VISAR: A Human-AI Argumentative Writing Assistant with Visual Programming and Rapid Draft Prototyping
Project Overview
The document explores the integration of generative AI in education, particularly through the introduction of VISAR, an AI-enabled tool that aids students in the prewriting and planning phases of argumentative writing. VISAR leverages visual programming and rapid prototyping to offer recommendations for writing goals, synchronize text and visual planning, and spark critical thinking by identifying counterarguments and supporting evidence. A user study indicated that participants found VISAR beneficial for idea generation, outline validation, and enhancing their argumentative writing skills, despite some noted usability challenges and limitations in draft quality. Additionally, the document discusses a broader range of generative AI applications aimed at improving argumentation skills, collaborative writing, and critical thinking. These AI systems provide adaptive support, generate content, and foster interaction among students during writing tasks, ultimately striving to enhance educational outcomes and deepen understanding of complex topics. Overall, the findings emphasize the potential of generative AI tools to transform educational practices by supporting students in developing essential writing and critical reasoning skills.
Key Applications
VISAR: A Human-AI Argumentative Writing Assistant
Context: Used in academic settings, targeting students in writing courses and those engaged in argumentative writing tasks. The tool provides real-time assistance in drafting essays and facilitates the planning and organization of ideas.
Implementation: Implemented as a web application using React and integrating large language models (LLMs) for writing assistance. The system generates prompts, discussion points, and counterarguments based on user input to help improve writing skills.
Outcomes: Facilitated improved argumentative writing skills, critical thinking, ideation, and organization of ideas. Enhanced user control and autonomy in the writing process.
Challenges: Users reported a steep learning curve, difficulties with the interface, and the need for more concrete suggestions. Additionally, it may produce generic content that lacks depth if not guided by user inputs.
ArgueTutor: An adaptive dialog-based learning system for argumentation skills
Context: Used in educational settings to assist university students in developing argumentation skills through interactive dialog.
Implementation: Implemented as a dialog-based system that allows students to interactively engage and improve their argumentation techniques.
Outcomes: Improved student engagement and understanding of argumentation techniques.
Challenges: Requires significant data to train effectively and may struggle with nuanced arguments.
DocuViz: Visualizing collaborative writing
Context: Designed for collaborative writing tasks in classroom settings, specifically aimed at students working on group projects.
Implementation: Utilizes visualization tools to enhance the collaborative writing process, allowing for clearer communication and organization among group members.
Outcomes: Enhanced collaboration and clarity in written work among students.
Challenges: Technical issues can arise with visualization tools, potentially hindering user experience.
Implementation Barriers
Usability
Participants faced challenges in using the interface, such as difficulties connecting nodes and navigating the visual editor. The presence of many features increased cognitive load, making it challenging for users to maintain clarity in their writing process.
Proposed Solutions: Suggestions for improvement include more intuitive controls, enhancements to the interface for better usability, and simplifying the interface to reduce cognitive strain through user-centered design.
Technical Barrier
Challenges in integrating AI tools with existing educational platforms and systems.
Proposed Solutions: Development of standard APIs and better collaboration between developers and educational institutions.
User Acceptance Barrier
Resistance from educators and students to adopt AI tools due to skepticism about their effectiveness.
Proposed Solutions: Conducting workshops and demonstrations to showcase the benefits and ease of use of AI tools.
Project Team
Zheng Zhang
Researcher
Jie Gao
Researcher
Ranjodh Singh Dhaliwal
Researcher
Toby Jia-Jun Li
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Zheng Zhang, Jie Gao, Ranjodh Singh Dhaliwal, Toby Jia-Jun Li
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