<![CDATA[This study developed and evaluated ChemisFun, an AI-assisted Android application for learning the Basic Laws of Chemistry. The application was developed using the ADDIE model and designed around Discovery Learning, incorporating animated videos, data-exploration activities, games, and a multiple-choice evaluation feature. AI supported content development only: Gemini AI created illustrations, ChatGPT assisted in preparing learning materials and questions, and Flow AI generated videos representing macroscopic, submicroscopic, and symbolic aspects of chemistry. The application and its AI-generated content were checked through validation by one material expert and one media expert, with test items judged by two assessment experts; mean scores for the material and media validation were 4.50 and 4.54, respectively (very feasible). A single-group pretest–posttest design was used with 72 undergraduate students of Universitas Negeri Medan who owned Android devices. The intervention was one 100-minute session (excluding the pretest, posttest, and questionnaire) in which two lecturers acted as facilitators and students used ChemisFun independently. The mean score on a 20-item multiple-choice test increased from 47.85 to 90.56 (Wilcoxon signed-rank test, p < 0.001); the mean normalized gain (N-gain) was 0.81 (high), with 52 students in the high and 20 in the medium category. Motivation, measured only once after the session, was very high (M = 4.69). These findings indicate that ChemisFun met the expert-feasibility criteria and that its use was followed by substantial learning gains and high reported motivation. Without a control group or delayed posttest, causal effects and long-term retention cannot be established.]]>
