The rapid advancement of artificial intelligence (AI) technology necessitates a fundamental shift in educational goals—from the simple transmission of standardized knowledge to fostering learners’ abilities to collaborate with AI, solve complex problems, and generate new value. This study implemented an AI-integrated chemistry class that emphasized the interpretation of regression models, targeting general high school students, to examine its impact on students’ AI literacy and knowledge-information processing competency. The class covered two topics—surface tension and gases—each organized into five stages: concept introduction, AI modeling, model interpretation, explanation introduction, and summary. Students’ AI literacy and knowledge-information processing competency were measured through pre- and post-tests and analyzed using the Wilcoxon signed-rank test to examine differences before and after the class. In addition, students’ perceptions of the class were collected through a post-survey and qualitatively categorized to identify response patterns. The results revealed statistically significant improvements in students’ AI literacy and knowledge-information processing competency. Moreover, students reported that the class supported their understanding of chemistry concepts and highlighted the importance of social interactions with teachers and peers in reducing cognitive load and enhancing learning effectiveness. They also demonstrated strong motivation to pursue further learning through advanced theoretical learning and additional data analysis activities. These findings suggest that AI modeling and interpretation practices, collaborative interactions, and the integration of follow-up enrichment activities are critical components of effective AI-integrated science education.