Abstract

This study presents a preliminary evaluation of the effectiveness of a procedure for checking AI-generated geographic information against multiple sources in supporting grade 11 students’ ability to evaluate credibility and construct geographic arguments. A quasi-experimental pretest–posttest study with a control group was conducted over four lessons with 82 students at one upper secondary school, with 41 students in each group. After adjustment for pre-intervention scores using analysis of covariance (ANCOVA), the experimental class scored 0.452 points higher than the control class in geographic argumentation on a 0–3 scale (95% CI [0.292, 0.612], p < 0.001; Hedges’ g = 1.35) and 0.455 points higher in credibility evaluation (95% CI [0.303, 0.607], p < 0.001; Hedges’ g = 1.13). These preliminary findings suggest that the procedure may help students verify information and use geographic evidence. Further research across more classes and schools is needed to corroborate these findings.

Keywords: Credibility of geographic information, generative AI, geographic argumentation, geography teaching, multi-source information checking.