Impact of AI chatbots on students’ learning performance and experience

Determine how learning with AI chatbots (including ChatGPT and custom GPTs) affects students’ learning performance and learning experience in school-based physics and mathematics education, specifying the nature and magnitude of any effects under realistic classroom conditions.

Background

The paper investigates the educational use of AI chatbots, particularly custom GPTs, in mathematics and physics contexts. While prior work suggests potential benefits for motivation and personalization, the authors note that rigorous evidence about the effects on learning performance and experience is limited.

This motivates a controlled study and highlights the need to precisely characterize how AI chatbot-based learning influences students’ outcomes and experiences.

References

However, despite this promise, the question of how exactly learning with AI chatbots affects students' learning performance and experience remains unanswered [5].

— Building Bridges: AI Custom Chatbots as Mediators between Mathematics and Physics  (2412.15747 - Lademann et al., 2024) in Section I. Introduction

The present study captured students' immediate perceptions of Beacon following use, but it does not establish whether sustained engagement leads to measurable improvements in confidence, conceptual understanding, assessment performance, or help-seeking behaviour over time.

— Reducing Barriers to Academic Support: Evaluating a Course-Specific RAG System for Addressing Help-Seeking Disparities in Higher Education  (2609.21600 - Gray et al., 18 Sep 2026) in Section 7, Future Work