Switch course ▾
Session 5. Training Colleagues and Keeping Judgement Central
Objective. Plan and teach a 45-minute colleague workshop with an offline version, define which decisions AI never makes, respond to resistance, and assemble a one-page school AI plan with review dates.
Training colleagues is where a good policy either becomes daily practice or stays in a drawer. Adults learn a new practice the way pupils do: by trying it on something that matters, getting feedback and returning to it. A leader also carries a quiet duty: to make sure that, as AI gets easier, people do not stop thinking. This lesson goes beyond the key ideas: how adults actually change practice, and how to keep judgement strong with simple checks and a regular review.
Key ideas
Train colleagues the way you teach
- Start from a real task colleagues care about, such as report comments, a quiz or a parent message, not from a tour of tools.
- Show it live in ten minutes, including the checking step, then let colleagues try it on their own material. Doing beats listening.
- Plan a follow-up two weeks later. Short, repeated sessions that return to classroom practice usually work better than one long talk.
A 45-minute workshop
- 5 min: why this task and what the policy says. 10 min: you demonstrate, with one deliberate error for colleagues to catch.
- 20 min: pairs practise on their own lesson materials. 5 min: share one risk each noticed. 5 min: each person commits to one try before a set date.
- Offline version: printed example prompts and a textbook page, so the workshop runs even with no internet or no devices.
Teacher judgement stays central
- Write down decisions AI never makes: marks and grades, discipline, special-needs placement, wellbeing concerns, reports to parents. AI may draft; a named teacher decides and signs.
- Watch for over-reliance: if a draft is accepted without change every time, the check has become a ritual. Keep an edit log and ask 'what did I change, and why?'
- Protect core skills: some tasks, such as writing a comment from scratch, stay hand-made so that professional judgement does not fade.
Resistance, and keeping the plan alive
- Scepticism is useful and enthusiasm is useful. Ask the sceptic to test the plan and the enthusiast to share a failure.
- Listen first, then offer a small, low-risk first step. Do not force: pressure makes people hide what they do.
- Assemble a school AI plan on one page: policy, pilots, vendor rule, equity steps, training dates and a review each term.
A training and review cycle
How adults change practice
Guidance on effective professional development describes several things that work together: building knowledge so that people understand why, motivating them, giving them concrete techniques to try, and helping new habits to stick through follow-up. A single talk covers only the first. So aim for a small shape: one real task, a short demonstration, practice on the participants' own material, a commitment to try once, and a meeting two weeks later.
Respect time. In many Uzbek schools a department has 45 minutes after lessons, not a day. Do less, and do it well. Use pairs, because teachers learn from each other and the quiet teacher is not exposed. Ask each colleague to bring a real lesson or worksheet and leave with something usable. End with a date, because a promise without a date is a wish.
Guarding judgement and skills
Judgement weakens when it is not used. If a colleague accepts every AI draft unchanged, checking has become a ritual. Simple safeguards help. Keep an edit log with one line: what I changed and why. Once a term, pick five AI-assisted outputs, such as comments or quizzes, and compare them with what the teacher would have written alone. Keep some tasks fully hand-made so that core skills do not fade.
Be explicit about decisions that stay human: marks and grades, discipline, special-needs placement, wellbeing concerns and reports to parents. AI may draft, a named teacher decides and signs. Treat enthusiasm and scepticism both as information. The sceptic often finds the real risk, and the enthusiast finds the real use. Review the school plan each term with evidence from classrooms, and change it when the evidence says so.
Examples from the classroom
The planted error
Scene. Malika runs a 45-minute workshop for six colleagues in a Samarkand school. Her demonstration is a Navruz reading text with a wrong date she planted. Only two of six colleagues notice it. She does not scold anyone; she says that this is why the checking step exists. In pairs, colleagues check each other's texts, and every pair finds at least one weak sentence.
Insight. A planted error teaches the need for checking without blame.
The edit log
Scene. Jasur has used AI to draft report comments for two terms. After the school adopts the edit log, he writes 'changed three comments: wrong reading level; too general; unkind tone'. At the end of the term he reads the log and sees that he edits less each month. He decides to write four comments from scratch to keep the skill sharp, and shares his log in the department meeting.
Insight. A one-line log makes over-reliance visible and keeps the teacher in control.
The sceptic as tester
Scene. In an Andijan school, Aziz is firmly against AI. The lead teacher, Dilnoza, asks him to test the department policy by trying to break it with five situations. He finds two gaps, including one about pupils' photographs, and the policy improves. At the next meeting Aziz presents the fixes. He is still cautious, but now he owns the policy.
Insight. Giving sceptics a testing role turns resistance into improvement.
Common pitfalls
- Running one long session with no practice and no follow-up.
- Trying to cover every tool instead of one real task.
- Treating a sceptical colleague as a problem to overcome.
- Letting the school plan go unreviewed so that it stays out of date.
Plan, Teach, Assemble
- Open the Activity Kit: the workshop planner, a SAMPLE department of six teachers with different views, and the school plan on a page (3 min).
- Pairs (12 min): choose one real task and fill in the 45-minute planner, including the deliberate error and the offline version.
- Trios (15 min): each person teaches a 4-minute slice of their workshop; the others give one thing to keep and one to change. No technology needed.
- Pairs (10 min): list the decisions AI never makes in your school and complete the school AI plan on a page with dates, using the work from Sessions 1 to 4.
- Second short activity (10 min): role-play two difficult colleague conversations with the scripts in the Kit ('I don't trust it' and 'I use it for everything'). Rotate once.
- Alone (5 min): write your school-based task from the Kit.
Your facilitator will guide you through this activity in the session.
Questions to discuss
- What did you cut from your workshop to keep it to 45 minutes, and why?
- Which decision on your list would be hardest to hold if the head asked AI to take it over?
- Which phrase in the role-play helped the colleague feel respected?
Reflect
- Who in your school would be the best partner to co-teach a first workshop?
- Which of your own teaching decisions would you never hand to AI?
- What will you stop doing to make room for this work?
Try it tomorrow
- Write the agenda of a 45-minute workshop on one task, with minutes for each part.
- Choose three colleagues and set the date of the follow-up meeting before the first session.
- Start your own edit log for one week and read it at the end.
Further reading and links
Teach colleagues through a real task and a short follow-up, name the decisions that stay human, and keep the whole plan on one page with review dates.