Adaptive Socratic AI Scaffolding Supports Critical Thinking in Indonesian Secondary School Physics Classrooms
This study demonstrates that an adaptive Socratic AI scaffolding system, developed through the ADDIE model and validated for content, significantly enhances critical thinking skills among Indonesian secondary physics students compared to traditional guided-inquiry methods by guiding reasoning rather than providing direct solutions.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the study of physics, the goal has always been more than just memorizing formulas or plugging numbers into equations. True understanding requires a student to look at a problem, question their own assumptions, weigh the evidence, and construct a logical argument for why a specific answer makes sense. This ability to think deeply and critically is essential for scientists, yet in many classrooms, students often learn to follow procedures without grasping the underlying reasoning. They can solve a math problem on paper but struggle to explain the physical principles behind it. For decades, educators have tried to fix this by using guided inquiry, a method where students investigate problems step-by-step, but even with these efforts, many learners still rely on rote memory rather than deep analysis. The challenge has been finding a way to support students as they struggle with these complex ideas without simply giving them the answers, a balance that is difficult for a single teacher to maintain in a crowded classroom.
A new approach emerging from Indonesia offers a solution by combining human teaching with a carefully designed artificial intelligence system. Researchers at the State University of Malang set out to build a digital tool that acts not as a tutor who provides solutions, but as a conversation partner who asks the right questions. This system uses a method known as the Socratic approach, where an intelligent agent guides a learner by asking probing questions that force them to clarify their thoughts, examine their evidence, and reconsider their conclusions. Instead of telling a student what is wrong, the system asks, "What assumption are you making here?" or "How does this piece of evidence support your claim?" To make this work effectively, the researchers added a layer of adaptability, meaning the system can detect how a student is thinking and adjust its questions accordingly. If a student is confused, the system offers more support; if the student is thinking clearly, the system steps back, allowing the student to work independently. This creates a dynamic learning environment where the technology acts as a temporary scaffold, helping students climb higher until they can stand on their own.
To test whether this idea works in a real-world setting, the research team developed a complete learning system for high school physics students. They built a digital laboratory where students could conduct experiments and solve problems, but with a twist: the experimental group used a version of the lab powered by this special AI, while a control group used a standard version without the intelligent questioning. The study involved 135 students from two different high schools in Southeast Sulawesi. Over the course of seven sessions, both groups worked through the same guided inquiry lessons. The students in the experimental group, however, interacted with four different AI agents, each designed to help at a specific stage of the learning process. One agent helped them define the problem, another guided them through the experiment, a third helped them analyze their data, and a fourth was available to answer general questions. Crucially, none of these agents ever gave a direct answer. Instead, they were programmed to ask questions that kept the students thinking, ensuring the learners did the heavy lifting of reasoning.
Before the classrooms were opened, the system was rigorously checked by experts in physics education and technology. These specialists reviewed the questions the AI asked and the way it supported students, rating the tool on a scale to ensure it was pedagogically sound. The experts gave the system a very high approval rating, confirming that the questions were well-designed to foster critical thinking and that the technology functioned reliably. Once the system was ready, the researchers measured the students' critical thinking skills before the lessons began and again after the seven sessions were complete. The results showed a clear difference between the two groups. While both groups started with similar levels of ability, the students who used the AI-enhanced system improved significantly more than those who used the standard lab. After adjusting for their starting scores, the experimental group achieved an average score that was more than ten points higher than the control group. This gap was substantial, indicating that the students who engaged with the questioning AI were not just learning more facts, but were actually becoming better at analyzing arguments, evaluating evidence, and constructing their own explanations.
The study suggests that the success of this approach lies in how the AI was constrained. The system was not allowed to act as a shortcut to the answer. By forcing students to articulate their reasoning and defend their ideas through a dialogue with the machine, the tool kept them engaged in the difficult work of thinking. The researchers found that when the AI was programmed to fade its support as students became more competent, it encouraged them to take ownership of their learning. This method proved that artificial intelligence, when used correctly, does not have to replace the teacher or the student's effort. Instead, it can serve as a disciplined partner that challenges learners to think deeper, turning the classroom into a space where the focus shifts from getting the right answer to understanding the path that leads there. The findings offer a promising path forward for physics education, showing that with the right design, technology can help students develop the critical thinking skills they need to navigate a complex world.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.