Research Summary: Changes in Help-Seeking Strategies Predict unaided Performance during AI-based Mathematical Learning

Research involving Grade-9 students using Generative AI (GenAI) for mathematical modeling practice indicates that changes in help-seeking strategies during AI interaction can predict subsequent unaided performance. The study investigated epistemic proactivity, focusing on self-regulated learning functions and help-seeking content, to understand how learners regulate GenAI tools in open-ended learning environments aimed at improving independent performance.

Why it matters

This research provides insights into the effectiveness and regulation of AI in educational settings, which is critical for future strategic planning in workforce development and educational technology integration. Understanding how learners interact with and benefit from AI tools can inform the design of more effective learning platforms and pedagogical approaches, ensuring that AI complements rather than replaces essential learning processes.

Key insights

• Generative AI systems are increasingly being utilized by students as learning companions.
• There is limited understanding of how students regulate GenAI tools in open-ended learning contexts.
• The study focused on Grade-9 students' use of a general-purpose GenAI for mathematical-modeling practice.
• Epistemic proactivity (learners retaining responsibility for knowledge-building) was a key lens, operationalized by self-regulated learning functions and help-seeking content.
• Changes in help-seeking strategies predict students' unaided performance after using AI for learning.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2606.28472