Beyond Answers: Large Language Model-Powered Tutoring System in Physics Education for Deep Learning and Precise Understanding
learningai-educationinstructional-design
Abstract
This paper proposes Physics-STAR, a framework for large language model (LLM)-powered tutoring system designed to provide personalized and adaptive learning experiences for high school students. Results showed that Physics-STAR increased students' average scores and efficiency on conceptual, computational, and informational questions. Students' average scores on complex information problems increased by 100% and their efficiency increased by 5.95%.
Summary
Relevance to Cognitive Load Theory
Physics-STAR exemplifies how AI tutoring can manage cognitive load effectively:
Reducing Extraneous Load
- The system provides step-by-step guidance rather than overwhelming students with complete solutions
- Reflective learning prompts help students process information more deeply
Managing Intrinsic Load
- Personalized difficulty adjustment matches content complexity to student ability
- The 100% improvement on complex information problems suggests successful load management
Implications for Instructional Design
The framework demonstrates several CLT principles:
- Worked Example Effect: Step-by-step guidance reduces cognitive load for novices
- Guidance Fading: Adaptive support can decrease as expertise develops
- Variability Effect: Personalization allows for appropriate variation in problem types
Key Findings
- Students showed significant improvement on both conceptual understanding and computational tasks
- Efficiency gains suggest reduced cognitive overhead in the learning process
- The approach helps develop critical thinking skills alongside content knowledge