Revolutionizing Science Education Evaluation Using a Vision Language Model of Effective Assessment and Supervision

Implementing the Independent Curriculum in Indonesia presents challenges in educational assessment, especially in efficiently evaluating students' essay responses. This study aims to develop and test an educational evaluation model based on artificial intelligence (AI) technology, especially the vision language model, which can automate and improve the process of assessing student learning outcomes in science. This study explores the potential of Vision Language Models (VLMs) as an innovative solution. This study uses a mixed sequential explanatory method. The subjects in this study were junior high school students. The data collection method used interviews. Data collection instruments with questionnaires. Data analysis techniques used were qualitative, quantitative, descriptive analysis, and inferential statistics. The study results are that integrating VLMs increases the efficiency and objectivity of assessment. This study concludes that VLMs can reduce teacher workload, improve feedback, and show synergy between technology and curriculum reform in the Independent Curriculum era. The implications of this study are very significant for the development of science and technology education; the use of vision language models (Vision-Language Models) in evaluating science education can increase the accuracy and objectivity in assessing student learning outcomes.