This study aims to examine the effect of Artificial Intelligence (AI) utilization on students' mathematical problem-solving ability among eleventh-grade students at SMAN 1 Wondama, West Papua. This study employed a quantitative approach with a quasi-experimental pretest-posttest control group design. The sample consisted of two classes, namely class XI Science 1 as the experimental group that used AI (ChatGPT) during the learning process and class XI Science 2 as the control group that received conventional instruction. Data were collected through a HOTS-based essay test of mathematical problem-solving ability, an AI-use questionnaire, and a student-activity observation sheet. Data were analyzed using normality, homogeneity, N-Gain, t-test, and effect size analyses. The results indicate that the improvement in mathematical problem-solving ability was higher in the class that used ChatGPT in a guided manner than in the conventional class (N-Gain of 0.62 versus 0.38; t = 3.15 > t table = 1.99; Cohen's d = 0.71), leading to the conclusion that guided AI utilization has a positive and significant effect on students' mathematical problem-solving ability. Questionnaire and observation data further indicate that this improvement was more pronounced among students who used AI reflectively than among those who merely copied answers, suggesting that usage attitude serves as a supporting factor alongside the main treatment.
Copyrights © 2026