Skip to main content Skip to navigation

Distinguishing AI-Generated and Human-Written Text Through Psycholinguistic Analysis

Project Overview

The document explores the rising integration of generative AI in education, emphasizing the importance of tools to differentiate between AI-generated and human-written texts to uphold academic integrity. It proposes a comprehensive framework that merges stylometric analysis with psycholinguistic theories, aiming to accurately identify text authorship by analyzing 31 stylometric features associated with cognitive processes. This framework reveals unique patterns in human writing, particularly concerning cognitive load, metacognition, and discourse planning, which contrast significantly with AI outputs. The findings underscore the necessity for educators to adapt to the challenges posed by AI in writing, ensuring that assessments and academic standards are maintained while leveraging the benefits of generative AI as a supplementary educational tool. Overall, the document advocates for the development of robust methodologies to navigate the complexities introduced by AI in educational contexts, fostering an environment where integrity and innovation can coexist.

Key Applications

StyloAI

Context: Educational settings, specifically in the context of assessing written assignments.

Implementation: Integrates stylometric analysis with psycholinguistic theories to develop tools for authorship verification.

Outcomes: Enhanced ability to distinguish between AI-generated and human-written texts, preserving academic integrity.

Challenges: AI-generated texts can closely mimic human writing, making detection challenging; also, the 'black box' nature of some AI models limits transparency.

Implementation Barriers

Technical barrier

The complexity of distinguishing AI-generated content from human writing due to the sophistication of AI models.

Proposed Solutions: Development of interpretable models that combine stylometric features with psycholinguistic theories.

Ethical barrier

Concerns regarding academic integrity and the implications of AI in educational assessments.

Proposed Solutions: Establishing clear guidelines and detection tools to maintain the authenticity of student submissions.

Project Team

Chidimma Opara

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Chidimma Opara

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