Short answer: use AI to test you, not to write your answers. Ask it for one question at a time on a named topic and level: a definition, a diagram you must draw and explain, a short data-response question, or a point you must evaluate. Answer from memory first, then check its feedback against your notes and specification, because AI often drifts to the wrong level or gets details wrong. Retrieval practice of this kind beats rereading for exams days or weeks away.
Why use AI for economics revision at all?
Economics exams reward the same few skills again and again: precise definitions, accurate diagrams, using data from an extract, and chains of analysis that end in a judgement. All of them improve with practice, and the bottleneck for most students is simply having enough questions to practise on. Past papers run out. AI does not.
There is now some evidence of what that looks like in a real course. At the University of Sydney, lecturers Huy Vu and Kadir Atalay built an AI practice bank for a large first-year unit, Economics for Business Decision Making, in Semester 2, 2025. It was a custom GPT trained on the unit's own tutorial questions, online quizzes and practice exams, and the teaching team reviewed the questions and worked solutions it produced. Of the students who responded to their survey, 97 used it before the test, which the authors report as 74% of consenting respondents. Among users who rated it, 65% found it moderately, very or extremely useful, and 72% came back more than once.
Two details from that pilot matter for anyone revising alone. Students valued it mainly as low-stakes practice with worked solutions, not as an AI novelty. And the most common request was for questions that looked more like the real assessment. The authors' advice to teachers applies to you too: quality and alignment with the exam mattered more than volume.
Students are already using AI heavily. The HEPI Student Generative AI Survey 2026, based on 1,054 full-time UK undergraduates, found that 95% use AI in at least one way and 94% use generative AI to help with assessed work. The question is not whether to use it, but whether it makes you do the thinking.
Definitions and key terms
Definitions are cheap marks that students lose through vague wording. "Price elasticity of demand is how much demand changes" will not get the mark; the responsiveness of quantity demanded to a change in price will. AI is a good drill partner here, provided it quizzes rather than lectures.
A prompt that works:
I'm revising A level economics, [exam board], topic: [market failure]. Give me one key term at a time and ask me to define it. Wait for my answer. Then tell me what an examiner would want that I left out. Don't give me the definition before I try.
Two refinements. First, paste in your own glossary or class notes and tell it to use only those definitions, because exam boards word things differently and the version you will be marked against is the one your teacher gave you. Second, keep a list of terms you got wrong and ask for those again two or three days later. Geoff Riley at tutor2u suggests AI-made flashcards on topics such as elasticity or fiscal policy and key term glossaries pulled from your texts; both work best when you are answering them, not reading them.
Diagrams: draw first, then ask
Diagrams are where chatbots are weakest and where students most need practice. Text-only AI can describe a diagram but often mislabels axes or shifts the wrong curve when it tries to draw one. So use it for the question, not the picture.
- Ask for a scenario. "Give me a scenario that requires an AD/AS diagram. Don't tell me which curve shifts."
- Draw it on paper. Label the axes, the original equilibrium, the shift and the new equilibrium.
- Explain it in writing. Two or three sentences: what shifted, why, and what happened to price level and real output.
- Ask for a check. Type your explanation back and ask what is missing. Then compare with the diagram in your textbook or notes, not the chatbot's description.
The common core diagrams (supply and demand shifts, a tax or subsidy, externalities, price floors and ceilings, AD/AS, the cost and revenue curves in theory of the firm) each deserve several rounds. The skill being tested is reading a scenario and choosing the right diagram, so mix topics rather than drilling one diagram twenty times; our guide to interleaving explains why mixing helps.
Data response and evaluation chains
Higher-mark questions want a chain: point, explanation, link to the question, then a judgement. AI is useful for generating the raw material and for pushing back on weak links.
- Data response. Ask for a short invented extract with a small data table (inflation, unemployment, a market's prices) and two questions: one that asks you to calculate or describe a change, and one that asks you to explain it using the data. Insist the data is labelled as invented.
- Chains of analysis. Give it a cause and ask it to stop you after each step: "A rise in interest rates..." You write the next link; it tells you if the step is missing or wrong.
- Evaluation. Ask for one argument, then ask for the conditions under which it would not hold: time lags, elasticity, the size of the change, how other agents respond. Write your own judgement and ask which assumption it depends on.
Write before you read. If you ask AI for a model answer first, you will recognise its points and believe you could have made them. Roediger and Karpicke (2006) found that students who restudied prose felt more confident, yet students who had taken recall tests remembered substantially more a week later. Revision that feels easy is often the kind that fades.
Where AI goes wrong in economics
- The wrong level. Ask about monopoly and you may get degree-level game theory. Say your level and board in every new chat.
- Invented statistics. Real-world examples earn marks, but AI-supplied figures for inflation, GDP or a policy's effect are often wrong or out of date. Take examples from your teacher, textbook or an official source.
- Diagram errors. As above, never learn a diagram from a chatbot's description alone.
- Confident marking. AI feedback on a 25-mark essay is a prompt for reflection, not a grade. Our post on whether AI can mark your answers covers why.
- Coursework rules. Using AI to quiz yourself is revision. Using AI-written text in work that is assessed is a different matter; check your school or university's policy.
The fix for most of these is the same as in the Sydney pilot: tie the questions to the material you will be examined on. Ace (acenow.app) is a study app that turns your own lecture PDFs, slides or YouTube lectures into multiple-choice questions checked against the source. Under each answer it shows the sentence from your file that backs it, and where your slides contain diagrams it can ask you to label them using screenshots from your own file. You choose the chapters; you can see how it works. For the wider picture, our guide to using AI for A level revision covers the same principles across subjects.
Frequently asked questions
Can ChatGPT help with A level economics?
Yes, as a question generator and a sparring partner for evaluation. Tell it your exam board and topic, ask for one question at a time, and answer before you see any model answer. Check definitions and examples against your notes.
Is AI good at economics diagrams?
It is good at setting diagram questions and checking a written explanation, and unreliable at drawing or describing diagrams accurately. Draw them yourself and check against your textbook.
How do I practise evaluation with AI?
Ask for one argument, then ask under what conditions it would fail. Write your own judgement first and only then ask which assumptions it depends on. That keeps the thinking on your side.
Is it cheating to revise economics with AI?
Quizzing yourself with AI is revision, much like using a textbook. Submitting AI-written text as your own in coursework or assessed work is not. Your school or university's policy is the rule that applies.
Sources
- Vu, H., & Atalay, K. (2026). Using an AI practice bank to support formative assessment in economics. Teaching@Sydney, The University of Sydney. educational-innovation.sydney.edu.au
- Stephenson, R., & Armstrong, C. (2026). Student Generative Artificial Intelligence Survey 2026 (HEPI Report 199). Higher Education Policy Institute. hepi.ac.uk
- Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: taking memory tests improves long-term retention. Psychological Science, 17, 249–255. DOI
- Riley, G. (2025). Economics teaching & learning with AI: tools for students & teachers. tutor2u. tutor2u.net



