Overview
The IB Maths AI Internal Assessment — the *Mathematical Exploration* — is worth 20% of your final Applications & Interpretation grade. It is a short independent piece of mathematics (about 12–20 pages) on a topic you choose, and for AI it plays to the course's strengths: real data, modelling, and technology. This guide covers the criteria, a structure that scores, and the pitfalls moderators penalise.
What examiners mark
For the IB Diploma Programme, the Exploration is currently marked on five criteria (20 marks total). Confirm the exact wording and mark split against your current *Mathematics: Applications and Interpretation guide*; the criteria are:
- Presentation — organised, coherent, concise.
- Mathematical communication — correct notation, defined variables, clear graphs and tables.
- Personal engagement — the exploration is genuinely yours.
- Reflection — critical evaluation of results, limits of your model, and next steps.
- Use of mathematics — correct, relevant mathematics at a level commensurate with the course.
Choosing a topic that scores
For the IB Diploma Programme, aI rewards applied, data-driven questions where mathematics models something real:
- Modelling — fit a function (linear, exponential, logistic, trigonometric) to data you collected or sourced, then use it to predict or optimise.
- Statistics — a well-designed test (correlation, chi-squared, regression) answering a question you actually care about, with enough data to be meaningful.
- Real context — sport, business, health, the environment, your part-time job. A personal data source lifts personal engagement.
Steer away from purely abstract proof-heavy topics — those belong in Maths AA. For AI, the mathematics should do something with real numbers.
Recommended structure
Introduction (the real-world question and why it matters to you) → data source and how you gathered/cleaned it → the model or test, built step by step with defined variables → technology used sensibly (GDC, spreadsheet, graphing software) → results and predictions → reflection on the model's accuracy, assumptions, and limitations → conclusion answering your question.
Common pitfalls
For the IB Diploma Programme, collecting data with no clear question; a model fitted but never interpreted or validated; statistics run without checking assumptions; graphs with no axis labels or units; a reflection that says "this went well" instead of critiquing the model; and exceeding the page guideline when concision is rewarded.
Criterion practice on MarkScheme
Write the draft early and self-mark it against the official criteria before your supervisor deadline. Strengthen the underlying methods with the free [Maths AI HL](/ib/courses/maths-ai-hl) and [SL](/ib/courses/maths-ai-sl) lessons, and [get an answer marked](/mark) to keep your exam technique sharp alongside the IA.
Frequently asked questions
This section covers Frequently asked questions — what IB examiners reward most often in past papers and coursework.
What makes a good Maths AI IA topic?
Something real you can gather or source data for, then model or test — with a clear question and enough data to draw a defensible conclusion. Personal, applied topics score best.
How much data do I need?
Enough to fit or test meaningfully and to discuss reliability — typically dozens of data points rather than a handful. Quality and relevance matter more than raw volume.
Do I have to use technology?
AI expects sensible use of technology (graphing tools, spreadsheets, your GDC). Show it supporting the mathematics, not replacing your understanding.