Transforming university lecture content into an enriched course using generative AI on AWS

This article describes how AWS collaborated with University of Technology Sydney to transform passive lecture recordings into engaging, structured learning modules using generative AI.

• Automatically segments 60–120 minute recordings into 15–30 minute topic-focused chunks, further divided into 30-second to 3-minute chapters
• Uses Amazon Transcribe for speech-to-text conversion and Amazon Bedrock with Claude for intelligent content generation
• Generates summaries, key points, explanatory content, and multiple-choice knowledge checks for each segment
• Employs three-stage pipeline: transcript preprocessing, TF-IDF timestamp recovery, and AI-driven topic grouping for reliable video segmentation
• Uses Amazon Bedrock structured outputs (constrained decoding) to enforce schema-compliant JSON output across thousands of invocations
• Implements human-in-the-loop workflow where educators review and approve all AI-generated content before publication
• Includes course-aware chat assistant powered by Amazon Bedrock Knowledge Bases for retrieval-augmented generation

The solution addresses declining student engagement with long-form content by delivering personalized, adaptive learning experiences at scale without requiring manual content creation by educators.