The Global Teacher AI Competency Map: What Educators Need to Know, Do and Lead
A teacher who can generate a lesson plan with a chatbot is not necessarily ready to evaluate that plan, protect student data, preserve intellectual effort, redesign assessment or lead colleagues through difficult decisions. Operational fluency is only one small part of professional competence.
The Global Teacher AI Competency Map organizes teacher development around five durable domains: human agency, AI literacy, ethical and safe practice, AI-enabled pedagogy and assessment, and professional learning and leadership. These domains matter more than mastery of a particular interface because tools, models and policies will continue to change. A teacher who understands the principles can judge the next system; a teacher trained only in prompts may become dependent on the current one.
The map draws on UNESCO’s AI Competency Framework for Teachers, the European Commission’s DigCompEdu framework and public education guidance from several jurisdictions. It translates recurring priorities into observable practice across four working stages: Aware, Guided, Independent and Leading. Those stage names belong to this Learning Coach model; they do not replace UNESCO’s official progression of Acquire, Deepen and Create.
Why a teacher AI competency framework must go beyond tool use
Schools often begin professional development with demonstrations: how to summarize a document, create a quiz, draft feedback or produce a worksheet. Demonstrations can reduce anxiety and reveal useful possibilities, but they do not answer the harder questions. Is the output accurate? Does it fit the curriculum? What data entered the system? Which students may be disadvantaged? Has the tool improved the learning design or merely accelerated production? Can the teacher explain and defend the final decision?
UNESCO’s AI Competency Framework for Teachers provides a stronger foundation. It describes 15 competencies across five dimensions—human-centered mindset, ethics of AI, AI foundations and applications, AI pedagogy, and AI for professional learning—and organizes development through Acquire, Deepen and Create. The framework protects teachers’ rights and agency while treating AI competence as a combination of knowledge, practice, values and professional judgment.
The European Commission’s DigCompEdu framework predates the current wave of generative AI, yet remains relevant because it connects technology with professional engagement, digital resources, teaching, assessment, learner empowerment and students’ digital competence. Read together, the two frameworks show why “prompt engineering for teachers” is too narrow. Competence includes selecting, refusing, adapting, explaining and governing technology as well as operating it.
A practical map should therefore help teachers locate evidence of their current practice, identify the next developmental move and avoid false confidence based on frequency of use.
The five domains of teacher AI competence
| Domain | Central professional question | Observable evidence | Common failure mode |
|---|---|---|---|
| Human agency | Does AI strengthen human learning and decision-making rather than displace them? | The teacher defines which thinking remains human, protects meaningful interaction and checks what students can do without the tool. | Efficiency becomes the default goal, even when cognitive effort or teacher-student relationships are weakened. |
| AI literacy | Can the teacher understand and verify what the system produces? | The teacher explains probabilistic output, checks claims and sources, recognizes limitations and compares tools against the task. | Fluent output is mistaken for reliable knowledge or neutral judgment. |
| Ethical and safe practice | Are privacy, fairness, transparency, accessibility and age-appropriateness protected? | The teacher minimizes data, follows policy, evaluates bias and provides fair alternatives. | Sensitive data is entered casually, hidden inequalities grow or responsibility is delegated to the vendor. |
| AI pedagogy and assessment | Does AI use improve instructional design and preserve valid evidence of learning? | The teacher aligns use with a learning objective, varies support, redesigns assessment and examines independent performance. | AI produces attractive resources while the quality of explanation, practice, feedback or assessment remains unexamined. |
| Professional learning and leadership | Can the teacher improve practice collectively and influence responsible school decisions? | The teacher documents evidence, shares limitations, supports colleagues and contributes to policy review. | Innovation remains isolated, anecdotal or driven by enthusiastic individuals without governance. |
These domains are interdependent. A teacher cannot exercise human agency without enough AI literacy to recognize when a system is unreliable. Ethical practice influences pedagogy because privacy constraints affect which tasks and tools are appropriate. Assessment design reveals whether the intended learning remains visible. Leadership determines whether good individual practice becomes a coherent institutional approach.
Domain 1: preserving human agency in AI-supported education
Human agency means that teachers and learners retain meaningful control over educational goals, decisions and interpretations. It is not a rejection of automation. It is the capacity to decide what should be automated, what may be assisted and what must remain a human responsibility.
For teachers, agency is visible when they define the purpose before choosing the tool, inspect generated materials rather than accepting them automatically and retain accountability for the lesson. For students, agency is visible when support expands understanding without concealing whether they can explain, remember or transfer the learning. The Student AI Independence Audit logic is useful here: a strong product is not sufficient if the learner cannot reconstruct the reasoning or apply it elsewhere.
Human-centered teaching also protects relationships. Some of the most important classroom decisions depend on knowledge of a student’s history, confidence, language, motivation and response to previous support. An AI system may summarize patterns, but it does not assume the moral and relational responsibility of the teacher. The professional question is therefore not whether AI can produce a recommendation. It is whether the recommendation deserves influence, what evidence supports it and who remains accountable for the consequence.
The OECD’s Digital Education Outlook 2026 reinforces this distinction by arguing that generative AI should enrich learning rather than replace cognitive effort or the human relationships at the center of education.
Domain 2: developing AI literacy for verification, not just prompting
Teacher AI literacy includes a functional understanding of how generative systems behave. Educators do not need to become machine-learning engineers, but they do need enough conceptual knowledge to interpret output responsibly.
A language model generates likely continuations from patterns in data; it does not retrieve truth in the same way as a curated database. It may produce accurate information, outdated information, fabricated citations or an unjustified synthesis in the same confident style. Output can vary with the wording of a prompt, model version and system settings. A teacher who knows this will treat plausibility as a reason to verify, not as proof.
Competent verification is disciplinary. A history teacher checks provenance and context. A science teacher checks whether claims are supported by evidence and whether causal language is warranted. A mathematics teacher inspects assumptions and intermediate steps. A language teacher distinguishes stylistic fluency from communicative appropriateness and authorship. Across subjects, the teacher should be able to explain why a generated response is useful, incomplete or unsuitable.
The objective is calibrated trust. Automatic rejection wastes legitimate assistance; automatic acceptance transfers professional judgment to a system that cannot carry professional responsibility.
Domain 3: making ethics and safety part of ordinary planning
Ethical AI practice becomes real through routines. Before using a system, teachers should know which accounts are approved, what data may be entered, whether student work is retained, how age restrictions apply and which alternatives exist for families or learners who cannot or do not wish to use the tool.
Privacy is only one dimension. Bias can appear in representation, language, examples, disciplinary assumptions and predictions. Accessibility can improve when AI supports reading, translation or expression, but it can also deteriorate when interfaces are incompatible with assistive technology or when automated recommendations confuse disability with low ability. Sustainability, labor conditions and intellectual property may also affect institutional choices.
The European Commission’s ethical guidelines for educators on AI and data encourage educators to examine the educational purpose, human oversight, transparency, fairness, privacy, safety and accountability of systems. The practical implication is that ethics should not appear only in an annual policy presentation. It should influence lesson planning, procurement, assessment and communication with students.
Domain 4: designing AI pedagogy and valid assessment
AI is pedagogically competent only when its role matches the learning goal. The same tool may be beneficial in one phase and harmful in another. Generating multiple examples can help students compare a concept; generating the final explanation may remove the very practice the lesson was meant to develop. Automated feedback can increase opportunities to revise; generic feedback may distract students from the misconception that matters most.
Teachers therefore need to decide whether AI will act as tutor, practice partner, critic, simulator, drafting assistant or administrative aid. Each role creates different opportunities and risks. The teacher should specify what the learner does before using AI, what the tool may contribute, what the learner must verify and what independent performance will follow.
Assessment requires the same precision. Schools should combine evidence from the product, the learning process, student explanation, transfer and AI-related judgment. Some tasks should be completed without AI, some with bounded support and some should assess critical use directly. For a fuller design model, see How Should Schools Assess Learning When Students Use AI? and the related guide to building student independence in the classroom.
Pedagogical competence is not measured by the number of AI activities used. It is measured by the quality of the learning design and the evidence that students are becoming more capable.
Domain 5: turning individual experimentation into professional learning and leadership
Professional leadership begins with inquiry rather than advocacy. A teacher can document a small use case, compare it with previous practice, collect evidence of student learning and report both benefits and limitations. This is more valuable than declaring a tool transformative after one successful lesson.
At department or school level, leaders can establish common language for permitted use, verification, disclosure, accessibility and data protection. They can create review cycles rather than permanent policies written around a specific product. They can also protect time for teachers to examine student work, test claims and share failed experiments as well as successes.
Leadership does not require positional authority. A teacher who helps colleagues interpret a framework, models a safer routine or brings student evidence into a policy discussion is already exercising professional leadership. The aim is distributed judgment: schools should not depend on one enthusiast, one vendor or one technical officer to decide what counts as educationally sound.
Four observable stages of development
The following matrix translates the five domains into a practical progression. The stages are not personality labels and should not be averaged into a single score. A teacher may be independent in pedagogy and still need guided support on procurement or data governance.
| Stage | What practice looks like | Evidence worth collecting | Best next move |
|---|---|---|---|
| Aware | The teacher recognizes opportunities and risks but relies on general rules or informal impressions. | A short explanation of one appropriate and one inappropriate use, plus recognition of a verification or privacy risk. | Learn one core concept and analyze one authentic classroom case. |
| Guided | The teacher uses an approved routine or tool with support and can explain the educational purpose. | An annotated lesson, verification record, risk check and student evidence reviewed with a colleague. | Repeat the routine in a second context and compare assisted with independent learning. |
| Independent | The teacher selects, adapts or rejects AI based on goals, evidence, policy and learner needs. | A sequence of lessons showing differentiated use, valid assessment and documented improvement. | Share the decision process, mentor a colleague and identify a school-level barrier. |
| Leading | The teacher or leader supports collective capability, evaluates systems and improves policy through evidence. | Professional-learning materials, evaluation findings, governance contributions and evidence of change across teams. | Establish review cycles, student participation and external comparison without turning the framework into compliance theater. |
Progression is not simply “use more AI.” A teacher may develop by refusing an unsuitable tool, reducing data collection, redesigning a task to restore student reasoning or challenging an unsupported institutional claim. Competence is the quality of judgment demonstrated in context.
A teacher evidence audit
Self-assessment becomes more credible when it is anchored in artifacts. Instead of asking teachers to rate confidence from one to five, a professional-learning conversation can examine a recent lesson, the generated material, the teacher’s verification process, samples of student work and the independent evidence collected afterward.
For each domain, the teacher identifies one claim about practice and one artifact that could support or challenge it. “I protect learner agency” becomes testable when the lesson shows which thinking remained with students and when students later explain or transfer the idea. “I verify AI output” becomes testable when the teacher can show what was checked, against which source and what changed as a result. “I use AI ethically” becomes testable when the selected tool, data practices, accessibility choices and alternatives are documented.
This evidence-based approach reduces two common distortions: confidence without competence and competence without confidence. Some teachers are comfortable experimenting but overlook important risks. Others exercise excellent professional judgment while underestimating their capability because they do not identify as “technical.” A map should make both patterns visible.
A 90-day professional-development pathway
Sustainable development needs repeated application, feedback and evidence. A single workshop can introduce language; it rarely changes practice by itself.
| Period | Professional focus | Practice task | Evidence of development |
|---|---|---|---|
| Days 1–30: establish foundations | Human agency, model limitations, privacy and verification. | Analyze one approved and one unsuitable use case; verify a generated resource against primary sources. | A documented decision showing purpose, risks, verification and what remained human. |
| Days 31–60: redesign teaching | Pedagogical role, learner effort, accessibility and assessment validity. | Teach one bounded AI-supported sequence followed by an independent explanation or transfer task. | Student evidence compared across assisted and independent conditions, with an instructional revision. |
| Days 61–90: build collective capability | Professional inquiry, shared routines and policy feedback. | Present the case to colleagues, include limitations and test the routine in another subject or class. | A revised shared protocol, peer feedback and a clear next evaluation question. |
The pathway should be adapted to jurisdiction, age group, subject and school policy. It should also include student voice. Learners often know where AI is useful, confusing or easy to misuse, and their accounts can expose gaps between written rules and actual practice.
What school leaders should do with the map
Leaders should use the map to organize support, not to create a punitive checklist. A school can begin by identifying which domains are already represented in policy and which remain invisible. Many policies address academic integrity while saying little about teacher workload, assessment validity, accessibility or professional agency. Others encourage experimentation without defining how evidence will be collected.
A balanced implementation provides approved tools and minimum safety expectations, but also creates space for subject-specific judgment. It distinguishes classroom experimentation from high-stakes assessment. It protects teachers from being required to use systems they cannot evaluate, and it avoids promising that AI will reduce workload before local evidence exists.
International frameworks are useful precisely because they reveal recurring principles across systems, not because they erase local differences. Comparative cases such as AI Schools Around the World: Eight Unusual Experiments in Learning can widen the discussion, provided that unusual models are treated as evidence to examine rather than templates to copy. The Australian Framework for Generative Artificial Intelligence in Schools, European guidance and national materials should be read alongside local law, curriculum, qualification rules and community expectations.
The Global Teacher AI Competency Map offers a common structure for that work. It helps educators move from “Can I use this tool?” to more professional questions: “Should it be used here? What evidence supports the decision? What remains human? What do students learn? What must the school govern?”
Educators who want to develop observation, diagnostic questioning, reflective practice and progressive learner autonomy can explore the Become a Learning Coach training pathway, the book Become a Learning Coach, and the wider Learning Coach English article library. The goal is not to turn teachers into technicians. It is to strengthen their capacity to understand, verify, teach and lead when AI is present.
