IB Math AA vs AI: What the Grade Boundaries Actually Tell You — and What They Don't

By Renzo Diaz··8 min read

Grade boundaries are not a difficulty score — they move from session to session as part of the IB's own grade-award process, so a course with a higher boundary is not automatically the harder one. The AA-versus-AI decision should be driven by university requirements, course fit, and realistic grade potential, not by which course has the lower number attached to a 7.

"Is AA harder than AI?" is probably the most common question families ask when choosing between the two IB Mathematics courses. It is also not the best question. A better one is: which course gives this particular student the strongest combination of university options, academic fit, and realistic grade potential? Those are not the same question, and answering the first one usually gets in the way of answering the second.

One reason the debate gets so confused is that students use grade boundaries as if they were difficulty scores. They are not. A Grade 7 boundary of 72% does not mean a course is harder than one with a 79% boundary, and it does not, by itself, tell you which syllabus is harder. The IB's own grade-award process explains why — and that distinction changes how this decision should actually be made.

The interesting question is not "which course has the lower boundary?" It is "what do the boundaries actually tell us — and what are we incorrectly asking them to tell us?"

1. A grade boundary is not a difficulty score

Start with the basic definition. After a student's component marks are aggregated and weighted, they are mapped onto the IB's one-to-seven grading scale, and the ranges corresponding to each grade are the grade boundaries. The important part is how those ranges get set. The IB's own research on the process describes grade award as "a rigorous process that involves the analysis of a range of different evidence to ensure the fairest and most comparable outcomes for students" — including statistically recommended boundaries alongside the judgement of senior examiners reviewing candidate work against grade descriptors.

A separate description of the same process puts it more bluntly: boundaries are set through "a combination of judgment, statistical evidence, grade descriptors and the G2 forms submitted by teachers." That is not "percentage correct equals difficulty." It is the outcome of an assessment process designed to hold a grade to a comparable standard across different exam papers and different years — which is exactly why the same percentage can mean different things in different sessions.

“Marks are how far a candidate has walked, but grades take into account how steep the path was.”

IB Assessment Guide for Teachers and Coordinators

2. Why the ranking keeps changing

If grade boundaries measured pure difficulty, the same course would always carry roughly the same boundary, and one course would consistently sit above the other. That is not what publicly tracked historical boundary data for AA and AI actually shows. Across different sessions, sometimes AA's Grade 7 boundary sits higher than AI's; sometimes AI's sits higher than AA's. There is no stable ranking to read off.

That is exactly what the grade-award process would predict, not a contradiction of it. Each session, each paper is judged on its own merits — how the candidates who sat it actually performed, and how demanding examiners judged that specific paper to be — and the boundary is set to hold the grade to a consistent standard, not to preserve a fixed percentage. A boundary that moves is the system working as designed, not a sign that the underlying course got easier or harder.

A boundary that moves from one session to the next is the system working as designed — not proof that a course got easier or harder.

3. The trap of comparing the wrong paper

This matters most in practice when students self-mark past papers. Large-entry subjects like Mathematics are examined across multiple time zones within the same session, which in practice means different candidates can sit different versions of the same paper. Each version gets its own grade boundary, set independently through the same process described above.

That means a percentage score on one version of a past paper does not automatically tell you what grade the equivalent score would earn on a different version — even from the same session. It is not that the IB is being inconsistent; different paper versions can have genuinely different assessment characteristics, and the grade-award process exists specifically to account for that. The practical habit is simple: when you check a past paper against a grade boundary, use the boundary published for that exact paper and session, not a generic table pulled from a search result.

4. So, is AA harder than AI?

Grade boundaries cannot answer that question. But that does not mean AA and AI are essentially the same course wearing different names — they are deliberately built around different mathematical emphases.

Mathematics: Analysis and Approaches leans into analytical and algebraic work — functions, trigonometry, calculus, and mathematical investigation, conjecture and proof, with real weight given to constructing and justifying a mathematical argument. Mathematics: Applications and Interpretation leans into mathematics in context — modelling, statistics, and the extensive use of technology to explore and build mathematical models. The difference is not "hard mathematics versus easy mathematics." It is closer to different kinds of mathematical thinking, applied to different kinds of problems.

The difference between AA and AI is not hard mathematics versus easy mathematics. It is different kinds of mathematical thinking.

5. AA and AI reward different strengths

This gives a far more useful way to compare the two courses than reading off a boundary. Ask which mathematical profile actually matches the student, not which course has the better reputation.

Neither profile is more capable than the other. They are different mathematical strengths, and the course that fits one profile poorly can still be a genuinely strong route for the other.

Layer 1 — Likely to find AA more natural

Comfortable with symbolic manipulation and algebra, enjoys pattern recognition and generalisation, at ease with abstract relationships and calculus, and can follow or construct a formal proof.

Layer 2 — Likely to find AI more natural

Enjoys seeing mathematics used in real-world contexts, likes solving practical and data-driven problems, strong with statistics and interpreting data, and learns well through technology-assisted exploration.

6. The myth that AI is "the easy option"

In some schools, AI gets treated as the mathematics course for students who are "not good at maths." Publicly tracked historical boundary data does not support using the boundary as evidence for that reputation — AI's Grade 7 boundary has sat both above and below AA's across different sessions, with no consistent hierarchy in either direction.

The reputation survives anyway because it is easier to repeat than to check. But the grade-award process is explicit that boundaries are set to produce comparable standards for that specific paper, not a permanent difficulty ranking of the syllabus behind it. Calling AI "easy" because it has a reputation for being more applied is a shortcut, not a finding. The course asks students to demonstrate mathematics differently — not less.

7. HL versus SL is a different decision

A separate distinction gets mixed into this debate: AA-versus-AI and HL-versus-SL are not the same choice. HL courses are studied with more depth and breadth than SL — the IB's own framework specifies 240 recommended teaching hours for an HL subject against 150 for SL.

That means the HL decision should not be reduced to "can I handle harder maths?" It is also a question of total workload across the whole Diploma. A student taking several demanding HL subjects is not making a mathematics decision in isolation; they are choosing how to allocate a finite amount of attention across the entire programme. A mathematically capable student does not automatically benefit from HL if their target courses do not require it — and choosing SL purely because it feels safer can create a real problem later if a target university requires HL.

8. The university question comes before the prestige question

University requirements are not uniform, which is precisely why blanket statements like "you need AA for economics" are too broad to be useful. The same university can require AA specifically for one programme while accepting either syllabus for another, and preferences shift by year and by department. A student should not discover the distinction between AA and AI after subjects are already chosen.

The specific requirements — which programmes require AA outright, which accept either syllabus, and which merely prefer one — are detailed in what universities actually require for IB Math AA vs AI, checked against current admissions pages rather than reputation.

9. The strategic case for AA

AA tends to make sense when three conditions hold together: the student's realistic target courses require or strongly prefer it, the student is genuinely comfortable with its mathematical style — algebraic manipulation, calculus, pattern recognition, abstraction, formal argument — and the student can sustain the workload alongside the rest of the Diploma. That is not an argument that AA is "better." It is an argument that AA can be the more strategically appropriate route for a specific student, and the difference between those two claims matters.

Choose AA because it fits the destination and the student — not because it carries a reputation for being the "smart" course.

10. The strategic case for AI

AI tends to make sense when the target university courses accept it and the student performs particularly well in modelling, statistics, contextual problem-solving and technology-supported work. This is not "choosing the easier option" — it is choosing a course whose mathematical demands line up with the student's actual strengths and future plans. For the right student, AI is an extremely strong mathematics route, not a fallback.

11. The most dangerous time to discover you chose the wrong course is two years later

Imagine a student who chooses AI SL because "AI is easier," earns strong grades, and then decides two years later to apply for a programme that requires AA HL. That is a problem that did not need to exist. The reverse happens too: a student chooses AA HL for its reputation, spends disproportionate effort surviving a course that does not match their mathematical strengths, and quietly sacrifices performance elsewhere in the Diploma to do it. Neither outcome is desirable, and both are avoidable if the subject choice is made before reputation and boundaries become emotionally loaded.

12. What should actually determine the decision

In practice, I use this order, and would encourage any family to use something close to it.

Layer 1 — University constraints

List the degree programmes the student may realistically apply to and check their published, current IB requirements directly — not what a friend, forum or schoolmate remembers.

Layer 2 — Course fit

Once you know which routes stay open, compare AA and AI based on how the student actually thinks and performs, not on reputation.

Layer 3 — Realistic grade potential

Ask where the student is more likely to produce high-quality work consistently — not which course has the lower boundary.

Layer 4 — Total Diploma workload

Consider the student's entire programme, not mathematics in isolation.

Layer 5 — Preference

Only once the constraints above are settled should personal preference become the deciding factor.

13. What grade boundaries are actually good for

None of this makes grade boundaries useless — it means they are useful for a narrower purpose than the one families usually ask them to serve. They are genuinely informative about how far a specific score sat from a specific grade on a specific paper, why the correct timezone and session matter when checking a past paper, and how much a threshold can move between sessions. What they cannot responsibly tell you is a permanent ranking of course difficulty. That is more information than a single number was ever designed to carry.

Sources

Sources 1 and 2 below are official IB documentation. The "walked/steep path" quote in Section 1 comes from the IB's own Assessment Guide for Teachers and Coordinators — a school- and teacher-facing document without a stable public IB web address, checked against multiple independent copies before being quoted here. The AA and AI course descriptions in this article are Renzo's own paraphrase of the publicly described course emphases, not a verbatim citation. This article deliberately omits session-specific numeric grade-boundary tables: the IB does not publish per-timezone boundaries on a public page, and no independently verifiable public source could be confirmed for specific recent figures — the argument here rests on the documented methodology and on publicly tracked historical patterns, not on unverifiable exact numbers.