AIDetection: A Generative AI Detection Tool for Educators Using Syntactic Matching of Common ASCII Characters As Potential 'AI Traces' Within Users' Internet Browser
Project Overview
The document explores the integration of generative AI in education, focusing on a web application named AIDetection, which assists educators in identifying AI-generated content in student essays by analyzing syntactic traces unique to generative models. Unlike traditional AI detection tools that rely on machine learning, which often exhibit inconsistencies, AIDetection employs a heuristic approach that enables bulk analysis of documents while maintaining privacy standards. The application addresses significant challenges in the educational landscape, particularly regarding academic integrity and the enforcement of policies related to AI use. Overall, the findings underscore the necessity for reliable tools to navigate the complexities of AI in educational settings, reflecting a broader concern for maintaining integrity while leveraging technological advancements.
Key Applications
AIDetection.info
Context: Higher education, specifically for college students in writing courses
Implementation: Educators use the tool to scan student essays for AI-generated content and check for acknowledgments of AI tool usage.
Outcomes: Facilitates monitoring of AI policy compliance, reduces grading workload, and provides clear reports on AI usage in student submissions.
Challenges: Does not provide definitive proof of AI content, risk of false positives, reliance on specific character encodings.
Implementation Barriers
Technical Barrier
Detection tools may yield false positives or cannot definitively identify AI-generated content.
Proposed Solutions: AIDetection uses heuristic-based methods to provide potential indications rather than definitive proofs of AI use.
Policy Barrier
Students often misunderstand or ignore acknowledgment requirements for AI usage.
Proposed Solutions: Educational interventions and clear communication of AI policies can help improve acknowledgment compliance.
Privacy Barrier
Concerns about data privacy and compliance with regulations like GDPR and FERPA.
Proposed Solutions: AIDetection processes data client-side, ensuring that student data is not uploaded to external servers.
Project Team
Andy Buschmann
Researcher
Contact Information
For more information about this project or to discuss potential collaboration opportunities, please contact:
Andy Buschmann
Source Publication: View Original PaperLink opens in a new window