What happened
In an eLearning Industry article, Muhammad Asif Raza Mir Shah proposes LEARNT, a six-step model for AI-supported learning: locate outcomes, engage independently, ask for guidance, review evidence, note revisions and AI use, and test performance without AI. The article distinguishes tutoring from answer generation.
Why it matters
For educators and L&D designers, that distinction shifts attention from polished submissions to demonstrable understanding. The article cites a 2025 undergraduate physics study in Scientific Reports in which, under the conditions studied, students using a structured AI tutor learned more in less time than students in an in-person active-learning class. It also describes a field experiment where unrestricted AI improved practice performance, but some students performed worse after its removal; a tutor with safeguards reduced that effect. These accounts do not show that any chatbot improves learning. They make scaffolding and independent assessment important design considerations.
What to do next
Specify permitted AI support for each activity, provide coaching prompts, and ask learners to make an initial attempt before seeking feedback. Have them check important claims against original sources, record which AI suggestions they accepted or rejected, and explain or apply their reasoning without AI. For workplace courses, heed the article’s caution about confidential and organizational information. Offer alternatives for learners who cannot or do not wish to use AI, and avoid making essential outcomes depend on a paid tool. Treat the reported findings as context, not a guarantee for another course.
About this briefing
Reviewed by TutorFlow Editorial. We link the primary source, preserve its publication date, and distinguish reported claims from TutorFlow analysis. Our commentary focuses on practical decisions for educators and training teams.
Discussion (0)
Open to everyone