KWIZIE: Scaling Engagement in Video-Based Learning with AI

This demonstration paper presents Kwizie, a web-based platform designed to enhance engagement with long-form educational video content (e.g. raw lecture recordings) using generative AI. Kwizie automatically transcribes videos, identifies key concepts, and segments them into pedagogically coherent chapters. After watching each segment, learners engage with brief quizzes integrated directly into the video frame based on the identified concepts and the contextual content of the corresponding video segment. By embedding these quick assessments within the viewing experience, the learner interaction maintains cognitive continuity and reinforces understanding of the key concepts at the end of each chapter. The system integrates OpenAI's Whisper for transcription, GPT-4o for concept extraction, and a custom text-tiling algorithm for topic segmentation. Kwizie supports scalable deployment across educational settings: it can be used as a standalone tool, embedded into LMS platforms, or exported to common formats. This paper highlights Kwizie's design and pedagogical affordances, demonstrating its potential to transform passive video watching into structured, interactive learning experiences at scale.