What you’ll be able to do

Right, so this one’s about using AI properly for homework. Not getting it to write your essay or spit out the answer to question 7b, but actually using it to understand the thing you’re stuck on.

By the end of this you’ll be able to set up a free AI tool, ask it to tutor you instead of solve for you, and check whether what it’s telling you actually holds up. That last bit matters more than people think. AI gets things wrong, especially with maths, so you need to know how to test it.

Before you start

Here’s the bit people skip and shouldn’t: check your school or university’s actual policy on AI. Some allow it for certain tasks, some ban it outright, some want you to declare when you’ve used it. This varies wildly and it changes often, so don’t assume. Ask your teacher, check the student handbook, look at what your institution has published. If a source is required, cite it properly, the same as you would a textbook.

The bigger point though, and this is the whole reason this guide exists, is that copying an answer defeats the purpose. You’re not doing homework to produce a correct answer for its own sake. You’re doing it so the concept sticks, so next time it comes up in an exam or in real life you actually know it. If AI just gives you the answer, you’ve skipped the bit where the learning happens. Anyway.

Get set up

You don’t need anything fancy here. A few of the mainstream tools do this job well, and most have a free tier that’s plenty for homework help.

ChatGPT has something called study mode, which is built specifically for this: it asks you questions rather than handing over answers straight away. That’s genuinely useful and worth trying first if you’re new to all this. You’ll need a free account, just an email address to sign up.

Gemini has a guided-learning mode that does something similar, and Claude can simply be told to act as a tutor rather than solve the problem; it follows that instruction well. Both have free tiers too.

If it’s specifically maths, Photomath is worth having on your phone. You point your camera at the problem and it shows the steps, which is handy for checking your own working rather than starting from scratch.

And if you’re at a school that uses Khan Academy, there’s Khanmigo, which is built from the ground up as a tutor rather than an answer machine. Worth asking your teacher if your school has access, because it’s designed with exactly this use case in mind.

None of these need anything more than an internet connection and, in most cases, a free account. Get one set up, pick your subject, and you’re ready.

Try it yourself

This is where the actual prompts come in. The trick is telling the AI how you want it to behave, not just what the problem is.

Start with this one, it’s the single most useful thing in this whole guide:

Don't give me the answer. Ask me what I think first, then give one hint at a time.

Paste that in before you share your actual homework question. It flips the whole interaction. Instead of getting a finished answer, you get asked what you already think, and then nudged along a hint at a time. It’s slower, which is the point.

Once you’ve had a go and got the right method but you’re not sure you’ve really got it, ask for a similar problem with different numbers:

Give me a worked example of this exact type of problem but with different numbers, and show me the steps.

This one’s brilliant for maths especially. You watch the method applied to a fresh set of numbers, then you try applying it yourself to your actual homework. If you can follow the worked example and then do your own version without help, you’ve probably actually learned it, not just copied a result.

Then, once you’ve done your own attempt, this is the one that catches mistakes:

Mark my answer and explain what I got wrong, step by step, without just giving me the correct answer straight away.

Paste your own working alongside this. A good response will point at the specific step where things went sideways rather than just saying “wrong” and handing you the fix. If it does just hand you the fix, ask it again and remind it to explain rather than solve.

A few small things that make a real difference: always tell the AI what subject and level you’re working at (GCSE maths is a very different conversation to A-level further maths), and if you’ve already tried something, paste in your attempt rather than starting from a blank page. The AI can only tutor around what you’ve actually done.

Check the result

This bit’s important because AI does get things wrong, and maths is a particular weak spot. It can produce a confident, nicely formatted explanation that’s just incorrect.

So check it. If it’s a maths problem, do the calculation yourself with a calculator or by hand once you’ve followed the explanation, and see if you land in the same place. Don’t trust the AI’s answer just because it looks tidy.

Ask yourself if you could explain the method to someone else without looking back at the chat. If you can’t, you haven’t actually understood it, you’ve just read it. Go back and ask for the same explanation again, differently, or ask it to break down whichever step didn’t land.

For essay-based or written subjects, check any facts, dates, or quotes it gives you against a real source. AI tools can state things with total confidence that turn out to be wrong, particularly names, dates, and specific figures. If you’re going to use anything factual it’s told you, verify it somewhere else first.

And a genuinely useful test: try the worked example with different numbers yourself, without asking the AI. If you get stuck at the same step every time, that’s your actual gap, and now you know exactly what to go back and ask about.

If it doesn’t work

Sometimes the AI just isn’t landing the explanation, or it keeps giving you the answer even after you’ve asked it not to. A few things to try: be blunter. Say “stop giving me answers, ask me a question instead” directly, mid-conversation, rather than assuming it’ll remember your first instruction forever. Long conversations drift.

If it’s still not clicking, try a completely different explanation approach: ask for an analogy, or ask it to explain the concept as if you were two years younger than you are. Sometimes a different angle works when the first one didn’t.

Now, the privacy bit, and this matters. Don’t put your full name, your school’s name, your home address, or anything else identifying into a free AI chat tool. You don’t know exactly how that data’s used or stored, and there’s no need to include it anyway. A maths problem doesn’t need your surname attached.

Similarly, don’t paste in anything containing other people’s personal information: don’t upload a photo of a group project with names visible, don’t share another student’s work without asking them, and don’t feed in anything from a paid resource or textbook that’s under copyright, since some of that content isn’t meant to be redistributed even into a chatbot.

And the obvious one, worth saying plainly: don’t submit AI output as your own work if your school’s rules don’t allow it. That’s not a grey area, that’s just against the rules, and you’ll likely get caught anyway since a lot of institutions now check for it.

Keep exploring

If you found this useful, there’s a natural next step once you’ve got your explanation sorted: turning what you’ve learned into something you’ll actually remember. There’s a guide on this site called “How to get AI to turn my notes into flashcards” that walks through exactly that.

If any of your homework touches on how websites or content get found online, there’s also “How to get AI to teach me about SEO”, which uses the same tutor-style approach as this guide.

And if you’re ever unsure whether something an AI’s told you is actually true, which, as covered above, you should be, there’s “How to get AI to fact-check something” that covers how to verify claims properly.

Sources and review notes

Review date: 15 September 2026 — every source above was opened and checked against the text on that date.