Skip to main content Skip to navigation

Modifying AI, Enhancing Essays: How Active Engagement with Generative AI Boosts Writing Quality

Project Overview

The document examines the role of Generative AI (GAI) in enhancing writing quality within educational contexts, focusing on its potential to support students in generating ideas, creating content, and revising their work. While GAI shows promise in improving writing outcomes, it also presents challenges for educators, particularly in evaluating the authenticity of student writing and the cognitive processes involved, as GAI can introduce linguistic biases. The study explores the correlation between GAI-assisted writing behaviors and the quality of essays produced, underscoring the necessity for students to engage actively with GAI tools to maximize their benefits. Ultimately, while GAI can facilitate the writing process, careful consideration is required to address the implications for assessment and the integrity of student learning.

Key Applications

GAI-assisted writing using tools like GPT-3

Context: Higher education, targeting university students engaged in writing tasks

Implementation: Students utilize GAI to seek writing suggestions, modify texts, and improve their essays based on behavioral patterns observed during writing sessions.

Outcomes: Active engagement with GAI leads to improved essay quality, including lexical sophistication, syntactic complexity, and text cohesion. Reduced linguistic bias when incorporating GAI-generated content.

Challenges: Teachers struggle to assess genuine cognitive engagement and the quality of student learning due to the integration of GAI in writing tasks.

Implementation Barriers

Assessment Challenges

Traditional assessment methods may not adequately evaluate student writing quality in GAI-assisted contexts, as they do not account for the collaborative nature of GAI involvement.

Proposed Solutions: Develop new assessment methods that focus on students' engagement with GAI during the writing process, such as analyzing writing logs.

Linguistic Bias

Generative AI models may perpetuate existing biases in language, which can be reflected in students' writing, especially for non-native speakers.

Proposed Solutions: Educators can provide training to help students recognize and mitigate biases in both GAI outputs and their own writing.

Project Team

Kaixun Yang

Researcher

Mladen Raković

Researcher

Zhiping Liang

Researcher

Lixiang Yan

Researcher

Zijie Zeng

Researcher

Yizhou Fan

Researcher

Dragan Gašević

Researcher

Guanliang Chen

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Kaixun Yang, Mladen Raković, Zhiping Liang, Lixiang Yan, Zijie Zeng, Yizhou Fan, Dragan Gašević, Guanliang Chen

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

Let us know you agree to cookies