Generating Reflection Prompts in Self-Directed Learning Activities with Generative AI
Dishita G Turakhia, Zane Mroue, Peiling Jiang, Stefanie Mueller · 2024
Self-reflection during maker activities is known to enhance conceptual comprehension and lead to better skill learning. While educational makerspaces commonly leverage reflective exercises guided by instructors, this practice often goes amiss in scenarios when makers interested in self-directed learning use online tutorials like Instructables. In this short paper, we explore the approach of using Large Language Models (LLMs), specifically OpenAI's GPT-4 for generating reflection prompts with the existing Instructable tutorials and aligning them with a list of learning goals adapted from prior work on maker skills learning. We built a system to generate Reflectables: Instructables designed for self-reflection while making. To exemplify the approach, we generated 9 Reflectables consisting of 128 prompts and evaluated them on seven attributes: goal-oriented, timely, contextual, sequential, multilevel, clear, and personalized. Our analysis highlights the benefits and limitations of this approach and points to further research directions for designing AI-based systems for reflection-focused self-directed learning in makerspaces.