Gerald Ajam

Explore / Toolkit / Moves

Moves for thinking about educational change

A move is the smallest useful unit of the toolkit: one question, asked on purpose, that changes what the situation looks like. Each comes from one of the three strands, and all of them work on the same kind of problem.

Status Working proposition

Published September 2026

The toolkit

There are ten here. My instinct is that a toolkit people can hold in their heads stops at about seven, so some of these will merge or go. Which ones is part of the testing.

Each move is tried on the same situation: a school that wants to increase the use of generative AI.

From ways of thinking

  1. Surface the assumption

    What must currently be believed for the present arrangement to make sense?

    On the AI situation The plan assumes more use means more benefit. Say that out loud and it stops being obvious.

  2. Change the scale

    Does the problem look different at the learner, classroom, school, system, or societal level?

    On the AI situation For a learner it is a question about effort. For a system it is a question about what an essay still certifies.

  3. Reverse the obvious

    What would the opposite assumption reveal?

    On the AI situation Ask where the school should deliberately use less AI. The answer names what it values most.

From complexity

  1. Change the boundary

    What happens when we include actors, effects, or timescales currently treated as outside the problem?

    On the AI situation Draw the boundary around the school and it is a tools question. Draw it to include exam boards and parents and it becomes an assessment question.

  2. Find the feedback

    What consequence of an action later changes the conditions that produced the action?

    On the AI situation If students use AI on homework, teachers set less homework that AI can do, and the pattern of use shifts again.

  3. Look for delay

    Which effects occur too late to influence the decisions that caused them?

    On the AI situation Productivity shows up this term. Any loss of writing fluency shows up in two or three years, after the decision is settled.

  4. Look for second-order effects

    What happens after the immediate intended result occurs?

    On the AI situation Teachers save time on feedback. Then what is that time spent on, and who decides?

From futuring

  1. Work backwards

    If a preferable future existed, what conditions would need to have become true before it?

    On the AI situation Picture students who use AI with good judgement. Before that, someone taught judgement, and assessment rewarded it.

  2. Introduce discontinuity

    What happens if a condition currently treated as stable changes abruptly?

    On the AI situation Suppose the free tier ends, or a regulator bans use with minors. What of the plan still stands?

  3. Ask what persists

    If the intervention disappeared tomorrow, what capability, practice, or knowledge would remain?

    On the AI situation The tools will change within a year. What should teachers and students still be able to do when they have?

One problem, four ways of seeing it
Singapore