AI for Educators: A 4-Module Workshop Playbook for Universities and K-12 Schools

Most “AI for teachers” workshops are tool tutorials. Educators leave knowing how to click buttons and with no changed practice. The useful workshops change how teachers plan, assess, and respond to students. Here is the four-module structure we run in AI Academy educator programs.

Quick answer

A workshop series that actually changes teaching practice covers four modules:

1. Foundations and mental model. What LLMs are, what they are not, where to trust them.
2. Curriculum and preparation. Lesson planning, differentiation, rubric generation, content localization.
3. Assessment and feedback. Item generation, formative assessment, feedback drafting, academic-integrity-aware design.
4. Personalization and intervention. Per-student tutoring, early warning, differentiated support.

The wrong way

The shape of a workshop that does not change practice:

• Two hours of “here is ChatGPT, try typing a prompt.”
• A parade of vendor tool demos.
• A panel on academic integrity with no concrete policy.
• No follow-up.

A week later, teachers use these tools exactly the same way they would have anyway — for minor task automation. Nothing institutional changes. This is why most “AI in education” spend produces no measurable outcome.

The four-module playbook

Module 1 — Foundations and mental model

Duration: Half-day.

Outcomes: Teachers can explain in one sentence what an LLM is, what it is not, and where their judgment is still required.

Topics:

• How LLMs work, at the level of “they predict next tokens based on statistical patterns” — no matrix algebra.
• Where they are reliable (language tasks, structured transformation) and where they are not (specific facts, recent events, mathematical precision).
• The “verify, never trust” heuristic for educator output.
• Data privacy and what not to paste into a public AI tool — including student work and grades.

Module 2 — Curriculum and preparation

Duration: Full day, hands-on.

Outcomes: Teachers produce a real lesson plan, with differentiation, for a real upcoming class.

Topics:

• Lesson plan scaffolding and rubric generation.
• Producing leveled versions of the same content for differentiated instruction.
• Arabic content localization — when machine output works and when native review is required.
• Standards-alignment prompts (local ministry frameworks, international accreditation).
• The “prompt library” approach — save good prompts, share with colleagues.

Module 3 — Assessment and feedback

Duration: Full day, hands-on.

Outcomes: Teachers produce a full assessment kit for a unit, plus a redesigned assessment approach that is AI-resilient.

Topics:

• Item generation: multiple choice, short answer, extended response.
• Rubric-aligned feedback drafting — AI generates the first draft, teacher approves.
• Formative-assessment design for quick-cycle feedback loops.
• Assessment redesign for academic integrity: in-person, oral, and process-based.
• Writing an assignment brief that makes AI misuse harder (course-specific artifacts, require specific citations, require in-class defense).

Module 4 — Personalization and intervention

Duration: Half-day + follow-up learning community.

Outcomes: Teachers build one personalization workflow for their class.

Topics:

• Per-student tutoring prompts and scaffolds.
• Early-warning systems: using LLMs to help process formative data and flag struggling students.
• Differentiated support plans generated from student profiles + teacher judgment.
• Parent communication drafting.

On academic integrity

Two truths:

1. AI detection tools do not work reliably. Their false-positive rates are high enough to unfairly accuse non-AI-using students, and their false-negative rates are high enough to miss actual use. They get worse as models improve.
2. Assessment redesign does work. In-person exams, oral defenses, process portfolios, and prompts that require course-specific artifacts all raise the cost of AI misuse to above the cost of doing the work.

Policies that rely on detection will be broken within a year. Policies that redesign assessment for an AI-ubiquitous world are durable. This is the single most important conceptual shift in the Module 3 workshop — and it is the one that generates the most institutional pushback before teachers try it.

Next step

The AI Academy offers this four-module program as an in-service training for universities and K-12 schools. Typical engagements are semester-long, combine instructor-led workshops with a learning community, and produce measurable changes in curriculum artifacts and assessment design.