EduChat: A Fine-Tuned AI Chat Assistant for Personalized Learning Support

View at publisher Summarise with VIRAT
Authors: Keerthiya T, Jenisha Delphin Mary J, Allin Geo
Cited by: 0
References: 2

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The study details EduChat, a Transformer-style chatbot specifically designed for academics that is user-friendly and fast to use, is unique in that it teaches first-year education students various concepts associated with programming, data science, and Artificial Intelligence (AI). In order for students to receive curriculum-aligned answers, short quizzes to assess their knowledge, and concise explanations about the curriculum’s introductory topics, EduChat utilizes the DistilGPT-2 model, a small model that makes it easy to train each unique topic through the fine-tuning of this chatbot using a curated dataset of 1,200 question-answer pairs and associated lecture note summaries and worked examples for each topic. The comparative evaluations of our fine-tuned models included ablations of the effects of curriculum alignment, window length, and data augmentations on the performance of the models. Overall, EduChat demonstrated a BLEU and ROUGE-L score of 0.72 and 0.76, respectively, and average results of usefulness of 4.42/5 from our human reviews. The unique contribution of this research is that it provides a reproducible resource-efficient workflow for creating a chat bot capable of achieving demonstrable pedagogical advantages through the process of fine-tuning chatbots using subject matterspecific training data that create a user-friendly and fast-to-use chat bot for students.

Fields of Study & Subjects

Intelligent Tutoring Systems and Adaptive Learning Artificial Intelligence in Education Artificial Intelligence in Healthcare and Education

Publication details

Journal / source
International Conference on Computing, Communication, Control and Cyber-Physical Systems (I5CPS)
Pages
1–6
Publication date
26 February 2026
Language
EN