Effect of Artificial Intelligence-Based Adaptive Learning on the Academic Achievement and Self-Regulated Learning of Secondary School Students: A Quasi-Experimental Study

Abstract

The promise of artificial intelligence (AI) in education is that software which adapts to each learner might deliver, at scale, some of the benefit of one-to-one tutoring. This study tested that promise in an Indian secondary classroom. Using a quasi-experimental pre-test–post-test non-equivalent control group design, it examined the effect of an AI-based adaptive learning programme on the academic achievement in Science and the self-regulated learning of Class IX students, and asked whether boys and girls benefited equally. Eighty students of a CBSE-affiliated school in Meerut formed the sample, split into an experimental group (n = 40) that learned a Science unit through an adaptive platform and a control group (n = 40) taught the same content conventionally, over six weeks. A researcher-made Science Achievement Test (KR-20 = 0.84) and an adapted Self-Regulated Learning Scale (α = 0.88) were administered before and after the intervention, and the data were analysed using the t-test and analysis of covariance. The experimental group significantly outperformed the control group on post-test achievement (t = 7.34, p < .01; d = 1.66), an advantage that survived adjustment for pre-test scores by ANCOVA (F = 48.60, p < .01), and reported significantly higher self-regulated learning (t = 5.62, p < .01; d = 1.27). Achievement gains did not differ significantly between boys and girls (t = 0.62, p > .05). The findings indicate that AI-based adaptive learning, used to support rather than replace the teacher, can improve both what students learn and how they manage their learning, with implications for the technology agenda of the National Education Policy 2020.