Most school AI training fails in the same way: an inspiring INSET, a flurry of interest, then nothing. Training that sticks is practical, specific to real tasks and followed up. Here is a plan that works in a term.
Before any training
- Pick one approved tool. Usually Copilot or Gemini on school accounts. Training on five tools trains people on none.
- Have the rules ready. A short staff acceptable use agreement covering data and checking outputs.
- Survey staff. What do they already use, and which tasks eat their time?
Who to train first
Not everyone at once. Order by workload and influence:
- Office team and SBM — the heaviest admin load and the fastest wins.
- Two or three teacher champions — one enthusiast, one sceptic.
- SLT — they set the tone and have their own paperwork. See AI for SLT.
- Everyone else, trained by champions with real examples from your own school.
What a half-day session should cover
- The rules in ten minutes — school accounts only, no pupil data unless approved, check everything.
- Five real tasks, hands on — a parent letter, a quiz, adapting a text for a lower reading age, a report comment from notes, summarising a document.
- How to prompt — role, task, context, format, with a shared prompt sheet.
- Where it goes wrong — confident errors, bias, and why checking matters.
- One commitment each — every member of staff leaves with one task they will use AI for this week.
See what to include in an AI INSET day for a fuller agenda.
Want staff training that actually changes how people work?
Tell me what you're trying to sort out — a tool, a policy, a parent question, a workload problem — and I'll give you a straight answer.
Follow-up is where it sticks
- Week 1: champions available for ten-minute drop-ins.
- Week 3: five-minute "what worked" slot in a staff meeting.
- Week 6: shared prompt library updated with staff's own prompts.
- End of term: re-survey — confidence, usage and time saved.
Don't skip safeguarding
Every session should include the safeguarding picture: filtering and monitoring, deepfakes and AI imagery, and companion chatbots. Your DSL should present it. See KCSIE filtering and monitoring for AI.
Measure it
Ask staff to estimate hours a week on three target tasks before and after. Even rough numbers show whether the training paid off, and give governors something concrete. See if every teacher saved two hours.
Buying training in
External training is worth it when it is hands-on, specific to your school and includes follow-up. Avoid generic "AI awareness" sessions. See AI INSET training providers and what whole-school AI training costs.
Want me to run it?
This is exactly the session I deliver: half a day in your school (or two online sessions), hands-on, built around your staff's real tasks, with the safeguarding and data rules baked in, and follow-up included. £395 for the whole staff body. Add the policy, staff agreement and governor briefing in the £495 KCSIE AI package and your school is covered.
Frequently asked questions
How should schools train staff in AI?
Pick one approved tool, set clear rules, train the office team and champions first, run a hands-on half-day on real tasks, and follow up over the term.
How long does AI training for school staff take?
A half-day hands-on session plus light follow-up over a term is enough to change how most staff work.
What should AI staff training cover?
The data and safety rules, five real tasks hands on, how to prompt, where AI goes wrong, and one commitment each.
Who should be trained first?
The office team and school business manager, then teacher champions and SLT, then everyone else using real examples from the school.
How do we know AI training worked?
Compare staff estimates of hours spent on target tasks before and after, plus confidence and usage surveys.