Bukit Timah Secondary School, Bukit Timah, Singapore

Abstract: Traditional chemistry assessment in Singapore secondary schools relies heavily on written examinations and practical tests, which may not capture the full spectrum of student learning. This paper presents the development and implementation of an AI-Enhanced Multimodal Assessment Portfolio (AI-MAP) system for secondary chemistry education. The AI-MAP system integrates five assessment dimensions—conceptual understanding, scientific inquiry and practice, chemical thinking and reasoning, communication and collaboration, and self-regulation and metacognition—aligned with Singapore's 21st Century Competencies (21CC) framework. Artificial intelligence serves as a "multimodal analysis engine" that analyzes diverse student artifacts, including written assignments, laboratory practical work, oral presentations, digital projects, and reflective journals. The system was implemented with 86 Secondary 3 students in a co-educational government secondary school over a 12-week "Chemical Bonding and Structure" unit. Findings indicate significant improvements across all five assessment dimensions, with particularly large gains in scientific inquiry and practice (d = 0.89) and self-regulation and metacognition (d = 0.82). Students reported increased engagement and a deeper understanding of their own learning progress. Teachers noted that the AI-MAP system reduced assessment workload by approximately 35% while providing richer, more comprehensive insights into student learning. The study contributes to the growing body of research on AI-enhanced assessment in science education and offers practical implications for Singapore secondary chemistry teachers seeking to implement more holistic assessment approaches.

Keywords: multimodal assessment; AI in education; chemistry education; portfolio assessment; 21st century competencies; secondary education