REINVENTING CURRICULUM THROUGH AI-BASED VISUAL FEEDBACK

Abstract

The paper introduces a unified system of AI-based visual feedback in order to redefine the process of curriculum design with the help of information-driven personalization and adaptive learning. The proposed system will integrate Convolutional Neural Networks (CNNs) to extract visual features, Long Short-Term Memory (LSTM) networks to model a temporal sequence, and the Explainable AI (XAI) to be interpretable. The model is the integration of multimodal learning data, including visual artifacts, behavioral logs, and contextual records, in generating real-time visual feedback to enable self-regulation of the learner and decision-making by the educator. The experimental validation of two large scale datasets has shown an accuracy score of 94, F1-score of 0.92 and Visualization Clarity Score (VCS) of 4.8, proving that it is effective in both performance prediction and in pedagogical transparency. Findings show that AI-based visual analytics would improve engagement, metacognitive awareness, and curriculum flexibility and would decrease the manual assessment work and increase the responsiveness of instruction. Ethical and explainable design of the system creates trust and accountability and it is appropriate to institutional deployment and to integrate the policies. The paper comes to the conclusion that AI-based visual feedback systems are a paradigm shift in the current educational field, which unites cognitive science, artificial intelligence, and educational governance in the direction of ongoing and learner-driven evolution.