MumtahinGPT: A preparatory tool for reducing anxiety and enhancing students' academic performance with retrieval-augmented generation and AI feedback
DOI:
https://doi.org/10.71085/sss.04.02.444Keywords:
Artificial Intelligence, Retrieval-Augmented Generation, Academic Anxiety, Oral Assessment, Academic Confidence, Teacher Education, STEM Education, Formative Assessment, AI FeedbackAbstract
In this study, MumtahinGPT is presented, a Generative AI powered tool, empowering students and researchers with intelligent, personalized external examination practice. The preparatory tool is developed using the Retrieval Augmented Generation (RAG) workflow. The tool is designed with the aim to boost confidence and reduce presentation anxiety. Similarly, it improves academic performance while delivering thesis defenses, synopsis and proposal defenses. To cope this AI integration helps MumtahinGPT to retrieve relevant information from student’s uploaded documents. It augments with external knowledge in order to generate personalized questions to assist in preparation. The tool also offers feedback on user responses so as to help students to refine their answers. The tool is built with Python utilizing Gradio as User Interface (UI) and a gemini-2.5-flash LLM powered by Google API. The functionality of tools includes uploading documents, chunks creation, questions generation, lifeline features feedback and final performance report generation. The results validate the effectiveness of the tool in reducing anxiety and uplifting the student’s confidence. Further, it also assists external examiners and high ups in asking insightful questions during the defenses and presentations. The study concludes with future enhancements and broader applications in educational scenarios.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Ghulam Jillani Ansari, Shahbaz Hassan Wasti , Muhammad Naeem Sarwar

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.



