How Students Should Use AI Productively in the Classroom

Matthew Esterman was once leading a forum with a group of students in Victoria when one girl mentioned how she does her maths homework. She works it out on a piece of paper, takes a photo with her phone, uploads it to an AI chatbot inside a social app and has a conversation with it about her work.

Matt hadn't come across that before. The closest comparison Matt could draw was sending your work to a tutor for feedback, or showing it to a parent or a peer. To Matt, getting feedback from this other kind of intelligence was something very new and different.

So how should students use AI? Productively, which means the student keeps doing the thinking. If a student hands all their effort and thinking over to an AI, they stop progressing in their learning. Productive use looks like a student leading the AI: asking it to explain something, checking the answer against another source, and working out what it got wrong or left out. The AI supports their thinking and leaves the thinking itself to them. Whatever the task, start with your school's policy on AI, and ask if there isn't one.

These ideas come from Matt, AI consultant, author and founder of The Next Word, an AI consultancy for education. Matt co-authored a book of the same name, a practical guide for teachers adopting AI, and was a teacher for over 20 years, trained in history with a long interest in e-learning. In the course, Matt sums up what students should aim for in one line: "AI helps me think… it won't do the thinking for me."

The ideas

1. Ask whether the AI use is productive.Whenever AI is an option, whether you're designing an assessment task or leading a lesson, Matt asks one question: "Are they being productive with it or are they being unproductive with it?" Students can ask themselves the same thing about their own tasks.
2. Watch for replaced thinking.The clearest sign of unproductive use is a student handing everything over. In Matt's words, "Are you replacing all your effort and thinking with AI? Well, that's pretty unproductive." The student isn't doing anything at that point, and their learning stalls.
3. Look for the student's own thinking.Matt sorts AI use into a set of zones, based on how much of the student's thinking you can see and who is in control. Matt is upfront that this is a personal way of thinking about it, with no academic backing: "this is my brain on a page."
4. Check it against another source.Ask AI to explain photosynthesis. Then, as Matt puts it, "check it against another source. What's missing?" Students open Britannica, Wikipedia or another website and compare the two. You can run this at the front of the room.
5. Let students be the critics.Have AI explain something like the water cycle, then ask students to rewrite it in their own voice. Matt puts it this way: "I find students much, much more harsh as markers and as critics on AI generated information." Tearing the text apart builds the habit of reading it critically.
6. Know the rules before the task starts.Matt is clear on this: "if a school that you're working in has a policy, you do have to follow it." If you're asked to mark, collect or monitor an assessment task, ask whether AI is OK to use for that particular task. Where there's no policy, ask, and Matt suggests keeping a quick checklist of what you tell students so they use AI ethically and don't open themselves up to data breaches.

AI in Education: Bots, Agents & The Revolution We Are InStart this course

Productive and unproductive use, side by side

Here's how the two look in practice, drawn from Matt's examples in the course.

Matt holds teachers to the same standard and treats AI output like a Google result: never use it straight away, always check your sources, because if we ask students to do it, we should too. On privacy, Matt asks teachers to think carefully about when they do and don't use AI with students, so students aren't sharing personally identifiable information with these systems.

Why this matters beyond your classroom

Students are taking these tools beyond the school gate and using them however they like, completely out of the school's control. Matt's response is to bring students into the conversation. Listen to how they're already using these tools, sometimes quite intelligently, and steer them towards the higher levels of use, so they do more with AI than the old copy and paste.

There's a wellbeing side to this too. Some students go to chatbots looking for a kind of therapy or companionship. It can be very light touch, but Matt notes it goes quite deep for some students.

How to use this in your classroom this week

• Run the photosynthesis check. Ask AI to explain a topic at the front of the room, open another source, and ask the class what's missing.
• Turn an AI explanation into a marking sample. Put it up the way you would any sample and ask what's good about it, what mark it deserves and why.
• Talk about how students are using it. When AI is allowed, talk with students about how much of the thinking is theirs and how much is the AI's.
• If you're a relief or casual teacher, and you're asked to collect or monitor an assessment task, ask whether AI is OK to use for it. If a student discloses something to you, even when you see them for one lesson, it still matters under child safety. Follow it up as you would any other disclosure, using your school's child safety procedures.

For national guidance, the Australian Framework for Generative AI in Schools (Department of Education) sets out how schools can use generative AI tools responsibly and ethically, for the benefit of students, schools and society.

Take it further with your students

This post covers one question. In AI in Education: Bots, Agents & The Revolution We Are In, Matt goes much further. Inside the course, you'll find:

• Matt's full model of AI use, zone by zone, and how to talk students through each one
• a structured debate between students and AI
• a paragraph scaffold students can build with AI when they're stuck
• what chatbots, wrappers and agents are, including how to build a simple agent of your own
• ways to use AI to cut your admin, from marking rubrics to quizzes at three levels
• a role play on future jobs, and what students can do better than a bot
• the ethics to weigh up, from privacy and bias to what "all my own work" means now

If you're newer to AI, Matt's first course, AI for Teachers, is a good place to begin. You'll find more courses for high school teachers in secondary teacher PD.

Matt likens a school's role here to drugs and alcohol education or healthy eating: there's only so much a school can say, but "we can provide a safe environment and the best information possible to the students."

HBWritten by Helen Bryant, Director of Education, ClassCover Learn. Helen has been teaching for over twelve years in both government and independent schools, across all stages from K-6.