How to use AI in the IB without crossing the line

Almost every student I tutor uses AI. I use it too: to build mock exams, generate practice questions and check worked solutions. The question isn't whether to use it. It's how to use it so it makes you better, without putting your diploma at risk.

What the IB actually says

The IB set out its position in 2023, in an appendix on AI tools added to its academic integrity policy, and has kept it since. In my words:

• AI isn't banned. The IB treats it as a tool students have to learn to use honestly, a bit like calculators or translation software.
• AI output is never your own work. The IB does not count anything an AI tool produced, even a small part, as the student's work.
• If you include AI-generated material, cite it twice. Mark it clearly in the text, in quotation marks if it's text, and list it in the bibliography. Leaving that out counts as academic misconduct.
• Cite at the point of use, with the prompt and the date. The in-text citation should show the prompt you gave and the date the tool produced the output. This applies whether you copied, paraphrased or edited what it gave you, and to images and graphs too.
• AI is a research aid, not a ghostwriter. The IB's examples of fair use are research-type tasks: asking a tool questions, then checking what it says against real sources, because it can make things up.

Your school's own policy can be stricter. If in doubt, ask your teacher before you use a tool for coursework, not after.

My rule of thumb

The line I give students is simple:

AI can quiz you, explain things differently and check your answer. It must not write for you. And you can't take its feedback one-to-one.

The last part matters. AI feedback is sometimes wrong, often generic, and doesn't know your teacher's expectations or the rubric as well as it sounds like it does. Treat it like feedback from a confident classmate: worth thinking about, not worth obeying blindly.

Genuinely useful ways to use AI

For revision (no integrity issue at all):

• "Quiz me on [sub-topic], one question at a time, and don't give me the answer until I've tried." Self-testing is the most effective revision method there is.
• Ask for a different explanation. "Explain the chain rule using a real-world example" or "explain osmosis to someone who missed the lesson".
• Check your worked answer, not by asking for the solution first, but by pasting your full working and asking where the error is.
• Generate flashcards from your own notes, then check them before adding them to Anki.
• Mock exams in the style of a paper, which you then mark against a real mark scheme.

A caution on AI-written multiple-choice questions: in my experience they're often easier than real exam questions and can have tells (like the correct answer being the longest). Use them for drilling facts, and use real exam-style questions to practise for the actual exam.

For coursework (IA, EE, TOK), carefully (the step-by-step version for internal assessments is in using AI in your IA):

• Brainstorming topics and checking whether a research question is precise enough. The question and the decisions must still be yours.
• Finding what to search for: keywords, which databases to try, which methods exist for measuring X.
• Feedback on clarity of a draft you wrote, used critically.

Not okay for coursework:

• Text, analysis or conclusions written by AI and submitted as yours.
• AI-generated data, results or "interviews".
• Rewriting your draft through AI so it sounds better, then submitting that version.

The same applies to the Extended Essay: the reflection sessions and viva are where your supervisor checks the work is yours, so keep your AI use visible there too. How the EE reflections and viva work.

Keep your process visible

The best protection is a process that's obviously yours, and it's also how you'd do good work anyway:

• Write in one document over time, so the version history shows your work growing.
• Keep your raw data, notes, photos and earlier drafts. In a field IA, timestamped photos of every sample are evidence and good practice.
• Know your work well enough to talk about it. Teachers authenticate IAs and EEs, often by talking to you about your choices. If you can explain every decision, there's nothing to worry about.
• If you used AI for something that ends up in the work, cite it in the format your school uses, with the prompt and date.

Remember too that AI detectors are unreliable (the IB itself says it won't treat a detector result as enough evidence on its own), and that's a reason to keep good evidence of your own process, not a reason to relax.

A note on interview and essay prep

In mock-interview or essay practice, I don't accept AI-generated answers from students, because the point is to find your reasons. After a few rounds of being pushed for their own thinking, students go from generic lines to concrete reasons that are genuinely theirs. That's exactly what examiners and interviewers reward.

Sources

• IB: Statement on ChatGPT and artificial intelligence in assessment and education (2023).
• IB: Artificial intelligence: questions and concerns (2023).
• IB academic integrity policy (updated March 2023), Appendix 6 on the use of AI tools: crediting AI in the text with the prompt and date, plus a bibliography entry.