Effect of AI-generated explanations on learning performance and transfer

Determine whether AI-generated explanations produced by the configured custom GPT for proportional relationships improve sixth-grade students’ learning performance in mathematics and physics, particularly their ability to transfer mathematical knowledge to the physical context, compared to textbook-only materials.

Background

The paper’s RQ5 asks about the effect of AI-generated explanations on learning performance and transfer from mathematics to physics. Results showed no significant differences in overall test scores, with mixed outcomes across subtasks.

Given these inconclusive findings, the authors state they cannot provide a definitive answer about the performance and transfer effects of the custom GPT-generated explanations under the paper conditions.