Let teachers set the limits for AI help
How much help should AI give a learner? Haq’s framework starts with four fields approved by educators, then sets the help allowed for each activity. Its examples show how the rules might work. They do not test whether students learn more.
What the source reports
Ijaz Ul Haq proposes a way to set the limits of AI help for a learning activity. Educators approve four fields about the work, including its goal and what the learner must own. The framework then uses rules to decide which kinds of help are allowed and which are not. The paper supplies worked examples to make the design concrete. It is not a study of student learning in K-12 classrooms. It does not show that teachers followed the rules in a live lesson or that students learned more. The AI Agency digest recommends trying the design with a clearly named learner decision and a prohibited full solution. That is a proposed check, not an outcome reported by the paper.
Original source
- Title
- A pedagogically guided LLM framework for scaffolding human-AI cocreation and preserving learners' agency and creativity
- Author
- Ijaz Ul Haq
- Publication
- arXiv cs.CY
- Date
- Tuesday, October 27, 2026