Staff Online AI Course
An internal, self-paced course for teaching staff, hosted on a secure the certification client endpoint separate from the public websites. The course is interactive: teachers practise with a live AI assistant inside the course itself, not only read about it.
*This document is the course content and the build specification. Sourced facts and inferences are separated where it matters. A source note is at the end recording what has not yet been published by the authorities.*
---
What this course is for
Teachers finish the course able to use AI in their work **safely, accurately, and within the boundaries set by the authorities** — and, above all, able to **validate what an AI gives them** before it reaches a lesson, a worksheet, or a student. Data validation and integrity are the centre of gravity of this course, because that is the one skill that separates a teacher who is helped by AI from one who is misled by it.
The course keeps strictly within the KHDA and UAE Ministry of Education position on AI in schools. Nothing in it encourages a use of AI that falls outside that position.
---
How teachers move through it
The course is self-paced. A teacher signs in, completes an autonomous self-placement check, then moves through the core path at their own speed. Stronger participants take the optional "Go Further" extensions attached to several modules. Everyone covers the same core.
**Autonomous self-placement check (opens the course, ~3 minutes).**
Not a gate and not a sorting into different courses — a short self-assessment so each teacher knows where they stand and the school gets a baseline picture of the group. The result is shown to the teacher and logged with their completion record.
Six questions, each answered on a simple scale (Never / Tried once or twice / Use it sometimes / Use it confidently):
1. I have used an AI assistant (such as a chatbot) for a work task.
2. I can write a clear instruction to an AI and refine it when the first answer is not quite right.
3. I check AI-generated facts, figures, or references against a reliable source before I use them.
4. I know which AI uses are allowed and not allowed in a UAE school.
5. I can explain to a student why a piece of AI-generated work is or is not acceptable.
6. I have used AI to prepare a lesson, a resource, or feedback for marking.
The teacher sees a plain result — "You are starting out," "You are building confidence," or "You are already comfortable" — with a line on where to focus. The school sees the distribution across staff. No one is excluded from any module by their result.
---
Module 1 — Where AI stands in the UAE, and why this matters for teachers
*Sourced. This module sets the national context so teachers understand they are joining something already national, not an internal experiment.*
The UAE has placed artificial intelligence at the centre of national strategy, and Dubai has made education part of that. Teachers should recognise the landscape they are teaching into.
- The UAE was the **first country in the world to appoint a government minister for Artificial Intelligence** — Omar Sultan Al Olama, Minister of State for AI, Digital Economy and Remote Work Applications — signalling that AI is treated as national infrastructure, not a passing trend.
- Dubai runs the **Dubai Universal Blueprint for Artificial Intelligence**, a citywide plan to accelerate AI adoption, tied to the **Dubai Economic Agenda (D33)**, with the aim of adding AED 100 billion a year to the economy through AI and digital solutions. One pillar is the appointment of Chief AI Officers across government entities.
- Under the directives of **Sheikh Hamdan bin Mohammed bin Rashid Al Maktoum**, Crown Prince of Dubai, the **Dubai Future Foundation** and the **Dubai Centre for AI** run public programmes to build AI skills across the whole population:
- **One Million Prompters** — an initiative to train one million people in AI and prompt-writing skills over three years, delivered as a free, hands-on course with an accredited certificate. Its four modules move from AI foundations, to writing better instructions, to productivity, to creative uses. It is the clearest public signal of the skill the government wants citizens and residents to hold: the ability to instruct AI well.
- **Global Prompt Engineering Championship** — described as the world's largest AI prompt-writing competition, with a prize pool of **AED 1 million**, held at the Museum of the Future, run by the Dubai Centre for AI under the Dubai Future Foundation. It began with categories in Coding, Art, and Literature and has since expanded, with a Golden Visa now among the rewards.
- **Dubai AI Week** — a citywide event (held in April 2026 at the Dubai World Trade Centre and other venues) bringing government, industry, and academia together, including an AI education programme run in collaboration with the KHDA to introduce students and educators to AI.
**Why a teacher should care.** The skill these programmes reward — writing a clear, precise instruction and judging the quality of what comes back — is exactly the skill this course builds. A teacher who can do this well is aligned with where the country is going, and is equipped to guide students toward these same opportunities.
> **Go Further:** Look at the One Million Prompters course at dub.ai. Note its four-module shape; this course borrows that arc and adds the school-specific validation and ethics layers KHDA requires.
*Integrity note carried into delivery: whether school students may personally enrol in One Million Prompters or enter the Championship is **not yet confirmed**. This module presents these as national context and as teacher development avenues, not as student entry routes, until the school verifies eligibility.*
---
Module 2 — What AI actually is, and where it gets things wrong
*Mixed: the description of how these tools behave is general knowledge; the school framing is ours.*
A short, non-technical grounding so teachers hold an accurate mental model.
- Generative AI produces the **most statistically likely** continuation of your words. It is a pattern-completer, not a source of truth. This single fact explains almost everything a teacher needs to watch for.
- Because of this, it can produce fluent, confident text that is **wrong** — invented facts, invented references, plausible-sounding figures with no basis. This is often called "hallucination." It is not a rare glitch; it is a property of the tool.
- It reflects the **biases and gaps** in what it was trained on, and its knowledge has a **cut-off**, so it can be out of date on recent events.
- It is **not private by default** — what you type may be processed on external servers. This matters enormously for student data (Module 5).
**The teacher's takeaway:** treat every AI output as a **draft from a fast, confident, sometimes-wrong assistant** — useful, but never trusted on sight.
> **Go Further:** Ask the in-course assistant a question about a very recent local event and notice how it handles the limit of its knowledge. Observe whether it flags uncertainty or answers confidently regardless.
---
Module 3 — Writing good instructions (practise live)
*Interactive module. The teacher works with the in-course AI assistant on the endpoint.*
The heart of using AI well is giving it a clear instruction. This module is hands-on: the teacher writes instructions to the live assistant and immediately sees the difference good instructions make.
Core technique, taught by doing:
- **Be specific.** Say the subject, grade, length, and purpose. ("A five-question starter quiz on the water cycle for Grade 4, with an answer key" beats "make a water cycle quiz.")
- **Give context and a role.** Tell it who it is writing for and in what voice.
- **Show an example** of what you want when you can.
- **Refine.** The first answer is a starting point. Ask for changes; narrow it; correct it.
- **Ask it to show its reasoning** for anything where correctness matters, so you can check the steps.
**Live practice tasks** (the assistant responds; the teacher iterates):
1. Turn a vague request into a specific one and compare the two outputs.
2. Take a produced worksheet and refine it three times toward something usable.
3. Ask the same question two ways and see how the instruction shapes the answer.
> **Go Further:** Take a real task from your own week and get the assistant to a genuinely usable result through refinement. Save the instruction you used.
---
Module 4 — Validating what AI gives you (the core of this course)
*This is the module the whole course is built around. Mixed sourcing: the validation discipline is ours and general good practice; the "human review" requirement is sourced to the UAE MoE manual.*
Most teachers do not yet know how to check an AI's output. That gap — not a lack of enthusiasm — is the real risk. This module makes validation a habit.
**Why this matters most.** The UAE Ministry of Education's position is explicit: schools and educators **must not rely solely on AI-generated outputs**, and both teachers and students are required to **verify the accuracy, relevance, and alignment** of AI-generated content with the approved curriculum **before** using it in lessons or assignments. Validation is not optional good practice here; it is the stated expectation.
**The validation habit, taught as a routine:**
1. **Assume nothing is true until checked.** Facts, dates, figures, quotes, and especially references and citations are the highest-risk items — AI invents these convincingly.
2. **Check against a reliable source.** For anything factual, confirm against a textbook, an official site, or curriculum material — not against another AI.
3. **Check curriculum alignment.** Correct in general does not mean right for your grade, your board, or your scheme of work. A teacher's judgement is the filter.
4. **Check for bias and tone.** Does it carry an assumption, a cultural slant, or a framing that does not fit a UAE classroom?
5. **Check it makes sense.** If a figure or claim feels off, it usually is. Trust the professional instinct and verify.
**Live practice (interactive):** the teacher is given AI outputs that contain planted errors — a wrong date, an invented reference, a figure that does not add up, a claim that is off-curriculum — and must catch them. The in-course assistant then walks through what a rigorous check would have found. This is the exercise that turns the principle into a reflex.
**Data integrity as the moat.** Frame this to teachers plainly: anyone can generate content; the professional skill that holds its value is the ability to *judge* it. A school whose teachers validate rigorously produces trustworthy material; that trustworthiness is the advantage AI cannot hand out for free.
> **Go Further:** Build yourself a short personal validation checklist you will actually run every time, and test it on a real AI output this week.
---
Module 5 — Ethics, safeguarding, and the rules that bind us
*Sourced to the UAE MoE "Safe and Responsible Use of Artificial Intelligence in Classrooms" manual (2026) and KHDA/national position. Framed as the boundary the school works inside.*
This module sets the non-negotiable boundaries. Teachers leave knowing what is allowed, what is not, and how to protect students.
Sourced rules from the national framework:
- **Human oversight is required.** Generative AI should only be used where **direct teacher oversight and clear instructional guidance** are present.
- **Age boundary.** Generative AI tools are **prohibited for students under 13, or below Year 7 / equivalent grade.** Teachers must know which of their classes this covers.
- **Assessment integrity.** Using generative AI during **formal examinations or official assessments is banned.** Presenting AI-generated work as a student's own — without disclosure or teacher approval — violates academic integrity, as does copying or rephrasing AI output without genuine understanding.
- **Human review of all content.** Schools and educators must not rely solely on AI outputs; all content passes through human review before use.
- **No inappropriate content**, and any use outside approved frameworks is subject to accountability under existing regulations.
Safeguarding the teacher must own:
- **Never put student personal data into a public AI tool.** Names, records, assessment data, anything identifying — these do not go into external chatbots. (This is also why the certification client is building its own internal, secure endpoint.)
- **Model honest use.** How a teacher uses AI teaches students how to use it. Disclosed, supervised, checked — that is the model.
- **Bring students in, don't shut them out.** The national guidance frames AI as something to explore *with* students under supervision, building judgement, not something to hide.
> **Go Further:** Locate the exact age/grade line for the classes you teach and write one sentence you could say to a class about honest AI use.
---
Module 6 — Good practice: AI in the working week
*Practical. General good practice, framed for the certification client. Several tasks are interactive.*
Where AI genuinely saves a teacher time — always inside the rules and always validated.
- **Lesson preparation** — starters, explanations, differentiated versions of a task, worked examples. *Validate for curriculum fit.*
- **Resource creation** — quizzes, worksheets, reading comprehensions, writing prompts. *Validate every fact and answer key.*
- **Marking support** — drafting feedback comments, suggesting rubrics, spotting common errors in a set of answers. *The teacher's judgement remains the authority; AI drafts, the teacher decides.* Never let an AI assign a final grade unchecked.
- **Communication** — drafting a parent message or a notice, then refining tone. *Validate for accuracy and appropriateness.*
- **Differentiation and inclusion** — adapting a text to a reading level, generating supports for varied needs.
**Live practice:** each teacher produces one genuinely usable resource for their own subject with the in-course assistant, runs it through the Module 4 validation habit, and keeps it.
> **Go Further:** Produce a small bank of three validated resources you will actually use this term.
---
Module 7 — Where the certification client is going, and where you fit
*School-specific. Sourced to KHDA framing; the certification client's own direction stated as fact by the school.*
Teachers should leave understanding that this course is one step in a larger, deliberate the certification client direction — and that they are part of it.
- **the certification client is refining its AI literacy programme in line with KHDA.** The school is building a structured approach to AI literacy that follows the authority's direction rather than running ahead of it. This course is the staff foundation for that programme.
- **The programme runs across the school's six core subjects**, so that AI literacy is not a bolt-on but woven through the curriculum. The six are:
- **MSCS**
- **Arabic A and Arabic B**
- **Islamic Studies**
- and the **Ministry of Education subjects — English, Mathematics, and Science** — for all students.
- **This aligns with the national direction of travel:** the UAE has moved AI into the school curriculum, and KHDA's own initiatives (including the AI education work seen at Dubai AI Week and the emerging Skills for Life framework, which names digital skills among its domains) point the same way. The certification client is positioning early and deliberately.
The closing message to the teacher: *you are not being asked to become a technologist. You are being asked to bring good judgement to a powerful tool, inside clear rules — and that is exactly what your profession already trains you to do.*
> **Go Further:** Note one place in your own subject where AI, used well and validated, could improve what students experience — and bring it to the next department discussion.
---
Completion
On finishing, a teacher has:
- a working, safe, validated way of using AI in their role,
- a personal validation checklist they actually use,
- a clear grasp of the rules that bind AI use in a UAE school,
- and a place in the certification client's KHDA-aligned AI literacy direction.
Completion and the self-placement baseline are logged for each teacher, giving the school an attendance and readiness record.
---
Build notes for Claude Code (the hosting layer)
*This is guidance for the terminal build agent. The course above is the content; this is how it should be delivered on the endpoint.*
- **Host on the internal secure endpoint, separate from the public sites.** Not indexed, access-gated to staff. Reuse the existing platform design system — read the existing site before building, do not invent a new visual language.
- **Interactive modules (3, 4, 6) call the Anthropic API** from the endpoint to power the live practice assistant. Keep the system prompt scoped to teacher-support tasks. Note that live practice consumes API credit per teacher — build the static shell first, wire the interactive sandbox into Modules 3, 4, and 6 where it earns its place.
- **Self-placement check is autonomous** — runs at first sign-in, stores result with the teacher's record, shows the plain-language result, never branches content.
- **Log completion and the baseline result per teacher** so the school has an attendance/readiness record (this is the evidence trail the priority plan calls for — "attendance logged each session").
- **No student data anywhere in the build.** The course teaches this rule; the build must embody it.
- **Content sourcing rule holds:** the sourced facts in Modules 1 and 5 must not be altered or embellished in the build. If a fact needs updating, verify against the authority, do not regenerate.
---
Source note — what is sourced and what is not
**Sourced and verified (July 2026):**
- UAE first country to appoint an AI minister (Omar Sultan Al Olama); Dubai Universal Blueprint for AI; D33 / AED 100bn aim; Chief AI Officers pillar.
- One Million Prompters (DCAI under DFF; ~1m people / 3 years; four modules; accredited certificate; free); Global Prompt Engineering Championship (AED 1m prize pool; Museum of the Future; Coding/Art/Literature original categories, since expanded; Golden Visa added); Dubai AI Week (April 2026, DWTC; KHDA education programme).
- UAE MoE "Safe and Responsible Use of Artificial Intelligence in Classrooms" manual (2026): human oversight required; under-13 / below-Year-7 prohibition; ban on generative AI in formal exams/assessments; academic-integrity provisions; requirement to verify accuracy/relevance/curriculum alignment and to conduct human review.
**Not verified / not asserted:**
- **Student eligibility** for One Million Prompters and the Global Prompt Engineering Championship. Presented as national context and teacher avenues only. Confirm before positioning to students.
- Exact 2026 Championship dates (sources listed some as TBC) — kept general.
- KHDA **Skills for Life** implementation detail — the framework is announced but no lesson allocation, assessment method, or documentation standard is published. Referenced only at the level the authority has stated (digital skills is a named domain), nothing beyond.
**Stated by the school as its own direction (not an external claim):**
- That the certification client is refining its AI literacy programme in line with KHDA, delivered across the six core subjects (MSCS; Arabic A and B; Islamic Studies; and the MoE subjects English, Mathematics, Science). This is the certification client's own programme direction, recorded as given by the school.