Teacher AI Competence in Asia
A teacher completes a two-hour generative AI workshop. She can write a prompt, generate a lesson outline and demonstrate several popular tools. The school’s professional-development dashboard now records another teacher as “AI trained”.
But what happens when an AI-generated explanation contains a subtle error? Can she recognise when student information should not be entered into a tool? Can she decide when AI improves a lesson and when it simply makes a task faster? Can she design student work that still requires thinking when generative AI is readily available?
Those questions are becoming more important than whether teachers have attended AI training.
Artificial intelligence is already part of teachers’ professional environment. The OECD Digital Education Outlook 2026 reports that 37% of lower-secondary teachers covered by TALIS 2024 used AI for their work. At the same time, international guidance increasingly emphasises that effective AI use requires professional judgement rather than simple familiarity with software.
For schools across Asia, the next challenge is therefore measurement: how can leaders tell whether teachers are actually becoming competent users of AI?
AI Usage Is Not the Same as AI Competence
Schools tend to measure technology adoption through indicators that are easy to count: workshop attendance, training hours, accounts activated, prompts submitted or the number of teachers using an approved tool.
These indicators answer an implementation question. They tell leaders whether teachers have encountered the technology.
They do not answer the educational question: can those teachers make sound professional decisions when AI is involved?
A teacher who generates every worksheet with AI may be a frequent user without being a competent user. Another teacher may use AI relatively rarely but demonstrate strong judgement about accuracy, privacy, assessment integrity and when human interaction matters more than automation.
Frequency is therefore a weak proxy for competence.
UNESCO’s AI Competency Framework for Teachers provides a more useful starting point. It describes 15 competencies across five dimensions: a human-centred mindset, ethics of AI, AI foundations and applications, AI pedagogy, and AI for professional learning. It also describes development through Acquire, Deepen and Create levels.
A September 2026 UNESCO IITE analysis went further by reviewing 36 teacher AI competency frameworks and examining how teacher competence can be implemented and evaluated. The growing attention to assessment is significant: education systems are moving beyond asking whether teachers have access to AI toward asking what competent practice actually looks like.
The Five-Evidence Test for Teacher AI Readiness
Schools do not need to reproduce an international competency framework as an internal scoring system. They can instead look for five forms of evidence that reveal whether teachers are becoming capable, responsible AI users.
| Evidence area | Question for school leaders | What stronger evidence looks like |
|---|---|---|
| Understanding | Does the teacher understand what the technology can and cannot reliably do? | The teacher can identify limitations, check outputs and explain why verification is necessary. |
| Judgement | Can the teacher decide when AI should or should not be used? | AI use changes according to the learning objective rather than being added automatically. |
| Pedagogy | Does AI strengthen learning rather than simply accelerate task completion? | Lesson design preserves explanation, reasoning, practice, feedback and student thinking. |
| Responsibility | Can the teacher recognise privacy, bias, attribution and integrity risks? | The teacher follows appropriate safeguards and can explain those decisions to students. |
| Adaptation | Can the teacher evaluate a new AI tool without depending entirely on vendor claims? | The teacher tests usefulness against learning needs, evidence, risk and institutional policy. |
The purpose is not to create another compliance checklist. It is to make competence visible through professional decisions.
1. Test Understanding With Imperfect Outputs
A demonstration in which teachers reproduce a prepared prompt reveals very little. A more revealing assessment gives teachers an imperfect AI output and asks what they would do with it.
For example, provide an AI-generated explanation of a familiar topic containing an unsupported assertion, an oversimplification or a source that needs verification. Ask teachers to identify what should be checked before the material reaches students.
This tests something more important than prompting technique: whether the teacher treats generated material as something requiring professional review.
The same principle applies to lesson plans, quiz questions, translations and feedback. Competence becomes visible when teachers can interrogate an output rather than merely produce one.
2. Assess Decisions, Not Just Technical Skills
Many AI-training programmes naturally begin with operation: opening a tool, entering instructions, refining a response and exporting material.
Schools should eventually test a harder ability: deciding whether AI belongs in the task at all.
Give teachers several scenarios. One might involve generating differentiated practice questions. Another might involve providing feedback on identifiable student work. Another could involve students using AI during an assessment intended to reveal independent reasoning.
Instead of asking, “Which prompt would you use?”, ask, “Would you use AI here, under what conditions, and why?”
Teachers do not need identical answers. Context, age group, subject and institutional policy matter. What leaders should look for is defensible reasoning.
3. Look for Evidence in Lesson Design
AI competence should eventually appear in teaching rather than remaining inside a training room.
A lesson observation does not need an “AI score”. Leaders can instead examine whether technology use supports the intended learning.
If AI provides students with immediate answers, what intellectual work remains for the learner? If it generates examples, does the teacher verify their suitability? If students compare AI responses, are they required to evaluate evidence, assumptions or reasoning? If AI assists teachers with planning, does professional judgement remain visible in the final lesson?
This distinction matters because successful completion of a task with generative AI does not necessarily mean that learning occurred. OECD’s 2026 digital-education analysis makes that distinction explicitly when discussing generative AI in education.
The same standard should apply to teachers: efficiency is useful, but efficiency alone is not pedagogy.
4. Make Responsible Practice Observable
Responsible AI training often becomes a presentation about privacy, bias and academic integrity. Schools can strengthen it by asking teachers to demonstrate decisions around those issues.
Can teachers distinguish information suitable for an AI tool from data that should remain inside protected institutional systems? Do they know how to respond when an output reproduces stereotypes or makes unsupported claims? Can they explain acceptable AI assistance to students before an assignment begins rather than investigating misuse afterwards?
UNESCO’s teacher framework places both human-centred practice and AI ethics alongside technical knowledge and pedagogy. That is an important signal: responsible use is not an optional compliance layer added after teachers learn the technology. It is part of competence itself.
5. Test Whether Competence Transfers to New Tools
Product-specific training expires quickly.
A teacher who has memorised the controls of one platform may need retraining when the interface changes. A teacher who understands verification, data sensitivity, instructional purpose and appropriate human oversight can apply those principles to tools that did not exist when the original training occurred.
This provides schools with another useful assessment method: introduce an unfamiliar AI application and ask teachers how they would evaluate it before classroom adoption.
They should be asking what the product does, what information it receives, what evidence supports the educational claim, what human review remains necessary and whether its use serves a genuine learning purpose.
The ability to ask good questions about unfamiliar technology may ultimately be a stronger sign of readiness than expertise in today’s most popular tool.
From Training Records to an Evidence Portfolio
A school does not need one high-stakes AI examination for teachers. A small evidence portfolio is likely to provide a more realistic picture.
- A short scenario-based judgement task can test ethical and pedagogical decision-making.
- An annotated AI output can demonstrate verification and critical evaluation.
- A lesson or activity plan can show how AI supports a defined learning objective.
- A classroom observation or professional discussion can reveal whether planned practice transfers into teaching.
- A reflection on an unsuccessful AI use can demonstrate professional learning rather than simply successful tool operation.
Schools can collect such evidence periodically instead of requiring everything at once. The objective should be developmental: establish what teachers can currently do, identify the next capability and provide opportunities to practise it.
This also complements broader teacher professional development. Training attendance remains useful evidence that an opportunity was provided, but it is only the beginning of the evidence chain. Asia Education Journal has previously examined the same problem across professional development more broadly: participation does not by itself demonstrate changed classroom practice.
Do Not Collapse AI Readiness Into One Percentage
Institutional dashboards encourage leaders to create a single figure: “82% of teachers are AI ready.”
That may be convenient, but it can conceal important differences.
A school might have teachers who understand responsible use but need stronger pedagogical applications. Others may be highly capable users who need better privacy practices. Department leaders may need competency in procurement and policy decisions that classroom teachers do not require at the same depth.
A profile is therefore more useful than a single readiness score.
School leaders can ask where capability is strong, where risk remains and what type of support is required next. The purpose of measurement is improvement, not creating a headline number.
When an Institutional AI Initiative Becomes a Measurable Achievement
Most teacher AI-development programmes belong in normal school improvement reporting. Training a cohort of teachers does not automatically constitute an exceptional achievement, and a large participation number does not prove educational quality.
Occasionally, however, a school group, university, education company or training provider may reach a separate milestone that is unusually large, sustained, specific and objectively verifiable. That may involve the scale of a defined initiative, its duration, participation under clearly stated conditions or another measurable institutional result.
Organisations exploring record recognition Asia should keep that claim separate from teacher competence.
Asia Record official guidance states that proposed achievements should be clearly defined, measurable and supported with evidence for verification. An organisation considering an Asia record application or planning to apply for Asia Record should therefore document the metric, conditions and evidence from the beginning rather than reconstructing them after an initiative becomes successful.
Likewise, Asia record certification of an approved milestone would document that particular achievement. Becoming an Asia Record holder would not certify the quality of the school’s AI curriculum, accredit its teachers or prove that every participant has achieved AI competence. Record certification Asia and educational quality assurance answer different questions.
Institutions researching how to get an Asia Record should preserve that distinction. Independent recognition can document an exceptional measurable fact; evidence of teacher capability must still come from what teachers understand, decide and do.
The Better Question Is What Teachers Can Now Do
AI professional development will increasingly become a normal part of education. That makes attendance numbers progressively less informative.
A stronger school should be able to explain what teachers can now do that they could not do previously: verify generated information, recognise inappropriate uses, redesign learning activities, protect learners, make responsible tool choices and transfer those principles when technologies change.
Those capabilities are harder to count than workshop registrations. They are also considerably more valuable.
The most useful measure of teacher AI readiness is therefore not the number of teachers who have completed AI training or even the number using AI every week. It is the quality of professional judgement teachers can demonstrate when AI becomes one possible participant in the educational process.