← All notesMaths AI · IB Maths AI Exam Traps: Mistakes Examiners Flagged⬇ Download PDF
IB DP Maths AI · Exam special · SL & HL · from the Nov 2024 and May 2025 subject reports · ~11 min read

IB Maths AI Exam Traps: Mistakes Examiners Flagged

The mistakes IB examiners flagged in the November 2024 and May 2025 Maths AI subject reports, grouped by topic, each with the fix, plus a last-week checklist.

Accuracy & roundingFinance (TVM)Models & functionsVoronoi & geometryStatisticsProbabilityHypothesis testsCalculus
RSby Riyaz Sir
mentorriyaz.in
No tag = SL and HLHL ONLY HL students onlyFORMULA BOOKLET given in the examM1A1 how marks are awardedhighlight = comes up again and againHANDY worth memorisingTest yourself at the end

In 30 seconds

  • Every trap below was named by IB examiners in the Maths AI subject reports for November 2024 and May 2025 (all time zones, SL and HL).
  • AI is not "just the calculator". The May 2025 HL report called it "a wake-up call": core algebra, interpretation and clear working decide the grade.
  • The biggest single loss is accuracy: answers to 2 significant figures, ignoring "two decimal places" (even when it was in bold) and rounding in the middle of a question.

By the end you can

  • Spot the most common ways IB Maths AI students lose marks.
  • Replace each one with the habit examiners want to see.
  • Use a last-week checklist before every paper.

1Accuracy and working

Common mistake2 s.f. answers. The November 2024 SL report is blunt: an unsupported final answer to lower accuracy than instructed gets zero marks, because it doesn't show a correct method or answer. Fix: write the unrounded value, then 3 s.f.
Common mistakeIgnoring the stated accuracy. Money to 2 d.p., "nearest integer", "4 s.f.", "as a percentage" or "nearest minute": all were ignored in May 2025. Fix: underline them in reading time.
Common mistakeTruncating instead of rounding. 24.28 read as 24.2, or 2.836 written as 2.83.
Common mistakeRounding in the middle. Rounding a normal probability to 0.0339 and then using it in a binomial calculation gave the wrong final answer. Fix: keep full GDC values until the end.
Common mistakeToo many digits. Giving a distance as 13 865.42462 km (to the nearest cm!) was penalised in May 2025 HL. Fix: 3 s.f. unless the context needs more.
Common mistake"Used GDC" as working. Recording button presses is not working; neither is a bare answer. Fix: write the equation, the distribution, or the TVM values, then the answer.
Common mistakeDegree/radian mode. AI SL trig is in degrees; a GDC that has been reset is usually in radians. HL radian questions done in degrees gave a flat-line graph with "no turning points".
Common mistakeCoordinates when only x is asked for (and vice versa). "Write down the x-values" means two numbers, not two points.

2Finance (TVM solver)

Common mistakeCompound interest formula when there are regular payments. If PMT ≠ 0, use the finance solver.
Common mistakeSigns. PV and PMT must have opposite signs for a loan (money in vs money out).
Common mistakeN in years instead of payments. A 5-year loan with monthly payments has N = 60, not 5.
Common mistakeFV for a repaid loan. It is 0. Many students left it blank or used the loan amount.
Common mistakeEarly repayment. Find the balance still owed first (FV after the payments made), then answer the new question.
Common mistakeInterest vs interest rate. "How much interest" is an amount of money: total paid − amount borrowed.
Common mistakeP/Y and C/Y. "Quarterly" compounding means C/Y = 4. Check both values every time. See GDC guide for keystrokes.

3Models and functions

Common mistakeInterpreting a gradient. "a = 0.362" is not an interpretation. Fix: "the population increases by 0.362 million per year". Refer to both variables and the units.
Common mistakeFunction notation. h(d) does not mean h × d, and f⁻¹(x) is the inverse, not the derivative.
Common mistakeInverse functions in context. For C⁻¹(1270), 1270 is the cost, so you're finding the distance that costs 1270.
Common mistakeIgnoring the context. An answer below 8.5 m when the dolphin was given as still above the water at 8.5 m. Fix: ask "is this realistic?" before you move on.
Common mistakePiecewise models. Confusing < and ≤ at the boundary, or substituting a value outside the piece's domain.
Common mistakeSinusoidal models. The amplitude is (max − min)/2, not the max; the principal axis is a horizontal line y = d; the period is twice the time from a maximum to the next minimum, not 365 because the graph "looks like" a year.
Common mistakeTime answers. 14.6 hours after midnight is 14:36. Convert decimal hours and give a clock time when asked.
Common mistakeSequences in context. Using n = 10 when the account starts at n = 0, and "complete years" means rounding down.

4Voronoi diagrams and geometry

Common mistakeThe boundary between two sites is the perpendicular bisector of the segment joining them, not the line through them.
Common mistakeMidpoints as two numbers. Write the midpoint as a coordinate pair (4, 3).
Common mistakeCalculating every distance when the Voronoi diagram already tells you the nearest site. Use the diagram.
Common mistakeCoverage questions. "Is the whole region within 3 km?" needs an explicit comparison: 2.83 < 3, so yes.
Common mistakeArea units. If 1 unit = 2.5 km, areas scale by 2.5² = 6.25, not 2.5.
Common mistakeNon-right triangles. Don't assume a right angle or an isosceles triangle. Use the cosine rule when the sine rule doesn't have enough information.
Common mistakeShortest distance from a point to a line is the perpendicular distance, not the distance to a midpoint or along an angle bisector.
Common mistake"AB" means the length, not the vector. About a third of HL students misread this in May 2025.

5Statistics

Common mistakeSpearman's rank with labels. Sheepdogs A, B, C are names, not ranks. Rank the scores, and give tied values the mean of their ranks.
Common mistakePearson's r to 2 s.f. r = 0.99 lost a mark; give 3 s.f. (0.993).
Common mistakeOutliers. Use Q₁ − 1.5 × IQR and Q₃ + 1.5 × IQR (not the median), mark each outlier separately on the box plot, and end the whisker at the last value that isn't an outlier.
Common mistakeMean from a frequency table by hand. Slow and error-prone. Use 1-variable statistics with a frequency list; check whether you need σₓ or sₓ.
Common mistakeExtrapolation. A prediction outside the data range is unreliable, whatever the correlation coefficient. Checking the model at one point doesn't make it reliable.
Common mistakeSampling methods. Stratified, systematic, quota and convenience were confused in several reports. State a specific disadvantage, not just "biased".

6Probability and distributions

Common mistakeBinomial conditions. Say them in context: a fixed number of trials, two outcomes, constant probability, independent trials.
Common mistake"At least 4". P(X ≥ 4) = 1 − P(X ≤ 3), not P(X ≤ 4).
Common mistakeThe normal distribution is continuous. P(X > 45) is not P(X ≥ 46).
Common mistakeDiscrete distributions. Don't assume all outcomes are equally likely, and don't use E(X) = np unless X is binomial.
Common mistakeFair games. Fair means the expected gain (winnings − cost) is 0.
Common mistakeNormal sketches. Bell-shaped, symmetric, labelled with the mean, the region shaded. invNorm gives the value to the left of the area.

7Hypothesis tests

Common mistakeχ² hypotheses. "No relationship" or "doesn't influence" is wrong. Write: H₀: A and B are independent. H₁: they are not independent.
Common mistakeConclusions without context. "Reject H₀" alone loses the mark. Say what it means for the situation.
Common mistakeWrong comparison. Compare the p-value with the significance level, or χ²_calc with the critical value. Don't compare a p-value with a critical value.
Common mistakeDegrees of freedom. Contingency table: (r − 1)(c − 1). Goodness of fit: categories − 1, minus 1 more for each parameter estimated from the data (HL).
Common mistakeExpected frequencies below 5. Combine categories before testing.
Common mistaket-tests. Choose one-tailed or two-tailed from the wording of H₁ ("greater" vs "different"); state the assumptions (normal populations). See AI write-up templates.

8Calculus

Common mistakeDerivatives of terms like 5/x². Rewrite as 5x⁻² first.
Common mistake"Increasing at x = 3.5". You must state that the gradient is positive there, not just give the number.
Common mistakeThe constant of integration. Find it from the context (for example V = 0 when w = 0) and include it.
Common mistakeMaximum of a function vs maximum of its derivative. The optimum is where the derivative is 0, not where the derivative is largest.
Common mistakeTrapezoidal rule brackets. ½ × h × [(y₀ + yₙ) + 2(y₁ + … + yₙ₋₁)]: the large bracket matters.
Common mistakePercentage error. Divide by the exact value; the approximate value goes in the numerator difference only.

9HL only HL

Common mistakeMatrices. Order of multiplication in composite transformations; using the inverse matrix to reverse a transformation; reflection and rotation matrices skipped by some schools.
Common mistakeMarkov chains. sₙ = Tⁿs₀ (not Tⁿ⁻¹); each column of T adds to 1; give the steady state as a vector, the eigenvector for λ = 1.
Common mistakeGraph algorithms. Show the method for Prim's (order of edges, stop after n − 1 edges); don't confuse it with nearest neighbour; for the deleted vertex bound, add the two shortest edges.
Common mistakeAdjacency matrices. Closed walks of length 4: add the diagonal of A⁴; a zero entry in A³ means no walk of length 3.
Common mistakeCoupled and second-order DEs. Find eigenvectors in the form (1, k); write Euler steps with the question's variables, show the first step.
Common mistakeSeparation of variables. Missing the minus sign when integrating 1/(1 − P), and substituting L = 0 instead of L = L₀.
Common mistakeConfidence intervals. Use t (not z) when σ is unknown; write the interval as a < μ < b; s²ₙ₋₁ is in the formula booklet.
Common mistakeLinear combinations. Var(X − Y) = Var(X) + Var(Y). Variances add even when you subtract.
Common mistakeSample means. X̄ ~ N(μ, σ²/n); forgetting the √n.
Common mistakeType I and II errors. A Type I error rejects a true H₀; its probability is the actual significance level of the critical region.
Common mistakeHypotheses in words. HL reports prefer symbols: H₀: μ = 25, H₁: μ > 25, or H₀: ρ = 0.

10Last-week checklist

  • Before each paper: angle mode set (degrees for most SL trig); finance app found; lists cleared; spare batteries.
  • During: about 1 minute per mark; attempt every part; use the given answer to re-enter; write all GDC inputs.
  • Every answer: stated accuracy; units; context in conclusions; a reality check.
Linking questionFor calculator steps see the GDC guide; for model answers to interpretation questions see AI write-up templates; for how much these marks are worth see the grade boundaries.

Test yourself — cover the answers first

Quick check

  1. A loan lasts 4 years with monthly payments. What is N?
  2. What is the boundary between two Voronoi sites?
  3. 1 grid unit is 2 km. A cell is 7 square units. What is its area?
  4. Write H₀ for a χ² test on gender and sport choice.
  5. X ~ N(50, 4²). Is P(X > 45) the same as P(X ≥ 46)?
  6. HL How do you find the variance of X − Y for independent X and Y?
Show answers
  1. 48
  2. the perpendicular bisector of the segment joining them
  3. 28 km²
  4. gender and sport choice are independent
  5. no; the normal distribution is continuous
  6. Var(X) + Var(Y)
Practise this chapter
IB Maths practice papers
Stuck on a question?

Send it to me on WhatsApp. I read every message.

Want regular help? Send this to your parent →
Share with a friendAsk Riyaz Sir

Independent notes written by Riyaz Sir. Not produced, endorsed or approved by the International Baccalaureate Organization. Syllabus codes refer to the IB DP subject guides.