The Impact of AI ‐Supported Think‐Pair‐Share Instruction on Students' Problem‐Solving Skills and Motivation in Learning Quadratic Equations: A Social Constructivist Perspective

ABSTRACT Background Quadratic equations are a cornerstone of algebra and serve as a gateway to advanced mathematical concepts. Despite their importance, students frequently struggle with conceptual understanding, procedural accuracy, and sustained motivation when learning mathematics, particularly quadratic equations. Traditional instruction often fails to address these difficulties, leading to persistent misconceptions and disengagement. Artificial intelligence (AI) shows promise in offering adaptive feedback and customised support, but research mainly focuses on its role in helping individual learners. On the other hand, the Think‐Pair‐Share (TPS) model, grounded in Vygotsky's social constructivist theory, has been proven effective in fostering collaboration, critical thinking, and communication. However, limited studies have explored how AI can be systematically integrated into structured collaborative learning approaches such as TPS to enhance both problem‐solving and motivation. Rationale and Objectives This study introduces a novel instructional model that embeds AI scaffolding into each phase of the TPS process. Rather than replacing teacher facilitation, AI functions as a supportive scaffold that delivers individualised prompts during the Think stage, facilitates peer negotiation in the Pair stage, and provides clarification or reinforcement in the Share stage. This integration ensures consistent, contextualised guidance while maintaining the collaborative essence of TPS. The study therefore, aims to examine the effectiveness of AI‐supported TPS instruction in enhancing junior high school students' problem‐solving skills and motivation in learning quadratic equations. Methods This study employed a quasi‐experimental design with a non‐equivalent control group. A total of 43 junior high school students participated, with the experimental group receiving AI‐enhanced TPS instruction and the control group following traditional TPS instruction with online Google searches. Data were collected using two instruments: a problem‐solving test to measure students' learning outcomes and a mathematics motivation questionnaire to assess intrinsic motivation, interest, and self‐efficacy. The data were analysed using analysis of covariance (ANCOVA) to control for pre‐test differences and determine the impact of the intervention. Results and Conclusions The findings indicate significant improvements in both problem‐solving skills and motivation for the experimental group compared to the control group. Integrating AI tools within the TPS framework not only enhances cognitive outcomes but also fosters greater student engagement, highlighting the potential of technology to transform mathematics education.