Effective AI training for teachers covers four things: how to teach students to question and verify AI output, not just operate it; how to judge when AI helps a lesson and when it does not; how to roll out AI without widening the gap between well-resourced and under-resourced schools; and how to keep human judgement, not automation, at the centre of the classroom. That is the structure behind the UNESCO AI Competency Framework for Teachers, and it is also, almost point for point, the argument a Malaysian academic just made in public.
On September 11, Dr Norsharina Zabidi, a senior lecturer at Universiti Utara Malaysia's School of Business Management, published an op-ed in Malay Mail with a line worth sitting with: AI can provide information, but teachers provide wisdom. Her argument was not that AI should be kept out of Malaysian classrooms. It was the opposite. AI belongs there, she wrote, but only if teachers remain the ones steering how it gets used, rather than students and schools figuring it out unsupervised because nobody trained the adults first.
That is a more specific claim than it sounds, and it happens to double as a checklist for anyone designing teacher professional development right now.
What does "teachers must lead AI" actually require?
Zabidi's piece breaks into four concrete demands, and each one maps to a training gap rather than a technology gap.
The first is AI literacy for students, taught by teachers who have it themselves. She is specific about what this means in practice: students need to learn to question AI-generated information, identify bias, protect their personal data, and verify sources. Notice what that list does not include. It does not say students need to learn to prompt better or format essays faster. The skill she is describing is critical evaluation, and a teacher cannot teach critical evaluation of a tool they have only ever used passively.
The second is teacher training itself, described as comprehensive rather than a one-off orientation session. Understanding AI's potential is the easy half. Understanding its limitations, where it hallucinates, where it flattens nuance, where it should simply be switched off, is the half most training programmes skip because it is harder to package into a single afternoon workshop.
The third is equity. Zabidi warns that AI in schools risks widening the gap between well-funded and under-resourced institutions unless infrastructure investment happens alongside training. This is the part of the AI conversation that gets the least airtime and the most consequence. A training programme that assumes every classroom has reliable devices and connectivity will work beautifully in a handful of flagship schools and fail everywhere else.
The fourth is the one the whole piece is built around: preserving the human elements of teaching. Not as a soft, sentimental add-on, but as the actual design constraint. If a training programme cannot explain what stays with the teacher and what can be delegated to a tool, it has not finished its job.
What should school leaders do about teacher AI training?
Treat internal training as groundwork for a future mandate, not a workaround for the absence of one. Governments in the region are already investing in AI access and student certification, and teacher training is a natural next piece of that build-out. A school that builds structured AI training for its own teachers now is not getting ahead of the ministry, it is building the kind of programme a future directive would likely ask for anyway, and it arrives at that point already running rather than starting from zero.
Build the "when not to use it" module before the "how to use it" module. Every programme referenced here, Zabidi's argument, Singapore's compulsory module, treats judgement about when AI does not belong in a lesson as core content, not an afterthought. Training that only covers operation produces teachers who can use AI, not teachers who can supervise its use.
Treat infrastructure and access as part of the training rollout, not a separate budget line. A training plan that assumes uniform connectivity across a school system will visibly fail in the classrooms that need the training most. Sequence infrastructure and professional development together, or the equity gap Zabidi warns about becomes a design outcome rather than an accident.
Anchor the programme to a recognised framework rather than building one from scratch. The UNESCO AI Competency Framework for Teachers exists precisely so schools are not reinventing this structure independently. It gives a shared vocabulary for what "AI literate teacher" means, which matters when a school is trying to justify a training budget to a board or a ministry.
Frequently asked questions
What did the Malaysian op-ed about AI in classrooms actually argue?
Dr Norsharina Zabidi, writing in Malay Mail on September 11, 2026, argued that AI belongs in Malaysian classrooms but only if teachers remain central to how it is used. She called for student AI literacy, comprehensive teacher training, equity-minded infrastructure investment, and preserving the human judgement that AI cannot replace.
Is AI training compulsory for teachers anywhere in Southeast Asia yet?
Yes. Singapore's Ministry of Education requires all primary, secondary, and junior college teachers to complete one AI literacy module by the end of 2026, with two further modules recommended by the end of 2027, under its Teacher AI Literacy Professional Learning Roadmap announced in August 2026.
What is the UNESCO AI Competency Framework for Teachers?
It is UNESCO's standard structure for what AI competency looks like for educators, organised across domains and competencies that run from basic understanding through to confident, ethical classroom application. It gives schools a shared reference point instead of building teacher AI training criteria from scratch.
Does teacher AI training have to wait for infrastructure to be fixed first?
No, but they should be planned together. Rolling out AI training without addressing device and connectivity gaps tends to benefit already well-resourced schools first, which is the exact widening gap Zabidi's op-ed warns against. Sequencing the two together, rather than treating infrastructure as a later problem, is what keeps a rollout equitable.
What is Malaysia's government already doing on AI in schools?
Two things, both focused on students rather than teachers so far. The Ministry of Education launched the TVET-AI programme in May 2026 with Multimedia University and Huawei Malaysia, giving students at five fully residential schools an AI certification. In July, Prime Minister Anwar Ibrahim announced free AI software access for 100,000 young Malaysians through the Rakyat Digital platform. Neither programme has a publicly reported teacher training component yet, which is the gap Zabidi's op-ed points at.
What should a school do if there is no national AI training mandate yet?
Start with an internal programme rather than waiting. Singapore's compulsory rollout shows what that looks like once a ministry builds it at national scale, and a school that starts structured teacher training early has more control over how "when not to use AI" gets taught, not just "how to use it," and arrives ahead if a similar mandate follows.
thegurucool builds teacher AI readiness for schools through structured professional development aligned to the UNESCO AI Competency Framework for Teachers. The TEACH-AI framework develops 14 competencies across 7 domains, from first exposure to confident AI-era assessment and classroom design.
Related reading: Malaysia's AI Education Plans for 2026: What's Coming, and What Schools Can Do Now · Singapore's 2026 Plan to Cut Teacher Workload With AI: What Changed · Mandatory AI Literacy Is Here. Teacher Training Still Assumes a Computer Lab.