AI training for teachers and school staff: what to teach first

Teachers should learn AI first on lesson planning, differentiated materials, feedback drafts and parent messages in Arabic and English, using their own curriculum and classes. Admin staff learn admissions replies and letters. Two rules apply from the start: the teacher decides every grade and comment, and student data goes only into tools the school has approved.

#Key takeaways

• Teachers get the most from AI on preparation: lesson plans, differentiated worksheets, rubrics and question banks.
• Feedback and grading drafts are useful only when the teacher reviews and owns every comment and mark.
• Parent messages in Arabic and English are a quick win for both teachers and admin staff.
• Student data and academic integrity need written rules before training starts, not after.
• A free course is enough for individual teachers; whole-school training is worth it when the school wants shared rules and measured use.

#What does a teacher's week look like with AI?

Teachers plan lessons, adapt materials for different levels, mark work, write report comments, reply to parents and attend meetings, often through Google Classroom or Microsoft Teams and a school messaging app or WhatsApp. Admin staff answer admissions questions, prepare letters and manage timetables. Many teachers already try ChatGPT or Gemini at home for a lesson idea or a worksheet.

AI helps with the preparation: a lesson outline aligned to the curriculum, three versions of one worksheet for different levels, a rubric, a parent letter in both languages. It goes wrong when a draft contains a factual error a student then learns, when feedback is pasted in without the teacher reading the work, or when names and grades go into a personal account.

#What should teachers learn first, and in which order?

1. lesson planning: outlines, examples and question sets from the school's own curriculum, checked for accuracy.
2. differentiation: the same task at several levels, including support for students learning in an additional language.
3. parent messages: clear letters and replies in Arabic and English, in the school's tone.
4. feedback drafts: comments against the rubric that the teacher edits, signs and owns.
5. later: report comment drafts, department question banks and admin tasks such as admissions replies.

Feedback comes fourth because it is the most sensitive. By then teachers have learned where AI makes mistakes in their own subject, so they read its feedback drafts critically. Starting with grading, because it promises the biggest time saving, invites the shortcut that parents and inspectors worry about most.

#Which real tasks should the training use?

Use next term's real work. Each teacher brings an upcoming unit, a class list with names removed, a recent piece of student work with the rubric and a parent message they found hard to write. Admin staff bring common admissions questions. Everyone leaves with materials they will actually use in class or send to families.

#What should school leaders measure before and after?

• How long teachers say planning, differentiation and parent messages take, from a short survey before and after.
• How many staff use the approved tool, and for which tasks, from its usage logs.
• Quality checks: errors found in ai-drafted materials during department review.
• Parent message response and complaint patterns, where the school already tracks them.

The NIST AI risk management framework, january 2023 treats human oversight and accountability as part of managing AI risk. In a school, that oversight is the teacher's professional judgement, and the measures should show it is still in place.

#What are the safe-use rules for student data and integrity?

• Student names, grades, health notes and family details go only into tools the school has approved.
• The teacher reads every piece of work before any feedback or mark is given, and owns the result.
• Each assignment states whether students may use AI, and how they should say so.
• Materials are checked for accuracy and bias before they reach a class.
• Student data is personal data under laws such as the UAE personal data protection law, and some schools fall under extra education rules.

#When is a workshop or free course enough?

If a few teachers want to try AI for lesson ideas, the free courses that tool vendors and teacher networks publish, plus a staff meeting on the school's data rules, are enough. Many good teachers learn this way, and nobody should feel they need a programme to start.

Whole-school training is worth it when the school or group wants shared rules on student data and integrity, consistent parent communication in Arabic and English, and evidence for governors or inspectors of how AI is used. That needs training on the school's own curriculum and policies, with use measured. For schools with staff in the EU, the EU AI Act, article 4 also requires a sufficient level of AI literacy.

#How Stennir runs this

Stennir runs school training as cohorts in the AI Operator Program, part of the AI training service: teachers by department or phase, admin staff separately, and an executive session for the principal and senior leaders. Trainers have worked in education, and every exercise uses your curriculum and documents, in Arabic and English. Usage is measured before and after, ending in an adoption report.

#What should a school leader do next?

Ask each department which preparation task takes the most effort, and check whether the school has written rules on student data and AI use in assignments. Then read AI for teaching and AI for school groups: admissions and parent messages. When ready, take the free readiness assessment or book a 30-minute call.

#Common questions