Switch course ▾
Session 2. Appropriate vs. Risky Uses
Objective. Classify proposed classroom uses of AI as green, yellow or red, and justify each choice.
Colleagues are already using AI for lesson ideas, and some students are using it for homework. Rules in schools are still forming, so you will often decide alone. The green, yellow and red zones are not a list to memorise; they are a way of thinking you can apply to a use nobody has listed. In this session you go beyond sorting: you see why detectors fail, how to handle student information, and how the same tool can sit in different zones depending on the task.
Key ideas
Green: low-risk helpers
- Brainstorming, outlines, rewording for reading level, first drafts of rubrics or letters.
- Nothing reaches students or parents until you have read and edited it.
- No personal data goes in.
Yellow: only with checking
- Anything where facts, numbers, translations or answer keys matter.
- Student use of AI, if the task, rules and disclosure are clear.
- Rule of thumb: you could be wrong in a way that harms learning, so verify before use.
Red: do not do this
- Sharing personal or sensitive data about students or families with outside tools.
- Handing a decision about a child — marks, discipline, placement — to a tool.
- Treating an AI detector as proof of cheating.
Four questions before you use AI
- Does this put anyone's personal data into the tool?
- Will a mistake harm a student, or only waste my time?
- Will I read, check and take responsibility for the result?
- Does it replace my judgement, or support it?
The zones at a glance
| Zone | Meaning | Example |
|---|---|---|
| Green | Use freely, then review | Brainstorm lesson ideas |
| Yellow | Use only after checking | AI-written quiz questions |
| Red | Do not use | AI decides student marks |
Same tool, three zones
Green
Use, then review
- Brainstorm lesson ideas
- Draft a letter with no names
- Reword text for reading level
Yellow
Check before use
- Quiz questions and answer keys
- Translations of terms
- Student use with clear rules
Red
Do not use
- Pasting student data
- AI decides marks or discipline
- Detector score as proof
Why AI detectors cannot be your evidence
A detector does not find proof of AI writing; it gives a guess based on patterns such as how predictable the wording is. Researchers have reported that such tools can wrongly flag writing by people who are not native speakers of English, and a commentary in the journal Patterns concluded that technical detection alone is insufficient and that clear norms for use matter more. Many students write in a foreign language, so the risk of accusing an honest student is close to home.
A fairer approach collects evidence about the learning, not about the tool. Ask for planning notes and drafts, let the student explain the argument aloud for two minutes, or set part of the task in class. Say clearly in advance what help is allowed and how it must be stated. Then a conversation, not a score, decides what happened.
Handling student information: a simple habit
Personal data is more than a name. A photo, a mark, a health note, a family situation, a home address or a written comment about one child can all identify them, especially in a small class where everyone knows everyone. Outside tools may store what you type, and you often cannot see or undo that. A safe habit is to leave identifying details out, describe the situation in general terms, and check your school or ministry rules before using any tool with real information.
Removing the name is not always enough. 'The only left-handed girl in Grade 3B who moved from Khorezm' still points to one child. Ask yourself: could a colleague guess who this is? If yes, rewrite more generally. Often you do not need the child's own words: paste a textbook passage or an invented example instead.
Examples from the classroom
Green: the Navruz letter
Scene. Dilnoza teaches Grade 4 in Samarkand and needs a short letter to parents about the Navruz concert: time, clothes, and a request for help with refreshments. She asks the tool for a friendly draft in Uzbek and gives no child's name. She reads it, shortens two sentences, fixes one phrase that sounds unnatural, and adds the real date herself. The letter is ready in ten minutes instead of thirty.
Insight. Green means low risk: no personal data, a human reads and edits, and a mistake costs only a little time.
Yellow: the quiz with an answer key
Scene. Aziz teaches Grade 8 algebra in Tashkent. He asks the tool for ten linear-equation questions with an answer key. Before printing, he solves every question himself. He finds two wrong answers in the key and one question with two correct answers. He repairs them and prints the quiz. The draft saved time, but only because he checked it.
Insight. Yellow means useful but unreliable: facts, numbers and keys must be verified, because a wrong key harms learning.
Red: the detector score
Scene. Malika teaches Grade 10 literature in a Russian-medium school in Fergana. She runs her student Madina's essay through a free AI detector, which reports a high chance of AI writing. Her first thought is to give a zero. Instead she asks Madina to bring her notes and explain the essay's argument. Madina does so clearly and shows her handwritten drafts. Malika keeps the mark.
Insight. A detector score is not proof, and a decision about a child stays with the teacher, based on real evidence.
Common pitfalls
- Thinking 'no names' is enough when the class, age, place or story still points to one child.
- Placing a task in green because it is quick; speed does not lower risk, the content does.
- Banning all student use without explanation, which pushes it out of sight, or allowing it with no rules at all.
- Pasting real student notes into a tool to write report comments faster.
Sort the Twelve Uses
- In groups of four, take the 12 use-case cards from the Activity Kit and three zone mats: green, yellow, red.
- Place each card on a zone. You must write one sentence of reasoning on the back.
- Swap with another group. Mark any card you would place differently, and say why.
- Write one new use from your own school and place it. Can a colleague find a flaw in your reasoning?
Your facilitator will guide you through this activity in the session.
Questions to discuss
- Which cards caused the most disagreement, and what did the disagreement reveal?
- Which use did you move to a safer zone after discussion?
- What rule would you want your school to adopt about student data?
Reflect
- Which green-zone use could save you the most time this month?
- Have you ever shared something in a tool or chat that you would now think twice about?
- Who in your school decides which tools are allowed?
Try it tomorrow
- Take one task you do this week, write the four questions beside it, decide its zone and note what would move it.
- Write a three-line rule for your class: what AI help is allowed, how to say you used it, and what is not allowed.
- Next time you use a tool, rewrite the prompt so no student could be identified, then compare the result with your first version.
Further reading and links
Ask four questions before you use AI: whose data, what harm, who checks, whose judgement.