IB Math AI: GDC workflow + 2029 syllabus outlook

IB Math AI is short for Mathematics: Applications and Interpretation. It is the applied half of the DP mathematics offering. Statistics. Modelling. Real-world contexts. The course leans on technology. A graphic display calculator (GDC) is required on every Math AI paper (Paper 1 and Paper 2 at SL; Paper 1, Paper 2, and Paper 3 at HL). A new syllabus is in development, with first assessment in May 2029. If you coordinate the DP, the next three years need a plan.
TL;DR
Key takeaway | What it means for your school |
|---|---|
GDC required on every Math AI paper (2 at SL, 3 at HL) | Treat GDC fluency as a Year 1 skill, not a Year 2 exam-prep skill |
New syllabus first assessment May 2029 | Use 2026/27 to orient the department once the IB releases the draft guide |
HL Paper 3 confirmed at 1 hour 15 minutes | Update internal mocks to the corrected duration this year |
Real-world applications are the defining feature | IA topics and teaching examples should pull from data students recognise |
More schools are switching cohorts from AA to AI | Placement decisions need a structured Year 11 process, not a default |
Statistics, networks, financial maths are AI-specific | Teaching staff need confidence in these topics, not just the calculus carry-over |
Extended questions trip students who can't recover early stages | Long-form practice from Year 1 protects against late-paper collapse |
The current state of IB Math AI in 2026
The 2021 Math AI syllabus is still active in 2026. The next cohort under the new syllabus sits their first exams in May 2029, so coordinators should plan for two full cohorts under the current guide before any transition affects assessment.

Math AI launched alongside Math AA in 2019, replacing the older Math SL, Math Studies, and Math HL split. The course runs at SL and HL with five topic areas: Number and Algebra, Functions, Geometry and Trigonometry, Statistics and Probability, and Calculus. Same headings as Math AA. Very different depth and emphasis. AI weights statistics and modelling heavily. AA weights algebraic technique and calculus.
External assessment is two papers at SL (Paper 1 and Paper 2) and three papers at HL (Paper 1, Paper 2, and Paper 3). The IA, the mathematical exploration, is worth 20 per cent of the total grade at both levels.

In 2026 the practical state of the course is steady. No structural changes this academic year. No new IA criteria. No shift in the calculator policy. The change to plan for is on the horizon, not in the current term.
The 2029 syllabus transition: What's coming
Math AI is part of the IB's wider curriculum review, with first assessment of the new syllabus scheduled for May 2029. The subject guide hasn't been published yet, so the right move in 2026/27 is preparation rather than redesign.
Here's what the IB has signalled across the review:
Stronger alignment with on-screen assessment
Continued emphasis on conceptual understanding alongside procedure
Tighter integration between the IA and the externally assessed content
Ongoing focus on real-world contexts in AI specifically
And here's what's not yet confirmed for Math AI:
The final topic list
Whether the paper structure stays at three papers HL / two papers SL
Confirmed weightings and durations under the new model
Whether the IA brief or marking criteria will shift
The practical implication for coordinators: don't redesign schemes of work for current students. Do schedule department time in the 2026/27 calendar specifically for the new subject guide once it lands.
HL Paper 3 duration correction: 1h 15min confirmed
HL Paper 3 in Math AI has a confirmed duration of 1 hour 15 minutes (75 minutes). Coordinators should make sure internal mocks reflect the corrected timing this year.
Paper 3 at HL is two compulsory extended-response problem-solving questions, with the GDC available throughout. The corrected 75-minute duration matters because earlier internal mock papers in some schools were running a different length, which distorts pacing practice.
What 75 minutes means in practice:
Roughly 37 minutes per question, but the structure usually rewards more time on the earlier scaffolded sub-parts
Students need to know what "stuck on part (b)" looks like at 12 minutes in, and how to move on
Mock invigilation should match this duration exactly. Practising 90 minutes for a 75-minute paper teaches the wrong pacing
Coordinator action item: confirm with the maths team that the most recent department mock used the corrected Paper 3 duration, and update the schedule for the next mock round.
GDC-required workflow on every paper
Every Math AI paper requires a GDC. Paper 1 and Paper 2 at SL, and Paper 1, Paper 2, and Paper 3 at HL. That makes calculator fluency a non-negotiable Year 1 priority, not a Year 2 polishing skill.
This is the structural difference from Math AA. In AA, Paper 1 is non-calculator. In Math AI, every paper assumes the GDC is in the student's hand. The expected workflows include:
Solving equations numerically when an analytical method is impractical or impossible
Running regression on bivariate data and interpreting r and r² in context
Working with statistical distributions (binomial, normal, and at HL Poisson) using the calculator's distribution functions
Numerical integration and derivative-at-a-point for the calculus topic
Sketching and analysing graphs to support written reasoning
Where students lose marks: they have the GDC, but they don't know how to evidence what they did on it. The markscheme rewards working that shows the student's process. "Graph sketched from GDC." "Values from regression." "Intersection from GDC." That habit of documenting the GDC step has to start in Year 1.
A digital exam environment makes this harder, not easier. Whether the GDC is a physical device next to the student or an embedded equivalent, the on-screen workflow has to leave room for both the algebraic answer and the GDC justification.
The AA-to-AI switch: Why more schools are making it
More international schools are moving cohorts from Math AA to Math AI when the AI course is the better fit for the degree pathway. This is a placement decision, and it works when it's led by the coordinator, not by parent preference.
The pattern is familiar. Year 11 placement traditionally puts the "stronger" maths students into AA and the "less confident" students into AI. That framing is wrong. The IB itself designed the two courses as parallel options for different post-DP destinations, not as a difficulty ladder. When navigating these decisions, families often turn to external support like math tutors on Tutorela to help clarify course requirements and supplement classroom learning.
AI fits students heading into:
Business and management
Social sciences (economics that emphasises statistics, psychology, sociology)
Biology, environmental sciences
Design, architecture, media
Any degree where modelling and applied statistics carry more weight than proof and calculus
What's driving the switch: universities have become much clearer about which course they want. A student who picked AA on prestige grounds and gets a 4 will struggle against a student who picked AI on fit grounds and gets a 6. The grade differential matters more than the course label for most degree applications outside the engineering, physics, and pure mathematics group.
I sat in on an early-Year-12 placement review last term where a DP coordinator caught a tutor coaching a student around what the tutor called "voting systems" as a Math AI study angle. The coordinator's job, mid-meeting, was to pull up the actual syllabus and explain, calmly, that voting systems aren't on it. The real HL-distinctive content sits in graph theory, networks, and Markov chains. That conversation is easier when the placement framework is documented before the tutoring industry gets to the student.

Coordinator practice that works:
Year 11 placement conversation with named alumni examples from the last three years
Shared placement criteria between the maths department and the careers/university advisor
A late-Year 11 or early-Year 12 transfer window with a documented process
Real-world applications: The AI subject's defining feature
Math AI's defining feature is its applied focus. Every topic is taught with a real context, and the IA explicitly rewards explorations grounded in data the student has gathered or sourced rather than purely abstract investigations.
This shows up across the course:
Statistics topics use real datasets, not invented ones
Modelling questions reference real phenomena: population growth, depreciation, projectile motion in sport, network design in logistics
Financial mathematics treats compound interest, loan amortisation, and inflation as genuine financial-literacy content
The IA criteria reward authentic data collection
What this means for teaching: a Math AI scheme of work that runs on textbook abstract examples misses the brief. The strongest department-level practice is to build a bank of context-rich problems your school can re-use. Local data, school data, regional data, to anchor each topic.
What this means for the IA: students who pick a topic they care about, gather their own data, and analyse it well consistently outscore students who pick a "safe" topic from an exemplar list. Coordinator action: in Year 1, push topic selection towards genuine interest, not safe replication.
Statistics, networks, financial mathematics: Math AI's distinct topics
The topics that mark Math AI as a distinct course rather than a lighter version of AA are its statistics depth, graph theory and networks at HL, and financial mathematics. And these are the topics where teaching teams often need the most support.

Statistics in Math AI goes substantially deeper than in AA at both levels:
Bivariate analysis with regression and correlation (SL and HL)
The χ² (chi-squared) test for independence and the two-sample t-test, both part of the SL syllabus under Topic 4.11
Confidence intervals (AHL 4.16). HL only.
The Poisson distribution (AHL 4.17). HL only.
That SL/HL split catches schools out every May. One example: a Year 2 SL student I was talking to last week had her Paper 2 mock back and was furious with herself. She'd written off chi-squared as "an HL topic" because that's what a private tutor had told her. It came up on her actual paper. She'd parked it in revision. The marks went with it. The fix is structural. The department's revision guide has to label every statistics topic with the right level, not rely on what the student picked up from informal sources.
The HL-distinctive content beyond statistics is graph theory and networks (AHL 3.14) and Markov chains (AHL 4.19). These are topics most DP maths teachers haven't taught before. They're not in most teachers' undergraduate experience the way calculus is.
Financial mathematics is sometimes treated as light content. It shouldn't be. Real compound-interest problems, depreciation, inflation, and loan repayment schedules are where students lock in marks that the AA cohort doesn't see.
The coordinator implication: department CPD for AI needs to cover these topics explicitly, not assume teachers will pick them up from the textbook. Investing here in 2026/27, ahead of the 2029 syllabus, is also the most future-proof CPD spend, because real-world applied content is the direction the new syllabus is signalling.
How AssessPrep handles Math AI
AssessPrep's IB question bank holds 500+ Math AI questions tagged by topic, subtopic, paper, calculator policy, and applied context.
The Math AI question bank inside AssessPrep is structured for the realities of the course:
GDC-aware delivery on every paper. The GDC tool sits alongside the on-screen question interface across SL Papers 1 and 2 and HL Papers 1, 2, and 3
Topic-level filtering down to subtopic, so a teacher running a Topic 4 Statistics test pulls only the relevant questions
Applied-context tagging (financial, biological, business, environmental) so coordinators can build mocks that mirror the cross-context flavour of the live paper
Lockdown mode and secure browser for high-stakes mocks
AI-assisted grading that flags likely-correct responses while preserving teacher judgement for the working
Across 800+ partner schools in 85+ countries, the workflow that saves the most teacher time is the combined GDC-aware delivery plus AI grading layer. For HL Paper 3 specifically, coordinators can build multi-part extended problems and track which sub-parts students are getting stuck on across the cohort. That's the diagnostic data paper-based mocks don't surface easily.
We don't have published cross-school data on which Math AI topics produce the biggest mark loss. That varies by cohort and by region. What we do have is the platform-level pattern. Where students consistently lose marks, the failure is usually GDC-evidencing or pacing, not topic-level knowledge.
The IA exploration with real-world data
The mathematical exploration is worth 20 per cent of the final Math AI grade at both SL and HL. The strongest explorations are built around real data the student has collected or sourced. Abstract or recycled topics consistently underperform.
The IA is marked against five criteria (presentation, communication, personal engagement, reflection, and use of mathematics) totalling 20 marks under the 2021 guide.
What works in Math AI specifically:
Topics grounded in something the student is involved in. A sport, a club, a part-time job, a community context
Data that's been collected first-hand or sourced from a real database, not invented
Mathematics that's at level. Using AI-level techniques like regression and chi-squared, not just plotting averages
A reflection section that genuinely engages with what the maths revealed, not a generic "I learnt how to use Excel"
What underperforms: re-runs of well-known exemplar topics, datasets that are too small to analyse meaningfully, and explorations where the maths is below course level.
Coordinator action: a structured Year 1 IA topic-approval process catches the underperforming choices early. The IA proposal stage should ask: what real data, what AI-level technique, what does the student care about?
Action checklist for Math AI coordinators
A practical list for the next 12 months:
Audit internal HL mocks to confirm Paper 3 duration is 1 hour 15 minutes
Run a GDC-evidencing audit. Pick a sample of recent Paper 2 mock responses and check how clearly students are documenting their calculator work
Confirm CPD time for the statistics, networks, and financial maths topics in the coming year
Schedule a 2026/27 department slot for the new subject guide once published
Review your Year 11 AA-vs-AI placement process and add named alumni examples from the last three years
Push IA topic selection towards real-data, personally engaged explorations earlier in Year 1
Check your question bank coverage with the Math AI question bank for topic-specific practice
Key Takeaways
Math AI in 2026 is steady. The 2021 syllabus runs through to first new-syllabus assessment in May 2029.
GDC fluency is a Year 1 skill across every Math AI paper (Paper 1 and 2 at SL; Paper 1, 2, and 3 at HL), not a final-term polish.
HL Paper 3 is confirmed at 1 hour 15 minutes. Update internal mocks now.
The AA-to-AI switch belongs to the coordinator and the degree pathway, not to prestige assumptions.
Statistics, networks, and financial maths are the topics that need the most departmental CPD investment.
If you're auditing your Math AI assessment workflow ahead of the 2029 syllabus transition, book a 30-day free trial of AssessPrep and put the Math AI question bank in front of your team this term.
Frequently asked questions
Is a calculator required for IB Math AI?
Yes. A graphic display calculator (GDC) is required for every Math AI paper. That covers Paper 1 and Paper 2 at Standard Level, and Paper 1, Paper 2, and Paper 3 at Higher Level. This is a structural difference from Math AA, where Paper 1 is non-calculator. Students need fluent GDC workflows from Year 1, including numerical equation solving, statistical distributions, regression, and graphical analysis. Marks are also awarded for evidencing GDC use in written working.
What is the difference between IB Math AI and IB Math AA?
Math AI focuses on applied mathematics: statistics, modelling, networks, financial mathematics. It uses a GDC on every paper. Math AA focuses on theoretical mathematics: algebra, calculus, proof. It has a non-calculator Paper 1. The two are parallel options, not a difficulty ladder. The right choice depends on the student's intended degree pathway. Universities specify which course they require for each programme.
Which statistics topics are HL-only in Math AI?
Many statistics topics commonly assumed to be HL-only are actually part of the Standard Level syllabus. The χ² (chi-squared) test for independence and the two-sample t-test are both in SL Topic 4.11. The genuinely HL-distinctive statistics topics under the 2021 guide are confidence intervals (AHL 4.16) and the Poisson distribution (AHL 4.17). Department revision guides should label each topic by level so SL students don't dismiss content they'll be examined on.
How does AssessPrep support IB Math AI teaching?
AssessPrep is the on-screen assessment platform used by 800+ international schools across 85+ countries. The Math AI question bank holds 800+ questions tagged by topic, subtopic, paper, calculator policy, and applied context. The platform supports GDC-aware delivery across all Math AI papers, lockdown mode for secure mocks, and AI-assisted grading that flags likely-correct responses while preserving teacher judgement on working and method marks.
Why do students struggle with extended questions in IB Math AI?
Extended-response questions, particularly on HL Paper 3, are scaffolded so later parts often build on earlier results. Students who can't resolve the initial stages of a question lose access to follow-on marks in the later parts. The failure is structural rather than topical. Practising long-form problem-solving from Year 1, and teaching follow-through technique explicitly so students can move on and still score, are the most effective preparations.
What is the weighting of the IA in IB Math AI?
The internal assessment, called the mathematical exploration, is worth 20 per cent of the final Math AI grade at both Standard Level and Higher Level under the 2021 syllabus. It is marked against five criteria: presentation, communication, personal engagement, reflection, and use of mathematics, totalling 20 marks. Explorations grounded in real, student-collected data consistently outperform recycled or abstract topics.
Pranjal Gupta
Curriculum-focused EdTech specialist
Pranjal Gupta writes for AssessPrep on IB Diploma Programme assessment design and digital exam readiness. He works with DP Coordinators across 85+ countries on the Math AA and Math AI courses, GDC workflows, and the transition to on-screen assessment.
Articles
Read more from our blog
The DBQ Rubric: All 7 points and How to grade each one
The DBQ rubric scores essays out of 7 in APUSH, AP World and AP Euro. See each point, a real DBQ marked point by point, and 3 common marking mistakes.
How to grade with AI: A step-by-step guide for teachers
Learn how to grade with AI in 5 practical steps, from rubric setup to final review. A teacher's guide to AI grading for IB, Cambridge, and Pearson schools.
10 mins read

