86% of the students in an urban school just outside Kuala Lumpur had failed mathematics. I stood stunned as the realisation slowly sank into the depths of my stomach.
What I witnessed was real.
Over the next decade, I visited many schools, and in most of them, we observed the same trend in mathematics education: most students are struggling. Malaysia’s last Sijil Pelajaran Malaysia (SPM) results confirmed this, with 52.1% of SPM candidates failing or struggling with maths.1 But why do students continue to struggle year after year with no end in sight?
For years, I have observed students, teachers, and school systems following the well-meaning interventions conducted by the Ministry of Education and non-profits in our public education systems. However, the needle hasn’t moved, at least not at the macro level. After a decade of observations, I have concluded that the fundamental issue we must address first is the high student-toteacher ratio.
Although the data suggests Malaysia has a studentteacher ratio of 18:12, this does not mean a teacher only handles 18 students. Instead, this ratio indicates that at any given time in school, there are 18 students with one teacher. In reality, teachers might teach five different classes during the school week, which translates to each teacher handling, on average, 90 students. However, I have rarely seen teachers manage so few students.
When more than 80% of students regularly fail maths, teachers face the daunting challenge of needing to assist a number comparable to that of a cinema full of children.
In the urban schools I have visited, it is more common to see teachers managing five classrooms with 40 students each, resulting in a total of 200 students for every teacher. When more than 80% of students regularly fail maths, teachers face the daunting challenge of needing to assist a number comparable to that of a cinema full of children.
Without first addressing the student-teacher ratio, teachers feel overwhelmed, systems are stretched to their limits, and administrators struggle to make training effective. Well-meaning initiatives are destined to flounder in a deluge of students who demand a teacher’s attention far more urgently than any project or mandate, as projects or mandates do not stare back into a teacher’s soul with pleading eyes that silently scream, “Help me.”
Teachers cannot help so many students, which means most students do not receive the personal attention they need to thrive. Overhauling our education system to lower the student-teacher ratio is an expensive endeavour that no developing economy can afford, but this could change in just a few years.
I see two solutions for addressing the impossibly large student-to-teacher ratio in every developing nation. The first is a technological solution, while the second arises from community activism involving an unexpected demographic. These two solutions have the effect of flooding our education ecosystem with teachers.
To promote long-term accountability, I suggest a third solution rooted in political will. All three options have the potential to be straightforward and can be executed in just a few years, benefiting millions of students in the Southeast Asian region.
Education for all, powered by AI
An AI tutoring platform, like Tupai.ai, helps tackle educational challenges in developing countries by addressing teacher shortages and resource constraints. Through personalised learning, it empowers students to build essential skills and knowledge. Scalable and accessible, this technology is transforming education, paving the way for brighter futures in underserved regions.
Source: cilisos.my
Overhauling our education system to lower the student-teacher ratio is an expensive endeavour that no developing economy can afford, but this could change in just a few years.
THE FIRST SOLUTION: Affordable AI tutors for every student
I would like first to address the elephant in the room—will AI replace teachers?
No, it can’t.
If replacing our parents with AI is unimaginable, the same is true for replacing teachers with AI. The most crucial role of a teacher is the love, nurturing, and care they provide to a young, impressionable mind—something AI cannot offer. The problem is that our time-strapped teachers, with their hundreds of attention-deprived students, struggle to dedicate the time and space necessary to love and care for our children.
Instead, AI is used as a tool, not to replace teachers but to liberate them from the drudgery of their daily routines, allowing them to love once more.
Personal attention is the most valuable asset when it comes to helping students thrive. However, in Malaysia, students seldom receive enough individual attention. To seek more attention, 88% of school students attend tuition.3 However, affordable tuition solutions place students in after-school classrooms with 20 to 40 others, resulting in minimal personal attention. Conversely, personal or private tutoring can cost hundreds per subject, making it inaccessible for large segments of the student population. This is where generative AI comes into play.
Let’s first discuss what a purpose-built AI tutor can offer to students. Students can:
- Learn with AI tutors using natural language conversations in their native language.
- Ask clarifying questions and receive thorough explanations.
- Do practice problems while discussing the problems.
- Prepare for exams with detailed step-by-step solutions.
However, the most significant role of an effective AI tutor is to uncover why students face difficulties. When a student struggles to grasp concepts, an effective AI tutor embarks on a journey to determine whether the student has missed earlier teachings. Once a knowledge gap is identified, the AI will provide instruction to fill the gap.
When custom AI models can instinctively detect and understand a student’s difficulties in mastering concepts, rather than relying on structured assessments, AI tutors will become an essential tool for both students and teachers.
What about the cost of having AI tutors? This will become significantly cheaper in the coming years, as demonstrated by OpenAI’s GPT-4o, which has a fivefold price reduction compared to the often less capable GPT-4.
However, running AI systems incurs costs and cannot be free for the general public unless initially subsidised by governments or non-profits. While AI tutors are comparatively more affordable than traditional alternatives, they remain costly when scaled to a million students or more. The bottom 20% of the socioeconomic ladder cannot afford such interventions. For them, a different intervention is needed. This is where the second solution comes into play.
The rise of peer tutoring
In Southeast Asia, peer tutoring is gaining popularity as a way to enhance learning. Students collaborate to strengthen their understanding, especially in under-resourced areas. This approach fosters mutual support, improves academic performance, and builds stronger communities. Peer tutoring helps bridge educational gaps and encourages active, collaborative learning across the region.
Image: ImageZoo / Alamy Stock Photo
When a student struggles to grasp concepts, an effective AI tutor embarks on a journey to determine whether the student has missed earlier teachings. Once a knowledge gap is identified, the AI will provide instruction to fill the gap.
THE SECOND SOLUTION: Free tuition in all schools, powered by students
We conducted a pilot programme in two secondary schools in 2015. Both schools have an average school-wide pass rate for mathematics of only 14%.
Yes, you read that correctly. Eighty-six per cent of students had failed basic mathematics.
We selected around 50 students from these two schools and arranged for them to stay behind after school once a week for seven months to help each other with maths. No teachers or adults participated in their after-school activities.
We organised the student tutoring community into a club with leadership structures and provided them with a template for conducting their weekly sessions. During these sessions, students enhanced each other’s leadership skills, much like Toastmasters members develop each other’s public speaking abilities.4
The results were stunning. The pilot class from SMK TIS improved from an average of failing to an average of A, and the pilot class from SMK PJ progressed from failing to a C-average.
Suppose our theory of change in public schooling systems revolves around increasing the number of teachers and lowering the student-teacher ratio. In that case, student tutors represent an untapped and accessible resource.
If there are 30 student tutors in the school, what could a teacher do with them? Using these student tutors could decrease the student-teacher ratio from 200-to-1 to 7-to-1.
With these first two solutions—utilising accessible AI tutors and students as teaching resources—we would successfully staff our schooling systems with enough “teachers” and provide students with the personal attention they need to thrive in school.
We’ll need one more thing to see long-term, largescale improvements to the entire education system. Arguably, the easiest switch to flip, yet the hardest decision to make, is to make our education data transparent to everyone.
THE THIRD SOLUTION: Make education data transparent
Without data, we are working in the dark. As a result, when I returned to my alma mater this year to run a pilot programme, I was surprised to find that 65% of the top class of 14-year-olds were either failing or receiving Ds and Es. This revelation comes from what was once regarded as one of the best schools in the district. First-hand accounts from other schools indicated similar concerning figures.
Advancing education through data
The Southeast Asia Ministers of Education Organization (SEAMEO) works to improve educational data collection through initiatives like the Southeast Asia Primary Learning Metrics (SEA-PLM), which assesses student outcomes across the region. SEAMEO also supports national education data systems to inform policies and enhance quality in underserved areas, though challenges remain in ensuring equitable data practices.
Photo: FORGET Patrick / Alamy
However, I obtained the data (and similar information) solely because I was physically present in the schools or knew individuals who worked there. In Malaysia, performance data for schools is not publicly available, which is why I was so surprised to learn what New Zealand’s Ministry of Education had done.5
I could click on a hyperlink on the MOE website for any school in New Zealand to see its performance. The data—student numbers, student engagement, student performance, and the percentage of students who went on to tertiary education—was all there.
Alas, such data is not available to us Malaysians. Why?
At a public forum, I posed this question in front of hundreds of attendees: “When can our school performance data be made public?” A high-ranking ministry official responded, “We must be very careful because such data can be weaponised.”
I replied, “Data can only be weaponised if it’s bad news.”
We need data to be open because only then can the government function effectively. An effective democratic system requires the public to hold the government accountable for its actions and results. Democracy requires public participation, and access to actionable data is essential for such participation.
Without data, the public will see no problem and will not engage. I’ve shared my data with everyone I’ve spoken to, and even experienced educators were shocked to learn about the alarmingly low maths passing rates in our public schools. The public needs to know how schools are performing before they can mobilise. When the public acts, governments respond alongside them.
The public needs to know how schools are performing before they can mobilise. When the public acts, governments respond alongside them.
To improve our education system, we must start by making our school performance data accessible to the public.
But isn’t all this change frightening? Frightening to governments, frightening to teachers, school administrators, and parents? Not if we rethink how we view supporting our children in their transformation.
Our children change when we love and care for them. Likewise, change feels less daunting when we stop shouting obscenities and pointing fingers and instead begin to support and nurture those in our education systems. Change feels less frightening when we value individuals until they are ready to change.
I have proposed three ways to quickly transform an education system for the better:
- Use AI to make personal tutoring accessible to underserved students.
- Systematise and scale student clubs—some schools have successfully done this.
- Release school performance data—this can be done with a simple signature from a senior politician.
To those who argue that transforming an education system is impossible, I ask this question: What is the impossibility associated with any of the three tasks mentioned above? Therefore, is rapid, large-scale change impossible? ∞
About Tupai.ai
Tupai.ai (www.tupai.ai) is a Malaysia-based AI-powered personalised maths tutoring platform. It aims to help secondary school students improve their performance in mathematics, starting with the SPM (Sijil Pelajaran Malaysia) exams, which are crucial in the Malaysian education system. The platform uses conversational AI to teach maths based on the local syllabus while adapting to students’ progress and addressing gaps in their understanding. This approach makes mathematics more accessible and engaging, especially for students who find traditional teaching methods challenging. Tupai.ai is currently in beta and offers a variety of plans, including a free tier with limited access and a paid plan that provides unlimited usage and additional resources such as exam preparation materials and adaptive learning features. Its ultimate goal is to address the massive underperformance of students in maths across developing Southeast Asian nations. Source: Genashtim website. https://genashtim.com/news/genashtim-2023.
EDMOND YAP
Edmond Yap is the co-founder of Tupai.ai and the founder of EduNation (now BACflix). EduNation was Asia’s largest free online repository of learning videos and exercises, offering over 6,000 educational videos and thousands more exercises available for free, providing at-risk children with valuable online educational support; today, the platform serves more than 200,000 children. Edmond earned his bachelor’s degree in civil engineering from the University of Adelaide in 2002.
FEBRUARY 2025 | ISSUE 13
MIND THE GAPS; BRIDGING THE GAPS
- Laporan Analisis Keputusan SPM 2023, Kementerian Pendidikan Lembaga Peperiksaan, lp.moe.gov.my/files/spm/2024/Laporan Analisis Keputusan SPM 2023.pdf. Accessed 2 Dec 2024.
- “Pupil-Teacher Ratio, Upper Secondary.” World Bank Open Data, 2020, data.worldbank.org/indicator/SE.SEC.ENRL.UP.TC.ZS?locations=MY.
- Kenayathulla, Husaina Banu. “Teachers’ Perceptions on the Effectiveness of Private Tutoring in Malaysia.” Revue Internationale d’éducation de Sèvres, 2014, https://doi.org/10.4000/ries.3801.
- Toastmasters International—Home, www.toastmasters.org. Accessed 2 Dec 2024.
- “Ministry of Education—Education Counts.” Find Your Nearest School, www.educationcounts.govt.nz/find-school. Accessed 2 Dec 2024.
