Forum: Maths syllabus needs to evolve to prepare students to be AI-ready

I applaud the announcements in Budget 2026 to spur artificial intelligence (AI) adoption by the workforce (PM Wong unveils $155b Budget with nationwide AI push, more support for firms and families, Feb 12).

As a computing professor and AI researcher, I can see the immense benefits that new AI tools with proper guardrails can bring.

But one area I see that needs strengthening is the mathematics syllabus in schools, especially for the years that students sit national exams. I examined the 2026 syllabuses for O-level mathematics, additional mathematics, A-level H2 mathematics and further mathematics, and found them all to be heavily geared towards engineering rather than computing/AI. Indeed, all the topics appear relatively unchanged from when I took these exams in the mid-1980s.

For computing/AI, the following topics are more foundational and useful, and more students should learn them at an early age: number theory (including modulo arithmetic and number bases), logic and proofs, set theory, matrix algebra, combinatorics, orders of growth, probability and statistics, calculus, basic graph theory and optimisation.

Singapore’s current syllabuses teach “traditional” topics more appropriate for those pursuing an engineering/science major: complex numbers, numerical methods, differential equations, advanced integration and hypothesis testing. These may be omitted or offered as specialised topics meant for academically stronger students.

Many university computing students who progress through the current syllabuses (even those who aced H2 mathematics) struggle in their first course in discrete mathematics, which is the foundation of computing/AI.

I have observed this from teaching such a course over many years. Pre-university students are well trained in using formulas and algorithms to calculate answers, and are very quick to do so. However, when asked why adding two odd numbers will always yield an even number, they are often unable to articulate simple proof. They have mastered the skills to do the “how” at the expense of learning to ask “why”.

When AI can exceed humans in doing the “how”, it becomes more important for humans to ask the “why”.

To be fair, I see that the Ministry of Education has introduced computational thinking (which should not be conflated with coding) to primary school pupils. While this is helpful, I feel it is more impactful to revise the mathematics syllabuses at both the O and A levels to prepare all students to be AI-ready (regardless of what majors they may pursue).

Terence Sim