Wordcraft: a Human-AI Collaborative Editor for Story Writing
Project Overview
The document discusses the integration of generative AI in education, focusing on Wordcraft, a tool that aids creative story writing by fostering collaboration between human writers and AI. It highlights various applications of Wordcraft, including planning, writing, editing, and rewriting, which leverage few-shot learning and dialog systems to enhance user interaction and improve overall writing quality. The findings indicate that while AI can significantly assist in creative processes, it also presents certain limitations. The outcomes suggest that when used effectively, generative AI tools like Wordcraft can complement traditional writing methods, enabling students to enhance their creativity and writing skills through a supportive and interactive environment. The document underscores the potential of AI to transform educational practices by providing personalized learning experiences that cater to individual needs in creative writing.
Key Applications
Wordcraft: a Human-AI Collaborative Editor for Story Writing
Context: Creative writing tool aimed at writers of all levels pursuing story creation.
Implementation: Developed a web-based text editor integrated with a dialog system for interactive writing support.
Outcomes: Facilitated diverse writing tasks, inspired creativity, and provided dynamic feedback to writers.
Challenges: Inconsistent output quality, sensitivity to prompt phrasing, and documented biases in language models.
Implementation Barriers
Technical barrier
Inconsistent output quality from the generative models, ranging from superb to nonsensical.
Proposed Solutions: Conducting formal user studies to understand writer needs, exploring fine-tuning methods, and addressing biases.
Ethical/Compliance barrier
Documented issues with bias and memorization present in language models.
Proposed Solutions: Improving training methodologies before wider deployment.
Project Team
Andy Coenen
Researcher
Luke Davis
Researcher
Daphne Ippolito
Researcher
Emily Reif
Researcher
Ann Yuan
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Andy Coenen, Luke Davis, Daphne Ippolito, Emily Reif, Ann Yuan
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