If Students Never Get a Voice in Class, What Are We Teaching Them About Democracy?
From having a say to having an impact: why student agency matters.
AI can get students to the summit. But did they climb?
Same summit, different learning: one student climbed, the other arrived by helicopter.
Imagine two students at the top of a mountain. Same summit, same view, same photo at the top.
One of them climbed. The other arrived by helicopter.
If all we look at is the destination, they’ve achieved exactly the same thing. But one chose a route, made mistakes, changed direction, kept going when it got hard, and learned something about climbing. The other just got there faster.
That’s what AI has suddenly exposed in education. AI can get our students to the summit. Very quickly.
It’s one of the questions I hear most from teachers, and I get it. Students go home and come back with an elegantly structured essay, impressive vocabulary and, miraculously, flawless punctuation.
So we start asking: Can I detect it? Should I ban AI? Back to pen and paper?
Sound familiar? Set the essay. AI writes it. We try to catch it. Students find a smarter tool. We set the next essay. And round and round we go.
The AI “cheating” cycle, and the way out of it.
But there’s another question hiding underneath: What were we hoping students would learn on the way up?
Because if all we assess is the finished product, AI is remarkably good at producing it.
“Write 500 words about climate change.” Done.
“Summarise this chapter.” Done.
“Give me five arguments for and against school uniforms.” Done.
The summit has been reached. We still don’t know if anybody climbed.
Tony Frontier’s work on AI with Intention makes a distinction I keep coming back to: using AI to complete schoolwork is not the same as using AI to support learning.
Maybe AI hasn’t just made cheating easier. Maybe it has shown how easily we mistake a finished product for evidence of learning. A polished essay tells me what exists at the end. It doesn’t tell me what happened in the student’s head along the way.
So the question shifts from “How do I catch the AI?” to “How do I make the thinking visible?”
Here’s one simple idea. Alongside the final essay, presentation or project, students hand in a Thinking Trail. Not another enormous reflection sheet, we all have enough paperwork. Just a few questions that show the route they took:
And my favourite:
What’s in the final piece because of YOU?
The Thinking Trail: assess the journey, not just the destination.
That interests me far more than “Did you use ChatGPT?” Now I’m looking for the student’s fingerprints: their decisions, questions, change of mind, mistakes, voice.
AI stops being something we try to catch hiding behind the product. It becomes something the student has to account for. And if AI did most of the journey? That’s much easier to find out.
This is where Catlin Tucker’s idea of Skills Before Tools comes in.
Back to the mountain. A real climber can read the terrain. They notice when a path looks unstable, when the weather is turning, when they’ve taken a wrong turn. Someone dropped off by helicopter can’t. They don’t even know if they’re standing on the right peak.
It’s the same with AI. The most important AI skill isn’t writing the world’s cleverest prompt. It’s knowing enough to say:
“I’m not sure that’s right.”
“That’s one perspective. What’s missing?”
“I can make this better.”
“Actually, I don’t need AI for this.”
Skills before tools: a climber can read the terrain. A helicopter passenger can’t.
Questioning, evaluating, checking, revising, explaining, making judgements. These aren’t AI skills. They’re human skills, and AI makes them more important, not less. When a tool can produce something polished and convincing in seconds, students need the judgement to decide whether it’s actually any good.
Sometimes, yes. If the agreement was “do this independently” and AI did it instead, that’s pretty straightforward.
But the more interesting question is: Where did the student do the thinking?
If AI generated the ideas, made the decisions and wrote the piece while the student just moved it from one screen to another, AI did the climbing.
If the student used AI to explore possibilities, then questioned them, checked facts, rejected suggestions and can explain their final choices, that’s different. The tool helped. The student still climbed.
The question that matters: where did the student do the thinking?
One more uncomfortable thought. If students can hand an assignment to AI and lose almost nothing from the learning… maybe we need to look at the assignment.
AI isn’t only challenging students. It’s challenging us to ask: Where in this task will students make decisions? Where will they struggle productively? Where does their own perspective matter? What evidence of learning am I actually looking for?
Those questions are much harder than downloading an AI detector. But they’re also much more interesting.
In Teaching in the AI Age, our 5-day Erasmus+ course at ITC International Training Center, we go beyond the tools and look at the teaching behind them. You’ll explore:
It’s practical, hands-on and full of discussion with teachers from across Europe. You’ll leave with strategies you can use straight away.
AI can get your students to the summit. Come and learn how to make sure they still climb.
From having a say to having an impact: why student agency matters.