This study examines the effectiveness of Problem Based Learning assisted by ChatGPT in improving students' mathematical critical thinking and mathematical resilience. The study employed a quantitative quasi-experimental design using a nonequivalent pretest-posttest control group involving three intact seventh-grade classes at UPTD SMP Negeri 1 Buntu Pane in the 2025/2026 academic year. The sample consisted of 65 students selected through saturated sampling. Data were collected using a four-item essay test of mathematical critical thinking and a 20-item mathematical resilience questionnaire. Data analysis included descriptive statistics, normalized gain, assumption tests, one-way ANOVA, post hoc comparison, and two-way analysis at a 5% significance level. The results indicate significant differences among the three learning models in mathematical critical thinking and resilience. PBL assisted by ChatGPT produced the highest normalized gain for critical thinking (0.584) and resilience (0.746). Post hoc analysis showed that PBL assisted by ChatGPT and ordinary PBL did not differ significantly in critical thinking, but PBL assisted by ChatGPT was significantly superior in mathematical resilience. These findings suggest that the pedagogical strength of PBL lies in promoting critical reasoning, whereas ChatGPT contributes additional value by providing adaptive scaffolding that supports students' confidence, persistence, and willingness to re-engage with mathematical challenges.
Copyrights © 2026