For PGCE trainees
You are learning to teach. AI can speed up the middle of the work — examples, visuals, practice activities — but it cannot do the thinking that makes you a teacher. You start. AI accelerates. You finish.
Your subject knowledge, the class in front of you, and what “good” looks like.
Options, visuals, games, hinge questions — labelled as drafts, not ready to teach.
Check it, adapt it, put your name to it. You remain responsible.
Ground rules
The same shape as the rest of the site: humans at both ends. These are the practical version for placement and assignments.
Your subject
The most useful thing you bring to a chatbot is not a clever prompt. It is the knowledge you already have — what this discipline cares about, how it is examined, and where pupils actually get stuck.
Your degree is the Direction. Before you open a tool, write down the three ideas students must grasp, the language the specification uses, and the mistakes you know they make. That is subject knowledge doing the work AI cannot.
Spend ten minutes with the spec, a mark scheme or an examiner report. Then ask AI to draft around those misconceptions — not around a generic topic title. The output will be closer first time, and you will be able to tell when it is off.
History is not “content plus a worksheet”. Science is not “a practical write-up”. Ask AI to accelerate the middle (examples, retrieval, a visualisation) while you hold the disciplinary move: argument, method, proof, interpretation, making.
Get three ways of explaining a hard idea, then say them out loud in your own words. If you cannot explain it without looking at the screen, you are not ready to teach it — however polished the draft looks.
What to use it for
AI is often used to produce more paper. Use it instead to make ideas visible, to turn practice into a short game, and to break misconceptions with something pupils can handle — then judge whether it is accurate enough to teach.
Abstract ideas land when pupils can picture them: a labelled diagram of a process, a storyboard of a historical decision, a “zoom in” from the whole system to one part. Ask AI for a description or a simple HTML sketch, then check every label against your subject knowledge before it goes on the board.
A short game (eight to twelve minutes) can do retrieval, vocabulary or method practice without another printed sheet. You set the learning; AI drafts the mechanic, rules and a paper or browser version. You still decide whether it is accurate and whether this class can play it.
Explanations bounce off a stuck idea. A hands-on or interactive task can make the misconception fail in front of them: a simulation that behaves wrongly until they change one variable, a set of examples they have to repair, a mini-lab on one shared screen.
Not every use has to be a game. AI is also good at the unglamorous middle: alternative examples, reading-age rewrites, hinge questions, retrieval quizzes from your own notes, model answers at different grades for you to judge.
Three of the educator prompts are built for this: visualise an idea, a little classroom game, and a misconception activity.
Visualise → · Little game → · Misconception activity → · PGCE assignment help →
For teacher education
A practical session for trainees on using AI without handing over the thinking — subject specialism, classroom tasks, and the prompts that keep judgment with the teacher.
Enter your email to download The Middle Principle PDF.