Practical AI training for everyone
Module 1
7 min read
Ask an AI for a lesson plan and you will get a competent-looking document you would never teach from. It will have objectives that do not match your curriculum, a timing structure that ignores your period length, and activities that assume resources you do not have. This is not a failure of the tool; it is the generic-in, generic-out problem from the free course, applied to the most context-dependent document in your professional life.
A lesson plan is unusually dependent on constraints, and a model has none of them unless you supply them. Before it can produce anything usable it needs: the curriculum point, the year group and their prior knowledge, the period length, the class size, the resources available, the specific misconceptions you expect, and what the students did last lesson.
That sounds like a lot to type. It is about four lines, you write it once per class, and you reuse it all term. This is the reusable-context idea from module 4 of the free course, and teaching is where it pays back hardest.
Your reusable class block
'Grade 8 science, 32 students, 45-minute periods, mixed ability with about six students working two years below level. Data projector and one class set of tablets. Curriculum: national science curriculum. Students are generally confident with practical work and weak on written explanation.' Paste this at the top of every planning conversation for that class.
The highest-value division of labour is not 'AI plans the lesson.' It is 'you decide the arc, AI produces the material.' You know that this lesson needs a hook, fifteen minutes of direct instruction, a paired activity and an exit ticket. That structural judgement is the professional part and it takes you ninety seconds.
What eats your evening is the filling: writing the hook, producing the worked examples, building the exit ticket, differentiating the worksheet three ways. Hand over that and keep the arc.
The prompt that works
'[class block]. Next lesson is on balanced and unbalanced forces. I want: a 5-minute hook that creates a question in their heads, 15 minutes of direct instruction with three worked examples increasing in difficulty, a 15-minute paired activity using the tablets, and a 4-question exit ticket that would tell me who has not got it. Write the actual materials, not a description of them.'
That last sentence matters more than it looks. Without it you get a plan that says 'introduce the concept with an engaging demonstration', which is a description of work, not work. 'Write the actual materials' is what turns a document into something you can walk into a classroom with.
One move worth building into your routine: before planning, ask 'what do students at this level typically get wrong about this topic, and why?' Experienced teachers carry this knowledge for their own subject and lose it the moment they teach outside it. The answers are reliable here because common misconceptions are extremely well documented; this is genuinely a strong use of the tool.
Then plan against them. A lesson designed around the three things students actually get wrong beats a lesson designed around the topic, every time.
Try it now
Write your class block for the class you teach most. Then plan your next lesson with it, using the structure above: you specify the arc, the model writes the materials. Time it against how long the same lesson usually takes you.
A few quick questions. You'll see an explanation after each one.
Question 1 of 4
Why does 'write me a lesson plan on fractions' produce something unusable?