Automating question generation from educational text
Project Overview
The document explores the application of generative AI in education, highlighting an automated question generation tool known as QGen, designed to ease teachers' workloads by producing multiple-choice questions (MCQs) from educational texts through transformer-based language models. The study indicates a strong interest among educators for AI-assisted tools that facilitate the creation of question-based activities (QBAs), which are crucial for both formative and summative assessments. The findings reveal that the proposed framework effectively generates high-quality questions, answers, and distractors, thereby improving the educational experience while maintaining the indispensable role of teachers. Overall, the use of generative AI in educational contexts not only supports teachers in developing assessment materials but also enriches student learning outcomes by providing tailored and relevant assessments.
Key Applications
QGen - Automated question generation tool for formative and summative assessments
Context: Schools, targeting teachers who prepare question-based activities (QBAs) for assessments in grades 3-8.
Implementation: Developed a modular framework using transformer-based models like T5 and GPT-3 for generating MCQs from textual content.
Outcomes: Reported high-quality generated questions with good grammar and well-formedness; effective in reducing teachers' workload and enhancing personalized learning.
Challenges: Concerns about latency, privacy, and reliability of the AI-generated content; variability in question quality depending on the model used.
Implementation Barriers
Resource Management
Teachers reported not having enough time to prepare questions for class and faced a lack of access to quality question-preparing resources.
Proposed Solutions: AI tools like QGen can automate question generation and implement AI-driven tools that generate high-quality questions from existing educational content, potentially alleviating the time burden on teachers.
Personalization and Adaptivity
There is a reported lack of personalized and adaptive question-asking resources tailored for each class.
Proposed Solutions: AI tools can be designed to better integrate with lesson content and existing digital tools to enhance adaptability.
Project Team
Ayan Kumar Bhowmick
Researcher
Ashish Jagmohan
Researcher
Aditya Vempaty
Researcher
Prasenjit Dey
Researcher
Leigh Hall
Researcher
Jeremy Hartman
Researcher
Ravi Kokku
Researcher
Hema Maheshwari
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Ayan Kumar Bhowmick, Ashish Jagmohan, Aditya Vempaty, Prasenjit Dey, Leigh Hall, Jeremy Hartman, Ravi Kokku, Hema Maheshwari
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