Moira Olmsted, a student of online courses at Central Methodist University in the United States, faced this problem. Every week, students are required to submit their reading notes for review. Olmstead says that her summary received 0 points because the AI checking tool indicated that her work was generated by artificial intelligence.
Olmsted challenged the accusations to her teacher and student coordinator, emphasizing that she has autism spectrum disorder and writes in a pattern that could be mistaken for artificial intelligence.
In the end, her work was reviewed and received a satisfactory grade, but the student still received a warning that the next time the software identified her work as plagiarized, there would be no second review.
The best AI spelling detectors are highly accurate, but they are not foolproof. Bloomberg tested two leading services - GPTZero and Copyleaks - on a random sample of 500 college application essays submitted to Texas A&M University in the summer of 2022, shortly before ChatGPT's release, effectively guaranteeing that they were not created by artificial intelligence.
The essays were obtained through a public records request, which means they were not part of the datasets on which the AI tools are trained. The services mistakenly labeled 2% of the essays as likely written by artificial intelligence.
According to students, researchers, and AI developers, neurodiverse students and people who speak English as a second language are at greater risk of being accused of plagiarism.
A 2023 study by researchers at Stanford University found that AI detectors were "nearly perfect" at checking essays written by eighth-grade students born in the United States, but they flagged more than half of the essays written by students whose first language is not English as AI-generated.