Impact of AI-Supported Learning on Physics Anxiety and Academic Achievement among Higher Secondary Students

Abstract

The rapid diffusion of artificial intelligence (AI) tools into classroom instruction has generated considerable interest in their potential to reshape both the cognitive and affective dimensions of learning, particularly in subjects that students traditionally perceive as difficult, such as physics. This paper examines the influence of AI-supported learning on physics anxiety and academic achievement among higher secondary school students. Physics is frequently associated with mathematical demand, abstract reasoning, and evaluative pressure, all of which contribute to subject-specific anxiety that can depress performance independent of ability. Drawing on Bandura’s (1997) social-cognitive theory, Vygotsky’s (1978) sociocultural theory, and the Yerkes-Dodson (1908) law of arousal and performance, the paper proposes that adaptive, low-stakes, and immediately responsive AI tools can lower debilitating anxiety while simultaneously strengthening conceptual understanding through personalised scaffolding. A review of recent empirical and meta-analytic literature indicates consistent, moderate-to-large positive effects of generative and adaptive AI tools on both achievement and anxiety-related outcomes across school levels (Alneyadi & Wardat, 2023; Liu et al., 2025; Ma & Zhong, 2025; Wang & Wei, 2025). The paper further outlines a quasi-experimental, pretest-posttest, non-equivalent control-group design suitable for testing these relationships among Class XI/XII physics students, along with the instruments, procedure, and statistical techniques appropriate for such a study. Implications for physics pedagogy, curriculum design, and teacher training are discussed, along with the limitations that should guide interpretation of AI-based interventions in psychologically sensitive subject areas.