Subscribe

How to learn a new topic with an AI tutor

AIooze Editorial

In one minute

  • You will be able to use an AI chatbot as a patient tutor that explains, quizzes you and checks your understanding, rather than just handing over answers.
  • You need a free account with an AI assistant and a topic you want to learn.
  • Setting up takes five minutes; a useful study session takes 20 to 40.
  • The habit that matters most: check what it teaches against a trusted source.

Why it is worth doing

If you ask an AI assistant for the answer, you get the answer and little else. If you ask it to teach you, step by step, then quiz you and make you explain the idea back in your own words, you do the thinking yourself. That effort is what makes something stick.

An AI tutor is also patient: you can ask the same question five times, or ask for a simpler example, without feeling awkward.

Step by step

  1. Pick a tool, and see if it has a learning mode. Several assistants now offer one. According to OpenAI’s announcement (29 July 2025, checked 4 Nov 2026), ChatGPT’s study mode works through problems “step by step instead of just getting an answer”, you start it by choosing “Study and learn” from the tools, and it is available to logged-in users on the Free plan as well as paid plans. Google’s Gemini help page (checked 4 Nov 2026) describes Guided Learning, which you choose from the tools menu on the web; you must be signed in, and the page does not say which plans include it. Anthropic’s Claude help page on styles (checked 4 Nov 2026) lists an “Explanatory” preset for “learning new concepts”. Menus change, so check each maker’s page if you cannot find the option.
  2. If there is no learning mode, ask for one. The prompts below work in any assistant, with or without a special mode.
  3. Tell it your starting point. Say what you already know, why you are learning it and how much time you have.
  4. Learn in small chunks. Ask for one idea at a time and do not move on until you can explain it back.
  5. Get tested. Ask it to quiz you, then answer without looking back up the page.
  6. Check what you learned against a textbook, an official course or the original source (more below).

Prompts that work in any assistant

  • “Explain [topic] as if I am completely new to it. Use one everyday example and no jargon. Stop after the first idea and check I follow.”
  • “Check my understanding with 3 questions. Ask them one at a time and wait for my answer before telling me if I am right.”
  • “Give me a worked example, then a similar problem for me to try on my own. Do not show the answer until I reply.”
  • “I will explain this back to you. Tell me what I got wrong or left out.”
  • “Which sources should I read to learn this properly? Name textbooks, official courses or organisations, not just websites.”

A worked example

Say Priya wants to understand compound interest before opening a savings account. (This is an illustrative example.)

She types: “Explain compound interest as if I am new to it. Use one everyday example, then stop and check I follow.” The assistant explains that compound interest means you earn interest on your interest, not just on the money you put in.

She then asks for a worked example. It shows £1,000 at 5% a year: £1,050 after one year, £1,102.50 after two, about £1,157.63 after three. Then it gives her a similar problem to try herself: £2,000 at 3% for two years. She works it out (£2,121.80) and the assistant checks her steps.

Next she asks: “Check my understanding with 3 questions.” One asks why money grows faster the longer it is left. She has to explain it in her own words, which shows her where she is still fuzzy.

Finally she asks which sources to read, then checks the formula against her bank’s own explanation or a recognised personal finance guide. The same pattern works for a spreadsheet skill: ask it to explain one function, give a worked example, then set you a small task to try in your own spreadsheet.

Watch out for

  • AI can be confidently wrong. Chatbots can state errors fluently, including made-up facts and sources. OpenAI itself says study mode may show “inconsistent behavior and mistakes across conversations”. Learn why in What is a large language model?
  • Check the sources it names. Open each one and make sure it exists and says what the AI claims. Cross-check key facts with a textbook, an official course or your teacher.
  • Follow your school’s or employer’s rules. Using AI to understand a topic is different from submitting its work as your own. If you are unsure what is allowed, ask your teacher, tutor or manager before you start.
  • Do not share private details. Leave out your full name, account numbers, health details, other people’s information and confidential work material.
  • Know your privacy settings. Your chats are sent to the company and may be used to improve its models. According to each maker’s help page (checked 4 Nov 2026): in ChatGPT you can turn off “Improve the model for everyone” under Settings, Data Controls, and Temporary Chats are not used for training (OpenAI); in Claude you can turn off “Help Improve our AI models” under Settings, Privacy (Anthropic); in Gemini you can turn off Keep Activity, though Google says chats are still saved for up to 72 hours (Google).

Go deeper

Sources

Related