Adaptive AI-Based Assessment Systems: Enhancing Academic Resilience and Reducing Test Anxiety in Higher Education

Abstract

This research investigates the impact of the ‘adaptive AI-based assessment system’ on the psychological and academic outcomes of university students with a particular emphasis on anxiety, self-efficacy, and academic resilience. Stress and lack of confidence are commonplace psychological responses for assessments due to their rigid and highly-stakes nature. AI-driven assessments, on the other hand, offer personalized assessments of varying levels of difficulty with real-time feedback and responsive question adaptation, fundamentally concerning the assistance and support offered to students. This study uses a quasi-experimental method on a sample of students enrolled in higher education and probes the differences of traditional assessment methods to evaluations with adaptive AI tools. Anxiety, self-efficacy and academic resilience were assessed in the sample, and the resulting data were analyzed through descriptive and inferential statistics and structural equation modelling. In comparison to the traditional assessment methods, the AI based assessments showed conclusively that students had less anxiety and reported greater self-efficacy and resilience. Resilience and self-efficacy were found to be mediating variables and were able to provide an explanation for the reason adaptive testing reduces emotional strain while simultaneously increasing academic confidence. This study asserts that the AI-driven assessment tools can not only enhance student's academic outcomes, but also their overall mental well-being during their stress periods. This reflects a change in the attitude of the higher educational institutions with an intention to offer students more humane and less stressful academic assessments.