Generative AI for Enhancing Active Learning in Education: A Comparative Study of GPT-3.5 and GPT-4 in Crafting Customized Test Questions
Project Overview
The document explores the application of Large Language Models (LLMs), specifically GPT-3.5 and GPT-4, in the realm of education, particularly in generating personalized test questions for Grade 9 mathematics while adhering to active learning principles. By employing an iterative method, the study demonstrates how these models can produce tailored questions that vary in difficulty and content, leading to improved student engagement and learning outcomes. Results indicate that GPT-4 excels in crafting challenging questions, while GPT-3.5 shows enhanced problem-solving capabilities after receiving guidance from GPT-4. This research highlights the significant potential of generative AI to personalize educational experiences and emphasizes the need for further investigation into its applications across diverse educational settings.
Key Applications
Customized test question generation using LLMs (GPT-3.5 and GPT-4)
Context: Grade 9 mathematics education focusing on active learning principles
Implementation: Iterative question-answering process where GPT-4 generates questions and GPT-3.5 responds, simulating a teacher-student dynamic.
Outcomes: Improvement in the ability to generate accurate and challenging questions; enhanced engagement and learning outcomes for students.
Challenges: Limitations in the ability of GPT-3.5 to utilize explanatory content effectively; need for more diverse educational contexts and subjects.
Implementation Barriers
Technical
GPT-3.5 showed limited improvement when provided with explanatory content, indicating potential issues in processing and utilizing such information.
Proposed Solutions: Further research into refining LLMs to better integrate explanatory content into their learning processes.
Scope
The study's focus was primarily on Grade 9 mathematics, limiting the generalizability of findings across other subjects and educational levels.
Proposed Solutions: Future research should expand the scope to include a wider array of subjects and educational contexts.
Project Team
Hamdireza Rouzegar
Researcher
Masoud Makrehchi
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Hamdireza Rouzegar, Masoud Makrehchi
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai