CoachGPT: A Scaffolding-based Academic Writing Assistant
Project Overview
The document introduces CoachGPT, an innovative AI-powered academic writing assistant tailored to support students, especially those with limited resources or who are second language learners, throughout the writing process. By employing a scaffolding approach, CoachGPT guides users from brainstorming ideas to performing grammar checks, offering real-time feedback that enhances learning outcomes. This tool aims to provide a more personalized and immersive experience compared to conventional writing assistants, effectively addressing challenges such as plagiarism risks and the lack of instructional support often found in traditional large language model applications. Overall, CoachGPT exemplifies the potential of generative AI in education by empowering students to improve their writing skills in a supportive environment.
Key Applications
CoachGPT: A Scaffolding-based Academic Writing Assistant
Context: Academic writing support for students, particularly those with limited educational resources and non-native English speakers.
Implementation: Web-based application that guides users through a structured writing process with feedback and suggestions.
Outcomes: Improved writing skills through personalized feedback and a structured approach to writing assistance.
Challenges: Potential for misuse leading to plagiarism; ensuring quality of feedback and assistance; accessibility for all users.
Implementation Barriers
Ethical
Risk of students using AI to generate essays without learning, leading to plagiarism and diminished motivation.
Proposed Solutions: Incorporating teaching elements in the AI's feedback to encourage learning and responsible use.
Technical
Large language models can forget context, have limitations in understanding domain-specific language, and may require improved user engagement.
Proposed Solutions: Implementing a scaffolding structure and user feedback mechanisms to enhance understanding and engagement.
Project Team
Fumian Chen
Researcher
Sotheara Veng
Researcher
Joshua Wilson
Researcher
Xiaoming Li
Researcher
Hui Fang
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Fumian Chen, Sotheara Veng, Joshua Wilson, Xiaoming Li, Hui Fang
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