ERIC Number: EJ1482073
Record Type: Journal
Publication Date: 2025
Pages: 9
Abstractor: As Provided
ISBN: N/A
ISSN: N/A
EISSN: EISSN-1545-679X
Available Date: 0000-00-00
Utilizing GPTZero to Detect AI-Generated Writing
Karen Paullet; Jamie Pinchot; Evan Kinney; Tyler Stewart
Information Systems Education Journal, v23 n6 p44-52 2025
Generative AI tools such as ChatGPT are now in widespread use and are often utilized by students to help in creating writing assignments intended to be written entirely by the student. This has spurred the need for AI detection tools such as GPTZero. This study sought to determine the accuracy of GPTZero's AI detection in identifying whether writing was created by a human, generated by AI, or a mix of both. Because many students now submit work that is a mix of both their original writing and AI-generated text, it has become more important to be able to accurately identify mixed-generated writing. The study analyzed 500 writing samples of human, AI, and mixed origin and utilized GPTZero's Deep Analysis to identify writing origin sentence-by-sentence in the mixed samples. Results from this study indicated that GPTZero accurately identified the writing origin of all samples, within an 89% to 93% accuracy rate of mixed-generated writing, and a 95-99% accuracy of writing that was written by a human or entirely by AI.
Descriptors: Artificial Intelligence, Writing Assignments, Deception, Program Effectiveness, Plagiarism, Accuracy
Information Systems and Computing Academic Professionals. Box 488, Wrightsville Beach, NC 28480. e-mail: publisher@isedj.org; Web site: http://isedj.org
Publication Type: Journal Articles; Reports - Research
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A
Grant or Contract Numbers: N/A
Author Affiliations: N/A