Intro to Git — Snapshots, Branches & Sharing
WorkshopyOfficial10 stepsQuiz9 interactiveDiagram
gitversion controlbeginners
An AI workshop for teachers: plan a lesson, adapt a text, write diagnostic questions and rubrics, then catch the errors before class.

WorkshopyOfficial
38 lessons · used by 2 teachers
Sign in to copy it to your workspace. It’s free.
9 steps · 2 quizzes · open any step, or read it here
This AI workshop for teachers shows you how to plan, adapt and check lessons with an AI assistant. You will write prompts that draft a lesson plan, rewrite one text for different readers and produce quiz questions, then practise checking each draft before it reaches your class.
The idea behind the whole session: the AI drafts, you decide. A draft can save you twenty minutes, and it can also carry a wrong fact or a wrong answer key straight into a lesson if nobody reads it. Press Play and watch one request travel through the three stages you will practise.
Interactive · From a request to a lesson you can trust — try it in the lesson
In this lesson you'll:
It takes about 70 minutes. You can do every step without an AI tool; if you have one open, try each prompt on a topic you teach next week.
Try it: press Play, then Reset and Play again, and name the one thing that was caught in the Check stage.
Takeaway: an AI gives you a first draft, and nothing goes to students until you have checked it.
Ask an AI for "a lesson plan on fractions" and you get a plan for nobody: no age, no goal, no timings, nothing about what your students already know. It isn't the tool's fault. It can only use what you tell it, so it fills every gap with a guess.
Five details do most of the work in a planning prompt:
Try it: start from the bare prompt, switch the details on one at a time, and watch which one changes the plan most.
Interactive · Prompt builder: what the AI knows about your class — try it in the lesson
The prior-knowledge detail turned the starter from "What is a fraction?" into a question that tests the exact mistake the class makes. The objective changed every activity. If you want the general parts of a good prompt (context, examples, constraints, asking again), the library lesson Prompt Writing Workshop: Get Useful Answers from AI covers them.
If you have an AI tool open, write your own five-detail prompt for a lesson you teach soon, and compare its plan with one from the bare topic.
Takeaway: the more the AI knows about your class, the less it guesses; what students already know matters most.
One of the most useful things an AI does for teachers is rewrite a text at a different reading level: the same content, shorter sentences, fewer long words, hard terms explained. It takes seconds, where doing it by hand for three groups takes an evening.
The risk is that the simpler version quietly changes the content. A model asked to "make it easier" may drop an idea it thinks is too hard, or swap a precise statement for a familiar but wrong one. In science, the classic slip is "plants get their food from the soil", which is the very misconception a photosynthesis lesson is meant to correct.
So check two things in every adapted text:
A good habit is to list the key ideas before you ask, and put them in the prompt: "Keep these ideas: light, carbon dioxide, water, sugar, oxygen."
Try it: move the slider from Level 3 down to Level 1 and watch the numbers. Then switch on the sample AI slip.
Interactive · One text at three reading levels — try it in the lesson
At every level the five key ideas stay. With the slip on, the text reads just as smoothly, drops carbon dioxide and teaches a misconception. Reading for flow would not have caught it; checking the key ideas did.
Takeaway: an adapted text changes its sentences, never its ideas; check the key ideas every time.
Ask an AI for ten multiple-choice questions and you'll get them fast. The right answers are often fine. The wrong answers (distractors) are where drafts are weakest: many are options no student would pick, so they test nothing.
A useful wrong answer matches a mistake students really make. Then a student who picks it has told you what they think, and you know what to reteach. A question with three such distractors works like a short conversation with every student at once.
Two ways to get better distractors from an AI:
Try it: the AI suggested six wrong answers for a physics question. Pick the three you would use and read what each one tells you.
Interactive · Wrong answers that diagnose — try it in the lesson
Three of the six diagnose a specific idea. Two are filler: a student picking them is most likely guessing, and you learn nothing. "Air resistance pushing up" sits in between, and it is also a trap for you: at the top of its flight the ball isn't moving, so there is no air resistance at all.
Takeaway: keep a wrong answer only if you can say what a student who picks it believes.
AI tools draft rubrics in seconds, and the drafts look finished: a neat grid, four levels, a word in every cell. Read the cells. Many say "Excellent conclusion", "Good conclusion", "Fair conclusion". A student told "Good conclusion" knows how you rated it, not what to do next.
Two checks make a drafted rubric useful:
When you ask, say so: "Describe what the work looks like at each level. Don't use quality words like good or excellent." Feedback written from such a rubric then says what to do next, not just how well it went.
Try it: switch "Describe what each level looks like" off and on, and read the feedback the student would get each time. Then add Neatness.
Interactive · A rubric students can act on — try it in the lesson
With descriptions, every line of feedback names a next step. Without them, the feedback is a row of grades in words.
Takeaway: a rubric level should describe the work, so the feedback can tell a student what to do next.
Four questions on what you've practised so far.
Question 1. You ask an AI for a lesson plan. Which addition changes the plan most for your class?
Question 2. An AI rewrote a science text for younger readers. What should you check first?
Question 3. What makes a wrong answer useful in a multiple-choice question?
Question 4. Which rubric descriptor can a student act on?
An AI writes fluent, confident text whether or not it is right. That's what makes its errors easy to miss: a wrong date, a wrong answer key or a made-up reference reads exactly like the true sentences around it.
Read every draft for five kinds of problem before you use it:
Try it: this handout was drafted for a lesson on the water cycle. It reads well. Tap each line you think is wrong.
Interactive · Spot the errors in an AI draft — try it in the lesson
Five problems in twelve lines, and none of them is obvious from the tone. If you had read it only for flow, a class would have heard that evaporation needs 100 °C and been marked wrong for answering 0 °C.
Takeaway: an AI draft sounds right whether or not it is; check facts, keys, objective, safety and sources every time.
Many AI tasks seem to need student details: adapt a text "for Maya", draft feedback on "Sam's essay". In almost every case the AI doesn't need to know who the student is. It needs to know what the student needs.
A habit that keeps you safe whatever tool you use:
Before you use any AI tool for school work, find out what your school or district allows, and whether the tool keeps what you type or uses it for training. School-provided tools are often set up differently from personal accounts. This lesson teaches habits; your school's policy comes first.
Try it: sort the seven items. Where an item can be made safe, open the safer version.
Interactive · Green, amber or red: student information — try it in the lesson
Most red items have a green version that gets the same help: describe the reading level instead of naming the reader, ask for an email template instead of pasting the email thread.
Takeaway: the AI needs to know what a student needs, not who the student is.
Six short exercises from the work you have practised. Pick an answer and check it; a hint costs nothing.
The first time you use AI for planning, it can take longer than doing it yourself: writing the prompt, reading the draft, fixing it. It gets faster when you stop starting from a blank box.
Keep a short document with:
The saved prompt makes the draft better; the checklist makes it safe to use. Share both with your department and you each save the time, and agree on the same checks.
Try it: put in your own subject, age and lesson length, pick the tasks you'd use AI for, and tick through the checklist that appears. Copy the result into a document you keep.
Interactive · Your AI routine: saved prompts and a checklist — try it in the lesson
Notice that the checklist grows with the tasks: questions bring the answer key, planning brings timings, and privacy and sources are always on the list.
Takeaway: save the prompts that work and run the same checks every time; that's what makes AI save you time.
Six questions on the whole session.
Question 1. An AI-drafted handout cites a book you don't recognise. What is the best next step?
Question 2. Which request gets help with a student's reading without sharing who they are?
Question 3. Before you use an AI-made quiz with a class, what should you check? (pick all that apply)
Question 4. Students who answer "At the top of its flight, no forces act on the ball" most likely believe:
Question 5. Which sentence in an AI-simplified science text must you fix before class?
Question 6. Why keep a saved prompt for a task you repeat, such as writing quiz questions?
Workshopy shares this lesson under CC BY 4.0. Anyone may use and change it, with credit to you. Anyone can learn it here, free. A copy in your workspace is yours to change and run live; your students never need an account.
Version 1Current
· First published
Copies are made from the current version. A copy you already have doesn’t change.
WorkshopyOfficial10 stepsQuiz9 interactiveDiagram
gitversion controlbeginners
WorkshopyOfficial16 steps2 quizzes15 interactiveDiagram
workshopyproduct touronboarding
WorkshopyOfficial13 steps2 quizzes12 interactive
dnageneticsbiology
WorkshopyOfficial10 steps2 quizzes10 interactive
teacher traininglearning scienceretrieval practice
Comments
No comments yet. Be the first to say how it went.