As artificial intelligence quietly reshapes the landscape of learning, a team of researchers across four universities has asked a question that grades alone have never been able to answer: what actually happens inside the mind of a student at work? Their method, called Movement Analysis, watches the living process of inquiry rather than its finished remains — and in doing so, suggests that the true measure of education has always been the journey, not the destination.
New AI Method Shifts Education Focus From Student Output to Learning Process
We need to understand the process behind the product.
Mark
So the core problem here is that teachers can't tell if a student actually learned something or just got an AI to do the work?
Mimi
Exactly. A finished essay tells you almost nothing anymore. You need to see the thinking that led to it—the questions asked, the false starts, the moments where the student pushed back or built on someone else's idea.
Mark
And MOVA watches all that happening in real time?
Mimi
It tracks the movements—the transitions between different kinds of thinking. When a group moves from just sharing ideas into actually integrating them, that's a movement. When they get stuck in social chat and never push deeper, that's visible too.
Mark
Did the chatbot actually make students think better, or did it just make them produce better work?
Mimi
The data suggests it genuinely deepened their thinking. Groups with the chatbot moved between inquiry stages more often and went deeper into integration and problem-solving. Groups without it stayed in shallower patterns.
Mark
What happens if a teacher actually uses this? How does it change what they do?
Mimi
They stop waiting for the final product and start watching the process. They can see where a group is struggling and offer help before the work is done. They can teach toward actual understanding instead of just grading outputs.
The Pulse
- Traditional grading is going blind: as AI tools do more of the visible work, final essays and projects reveal less and less about whether genuine learning ever took place.
- A study of 108 university students exchanging over 1,600 messages exposed a troubling pattern — groups without AI assistance tended to circle the shallows, never quite breaking through into deeper, integrative thinking.
- The AI-assisted groups moved differently, transitioning more fluidly between questioning, exploration, and synthesis — suggesting that the right tool, used well, can unstick a mind rather than replace it.
- MOVA — Movement Analysis — offers teachers something they have never had before: a map of how thinking unfolds in real time, making visible the moments when students need support before the moment has already passed.
- The stakes extend well beyond one study: if assessment doesn't evolve alongside AI, education risks measuring only what machines can produce, losing sight of the critical, collaborative thinking that defines genuine human learning.
As artificial intelligence quietly reshapes the landscape of learning, a team of researchers across four universities has asked a question that grades alone have never been able to answer: what actually happens inside the mind of a student at work? Their method, called Movement Analysis, watches the living process of inquiry rather than its finished remains — and in doing so, suggests that the true measure of education has always been the journey, not the destination.
Teachers have long faced a quiet impossibility: a polished final submission tells them almost nothing about what actually happened in a student's mind. Did the group wrestle with hard ideas together, or did they simply let an AI carry the weight? As artificial intelligence becomes a natural part of how students work, grading only the end product has grown nearly meaningless.
Researchers at The Education University of Hong Kong, collaborating with colleagues from Monash University, the University of Wisconsin–Madison, and Northwest Normal University, decided to look at the problem differently. Dr. Shen Ba and his team developed Movement Analysis — MOVA — a method that tracks how students move through the actual stages of inquiry: questioning, exploration, social connection, integration, and resolution. Rather than waiting for the final paper, MOVA watches the thinking as it happens.
The team tested the method with 108 university students working in 16 groups, analyzing 1,617 messages exchanged over the course of their collaboration. Some groups had access to GPT-4-powered chatbots; others did not. The difference was striking. AI-assisted groups moved more fluidly between thinking levels — asking deeper questions, pushing into integration, and arriving at real solutions. Groups without the chatbot tended to stay in the shallower stages, circling social connection and basic exploration without breaking through.
The implications reach far beyond a single study. If teachers see only the finished essay, they miss where a group got stuck, what support might have helped, and whether any student was genuinely engaged. As Ba put it, assessment must evolve alongside AI — not to catch students cheating, but to understand the process behind the product. Published in Computers & Education, the research points toward a future where teachers can intervene at the exact moment a student needs guidance, nurturing critical thinkers rather than simply grading the outputs that AI can now generate on demand.
Teachers have a problem that no final exam can solve. A student submits a polished essay, a finished project, a complete solution—and the teacher has no way to know what actually happened in the student's mind. Did the student think through the problem step by step, or did an AI chatbot do the heavy lifting? Did the group collaborate and push each other's thinking, or did they just copy and paste? As artificial intelligence becomes woven into how students work, the old way of grading—looking only at what comes out at the end—has become nearly useless.
Researchers at The Education University of Hong Kong set out to answer a harder question: What if we could see the learning itself, not just the product? Dr. Shen Ba and his team, working with collaborators from Monash University, the University of Wisconsin–Madison, and Northwest Normal University, developed a method called Movement Analysis, or MOVA. It sounds technical, but the idea is simple. Instead of waiting for the final paper, MOVA watches how students move through the actual work—how they ask questions, explore ideas, build on each other's thinking, integrate what they've learned, and arrive at solutions. It's like watching the thinking happen in real time.
To test the method, the researchers gathered data from a university course with 108 students working in 16 groups. Over the course of their collaboration, these groups exchanged 1,617 messages. MOVA tracked their movements through distinct stages of inquiry: questioning, exploration, social connection, integration, and resolution. The researchers then introduced AI chatbots powered by GPT-4 to some groups and left others to work without that tool. What they found was striking. Groups with access to the chatbot showed something the others didn't: they moved more fluidly between different kinds of thinking. They asked more questions, explored more deeply, and crucially, they didn't get stuck. They moved from simple idea-sharing into the harder work of actually integrating those ideas and building toward real solutions.
Groups without the chatbot fell into a different pattern. They tended to circle in the same modes of thinking—staying in social connection or basic exploration without pushing forward into the deeper, more demanding stages of inquiry. Some groups showed longer, more dynamic movements through the learning process, reflecting sustained engagement and complex thinking. Others remained confined to the surface, never quite making the leap from generating ideas to actually doing something with them. MOVA made these patterns visible in a way that looking at the final product never could.
The implications ripple outward from this single study. If teachers only grade the final essay or project, they miss the entire landscape of how a student thinks. They can't see where a group got stuck, what kind of support might have helped them move forward, or whether a student was genuinely engaged in the work or just along for the ride. Ba put it plainly: "As AI changes how students learn and complete academic work, assessment also needs to change. We need to understand the process behind the product." This isn't about catching students cheating. It's about fundamentally rethinking what it means to assess learning in a world where AI is a tool students actually use.
The research, published in Computers & Education, suggests that the future of education assessment lies not in the final output but in the journey. Teachers who can see how students move through inquiry stages—how they respond to feedback, how they collaborate, how they think—can offer support at the exact moment it matters most. They can guide students toward becoming reflective, critical thinkers rather than just producers of acceptable work. In an era when AI can generate polished answers on demand, the real measure of learning is no longer what students can produce. It's how they think, how they work together, and how they grow.
Notable Quotes
As AI changes how students learn and complete academic work, assessment also needs to change. We need to understand the process behind the product.— Dr. Shen Ba, assistant professor at The Education University of Hong Kong
Want the full story?
Read the original at Phys.org ↗
Technology