Efficacy of Detecting AI Plagiarism in Higher Education

Abstract

Whether the text has been directly copied from a source and presented as the author’s work, or if the author poorly paraphrased their source, the result is plagiarism.  Plagiarism stems from a desire to get a good grade. It is the result of being sloppy, not having confidence in one’s ability to paraphrase correctly, a lack of understanding of what needs to be cited, thinking they are supposed to reproduce what the experts have said rather than synthesize the expert's opinions, they panic as the deadline approaches and of course they are lazy.    Plagiarism is not new to academia and detecting it has been made easier with plagiarism-detecting tools and simply conducting effective searches on the Internet.  However, new plagiarism methods are being developed as technology grows, which are significantly harder to detect.  Students now use artificial intelligence to write their academic papers.  Detecting if AI was used in producing an academic paper is difficult because the paper has been properly paraphrased and/or cited. New tools like GPTZero and ZeroGPT have been developed to address this issue, but these tools are untested, and academic faculty don’t want to pin a plagiarism charge on a student based on an untested tool.  This study looks to address this issue and determine if either or both tools effectively determine whether AI has written something.