Utilization of Ai-Assisted Instruction in Enhancing Mathematics Engagement and Achievement of Senior High School Students

Abstract

This study examined the utilization of AI-assisted instruction in enhancing the mathematics engagement and achievement of Grade 11 senior high school students in Data and Probability. A quasi-experimental pretest-posttest design with partial counterbalancing of instructional conditions was used among 131 students. The study compared traditional instruction, MathGPT-assisted instruction, and Wolfram Alpha-assisted instruction in terms of behavioral engagement, emotional engagement, overall learning experience, mathematics achievement, and perceived usability and accessibility. Mathematics engagement and usability/accessibility were measured using Likert-scale questionnaires, while mathematics achievement was measured through pretest and posttest scores based on six identified competencies. Weighted mean, standard deviation, Friedman test, and paired-samples t-test were used to analyze the data. Results showed that all instructional approaches produced high levels of mathematics engagement. MathGPT obtained the highest behavioral engagement rating (WM = 4.07, SD = 0.80), while Wolfram Alpha obtained the highest emotional engagement rating (WM = 3.93, SD = 0.89). For overall learning experience, MathGPT received the highest weighted mean (WM = 3.97, SD = 0.89), followed by Wolfram Alpha (WM = 3.95, SD = 0.84), and traditional instruction (WM = 3.75). Wolfram Alpha and MathGPT both obtained the highest usability and accessibility rating (WM = 4.03). Paired-samples t-test results indicated significant improvement between pretest and posttest scores across all competencies and treatment groups, with p-values below 0.05. However, the Friedman test showed no significant difference in mathematics engagement among the three instructional approaches, χ²(2) = 2.42, p = 0.299. The study concludes that traditional instruction, MathGPT, and Wolfram Alpha all contributed positively to mathematics engagement and achievement, with AI-assisted instruction showing slight descriptive advantages. An enhanced AI-assisted mathematics instructional plan is proposed to guide responsible, teacher-mediated AI integration in Data and Probability.