It is 8:40 on a Tuesday morning and the room is loud with thirty students deciding what to write about. A tool on your laptop promises to grade every essay in ninety seconds. Another will summarise the chapter. A third turns speech into text for the student who finds handwriting exhausting. Any of them could help. Each asks something in return: data, attention, or a shortcut past the thinking you were trying to teach.
None of that needs a policy manual the length of your arm. It needs a habit. Look at the tool the way you would look at a new textbook, and ask who benefits, who is exposed, and whether the lesson is genuinely better with it than without it.
Start With the Data, Not the Demo
The demo is always smooth. What matters is what happens to the information students put in. A grammar checker may keep a copy of a fourteen-year-old's writing on a server somewhere. A voice-to-text app may store the recordings. A quiz builder may want names, birthdates and email addresses before it will let you print a worksheet.
Ask three blunt questions before you sign up for anything:
- What is the minimum information this tool needs from my students? It is usually far less than the signup form suggests.
- Where does that information go, how long is it kept, and is it used to train the company's models?
- Is there a class-code or anonymous option that avoids individual accounts altogether?
If the answer is buried somewhere, that is your answer. Check the tool against your school's own rules, and when it touches sensitive records, ask whoever handles data protection at your institution rather than deciding alone. This is not paperwork for its own sake. It is the difference between a useful classroom aid and a leak you have to explain to a parent.
Test It Against Your Own Students
Bias is easiest to spot with examples sitting in front of you. Before you hand a tool to a class, spend fifteen minutes pushing against it.
Ask an image generator for a scientist, a nurse, a farmer, a person using a wheelchair, and see what comes back by default. Run speech recognition with a few different accents, including the ones in your room. Feed a translation tool a sentence full of slang and read the result out loud. Give a writing assistant two versions of the same paragraph — one in formal academic English, one in the language patterns some of your students actually use at home — and note which one it calls incorrect.
Then decide what to do with what you find. Sometimes the fix is a class discussion: look at what this produced. What is missing, and why? That conversation teaches more about how these systems work than any lecture. Other times the fix is simply not using that tool for that task.
Protect the Productive Struggle
Some difficulty is the point. A student who wrestles with a thesis statement for twenty minutes and produces a clumsy one has learned something that a polished sentence would have taken from them.
A rule of thumb that holds up well: use these tools where they remove an obstacle that is not the learning goal, and hold back where they remove the learning goal itself.
- Good fit: speech-to-text for a student who knows what they want to say but cannot get it onto paper; text-to-speech for reading support; practice problem generators that give immediate feedback on a skill you have already taught; vocabulary help while drafting.
- Poor fit: generating the first draft of an essay you are grading for argument; producing the conclusion of a lab report; answering the questions on a reading check meant to show whether the reading happened.
Make that boundary visible. Tell students which parts of an assignment must be their own work, where assistance is allowed, and ask them to note what they used. Clarity works better than policing.
A Checklist for the Next Tool You Try
Before you assign it
- Name the learning goal in one sentence. If the tool does not serve it, stop there.
- Confirm what student data is collected, where it is stored, and whether anonymous accounts are possible.
- Run three or four test prompts, including one from your own subject, and read the output critically.
- Check it against your school's rules and any consent requirements for the age group you teach.
- Decide the alternative for students who will not or cannot use it. There has to be one.
While students use it
- Say out loud what the tool is for and what it is not for.
- Watch whether students are reading the output or just accepting it.
- Save a few examples of wrong or strange results for a five-minute discussion afterwards.
After the lesson
- Ask two questions: what did this help you do, and what did it get in the way of?
- Compare the work with what the same class produced without the tool. If nothing changed, it may be costing more than it gives.
- Write your decision down in one line, with the date. In June you will not remember why you kept it or dropped it.
Keep Families in the Conversation
Parents and carers hear a great deal about these tools and very little about what happens in your classroom. A short note home — which tools are used, for what, what information they require, and how to request an alternative — prevents most of the awkward conversations later. Keep it in plain language and avoid promising more than the tool's own documentation supports.
With students, transparency beats prohibition. If you explain that you need to know which parts of a piece of writing are theirs, most will tell you honestly, provided they are not afraid of getting in trouble for asking the question.
Start With One Tool and Four Weeks
Choose a single tool, one class, and a clear purpose. Use it for a month. Keep a short log — what you used it for, what data it needed, what the students said — then decide whether to keep it, change how you use it, or drop it.
That pace is slower than the marketing and faster than a committee. It also keeps the decision where it belongs: with the teacher who knows the room, the students, and the lesson that was actually supposed to happen.
Photo: sweetlouise / Pixabay
