ReflexAI: Optimizing LLMs for Consistent and Constructive Feedback in Reflective Writing

Anand Bhojan, Tan Li Xin

2025

Abstract

Creative Media courses often require students to iteratively gather peer playtesting feedback, respond to it, and document their reflections. To streamline this process, iReflect, a web application, was developed in our previous work. Research indicates that high-quality reflective writing correlates with improved academic performance. To support this, iReflect leverages Large Language Models (LLMs) to provide automated feedback on students’ reflective writings. However, LLMs face challenges such as inconsistency and inaccuracies in feedback. This research explores methods to enhance the quality of LLM-generated feedback for reflective writing. Findings reveal that repeated queries and in-context learning enhance the consistency and accuracy of feedback scores. Additionally, integrating key elements of constructive feedback into the prompts enhances the overall effectiveness and utility of the feedback.

Paper Citation

in Harvard Style

Bhojan A. and Xin T. (2025). ReflexAI: Optimizing LLMs for Consistent and Constructive Feedback in Reflective Writing. In Proceedings of the 17th International Conference on Computer Supported Education - Volume 2: CSEDU; ISBN 978-989-758-746-7, SciTePress, pages 387-394. DOI: 10.5220/0013430800003932

in Bibtex Style

@conference{csedu25,
author={Anand Bhojan and Tan Xin},
title={ReflexAI: Optimizing LLMs for Consistent and Constructive Feedback in Reflective Writing},
booktitle={Proceedings of the 17th International Conference on Computer Supported Education - Volume 2: CSEDU},
year={2025},
pages={387-394},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013430800003932},
isbn={978-989-758-746-7},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 17th International Conference on Computer Supported Education - Volume 2: CSEDU
TI - ReflexAI: Optimizing LLMs for Consistent and Constructive Feedback in Reflective Writing
SN - 978-989-758-746-7
AU - Bhojan A.
AU - Xin T.
PY - 2025
SP - 387
EP - 394
DO - 10.5220/0013430800003932
PB - SciTePress