step-by-step Socratic questioning and tools that read handwritten equations.

Voice suits a few specific jobs:

- Explaining back. Say the concept aloud and let the tutor respond. Speaking forces you to finish sentences you would otherwise leave vague in your head.

- Walking through a derivation. Ask “why that step?” the moment it confuses you, instead of waiting until the end.

- Review on the move. Captions help when you are somewhere noisy or quiet and want to read along.

One limit from the same Oxford report is worth planning around: students noted that general chat tools struggled to produce diagrams. For physics, chemistry and biology, where the picture carries half the meaning, a tool that draws as it speaks fills that gap.

## Math, physics, chemistry and biology: what each needs

The workflow stays the same, but the emphasis shifts.

- Math: attempt first, then compare line by line. Ask the AI to find the first wrong step in your working, not to redo the problem. Re-derive anything you cannot follow.

- Physics: ANU’s guide suggests asking AI to act as a physics professor and [explain a topic in short steps, testing your understanding after each one](https://www.mso.anu.edu.au/~pfrancis/How_to_Use_AI.pdf). Always ask for units and a sanity check on the answer’s size.

- Chemistry: ask for the particle-level picture behind each equation, then do the stoichiometry yourself.

- Biology: ask for processes as sequences (what triggers what), then redraw the pathway from memory.

For a wider look at how AI fits into a course, see [AI-assisted learning: what works](/blog/ai-assisted-learning). And since mistakes are part of the loop, [learning from mistakes](/blog/learn-from-mistakes) is worth a read after your first unaided attempts go wrong.

## Rules, privacy and honest use

Three guardrails keep this workflow clean.

Follow your course’s AI policy. Policies differ by class and instructor, so read the syllabus and ask if it is unclear. Use AI to understand material, not to produce work you will submit as your own. Receiving an answer is not the same as understanding, and ANU’s Paul Francis warns that students who only use AI to do the work [will fail the exams](https://www.mso.anu.edu.au/~pfrancis/How_to_Use_AI.pdf).

Protect your personal information. The York Region District School Board’s student guidelines say not to enter [full names, student ID numbers, home addresses, phone numbers or dates of birth](https://www2.yrdsb.ca/student-support/artificial-intelligence/student-ai-guidelines-grades-7-12), and not to upload personal photos or audio of yourself or others. Pike Parent University in Indiana puts it in one line for families: [AI is not private](https://ppu.pike.k12.in.us/artificial-intelligence-ai-for-fami). It also notes that students generally must be at least 13 to create accounts on major AI platforms. If you are a high school student, ask a parent or guardian before signing up for any tool. Our guide to [AI tutors for high school students](/blog/best-ai-tutors-for-high-school-students) covers how to choose.

Do not let it write your notes. Melbourne’s guide cautions that having AI write your notes or transcribe for you means [missing the chance to engage with the material](https://students.unimelb.edu.au/academic-skills/study-skills/learning-with-genai/studying-with-genai). Summaries are a start. Your own notes are where memory forms.

## Where BoredProf fits

The five stages work in any chat window, but they take discipline to keep up. BoredProf is designed around them. You upload your lecture slides or a photo of a problem, it explains the idea by voice and draws the steps and figures on a whiteboard, and you can interrupt with follow-up questions, use captions, save lessons to revisit, and get feedback on your handwritten working. Subjects on the site are math, physics, chemistry and biology.

The free plan includes one lecture from one course, with usage caps, and needs no card. Pro is 29.99 CAD per month and Max is 59.99 CAD per month, both with expanded access; check the [BoredProf pricing page](https://boredprof.com) for current limits and terms. Users must be 13 or older under the [terms](https://boredprof.com/terms.html), and the [privacy policy](https://boredprof.com/privacy.html) explains how data is handled. BoredProf is for understanding and checking explanations. It does not promise grades or perfect accuracy, so keep Stage 5 in your routine. You can try the [public demo](https://boredprof.com/#demo), which uses prewritten sample answers drawn on the real board, or [start free](https://app.boredprof.com/?start=1). If you want to see worked problems first, browse the [examples library](https://boredprof.com/examples/what-do-i-need-on-my-final).

## FAQ

### Is it cheating to use AI for studying?

Using AI to understand material is usually allowed, but policies vary by course and instructor. Read your syllabus and ask if it is unclear. Submitting AI-written work as your own, or using it on assessed work where it is banned, is a different matter.

### What is the best prompt for studying with AI?

A good prompt names your course, pastes the exact slide or problem, says what you already tried, and tells the AI to hint instead of solve. Add a request for one check question and one similar practice problem with the answer held back until you attempt it.

### Can AI explain my lecture slides?

Yes, if you give it the slides and ask it to stay within them. Ask it to point to the slide behind each step so you can verify. BoredProf accepts slides, PDFs, Word and PowerPoint files and explains them by voice with whiteboard drawings.

### How do I know if the AI made a mistake?

Compare each step with your lecture notes or textbook, re-derive one step by hand, and check units and sizes of answers. Math is the area where errors are most often reported, so treat every explanation as a draft to confirm.

### Can high school students use AI tutors?

Most major AI platforms require users to be at least 13, and BoredProf’s terms require age 13 or older. Check the tool’s terms, talk to a parent or guardian, and never enter personal details such as your full name, address or school ID.

## Sources

- [Guidelines for using AI in studying, EPFL](https://www.epfl.ch/education/teaching/index-html/ai-teaching/guidelines-for-using-ai-in-studying/)

- [How to Use AI, Paul Francis, Australian National University](https://www.mso.anu.edu.au/~pfrancis/How_to_Use_AI.pdf)

- [How to Use AI Without Hurting Your Learning, Clemson University](https://media.clemson.edu/asc/using-ai.pdf)

- [Studying with AI, Weingarten Center, University of Pennsylvania](https://weingartencenter.universitylife.upenn.edu/studying-with-ai/)

- [AI Toolkit, Duke Academic Resource Center](https://arc.duke.edu/ai-toolkit/)

- [Class Prep with AI, Duke Academic Resource Center](https://arc.duke.edu/prepare-class-ai/)

- [Generative AI for Academic Study, UNC Learning Center](https://learningcenter.unc.edu/tips-and-tools/generative-ai-for-academic-study/)

- [AI for Teaching and Learning, Johns Hopkins University Libraries](https://guides.library.jhu.edu/c.php?g=1531299&p=11476047)

- [Using AI for Help Without Losing the Learning, Oregon State University](https://success.oregonstate.edu/learning-corner/studying/using-ai-without-losing-learning)

- [Studying with GenAI, University of Melbourne](https://students.unimelb.edu.au/academic-skills/study-skills/learning-with-genai/studying-with-genai)

- [How to use AI for homework, BBC Bitesize](https://www.bbc.co.uk/bitesize/articles/zf7xkhv)

- [How students used GenAI during their Astrophoria Foundation Year, University of Oxford](https://oerc.ox.ac.uk/ai-centre/ai-in-education-case-studies/how-students-used-genai-during-their-astrophoria-foundation-year)

- [Student AI Guidelines (Grades 7-12), York Region District School Board](https://www2.yrdsb.ca/student-support/artificial-intelligence/student-ai-guidelines-grades-7-12)

- [Artificial Intelligence for Families, Pike Parent University](https://ppu.pike.k12.in.us/artificial-intelligence-ai-for-fami)

- [AI Testing Methodology: How to Test AI Tools Properly](https://blog.aimaestrohub.com/blog/ai-testing-methodology)
