Hou, Xiao, Liu & Mueller (2026) — Lawrence Technological University / Michigan Technological University. 📄 Full text (arXiv)
Investigates how instructional guidance shapes student-AI interaction in construction engineering education. Introduces a five-step prompting framework grounded in Generative Learning Theory (GLT) to guide learner interaction during review activities. Three conditions tested in a controlled experiment: slide-based learning, unprompted AI-supported learning, and prompted AI-supported learning.
Key findings: Performance differences concentrated on tasks requiring explanation and reasoning (higher-order cognitive outcomes). The prompted condition achieved significantly higher open-ended scores (≈2–3 points on 18-point scale, p<0.01). No significant differences in multiple-choice recall across conditions. Unprompted AI use did not outperform traditional slide-based review.
Demonstrates that the effectiveness of AI-supported learning depends critically on how interaction is structured — a simple prompting framework grounded in learning science can significantly improve higher-order cognitive outcomes. Provides a basis for integrating learning science principles into GenAI systems for education.
Related Pages
- scaffolding — Structured guidance and prompting for AI-assisted learning
- active-learning — Active learning engagement with AI systems
- personalized-learning — Personalized learning through structured AI interaction
- feedback-loop — Feedback loops between AI and learners
- higher-ed — Higher education AI integration
- ai-learning-transfer — Cognitive transfer facilitated by structured AI guidance