On 6 October 2023, Marley Stevens, a student at the University of North Georgia, submitted an essay to Kevin Ellison, a lecturer who taught a criminal justice course she was taking.1 Thirteen days later, on 19 October 2023, Mr. Ellison sent Miss Stevens an e-mail message telling her that two plagiarism detection programs had found evidence that her paper had been written with the aid of artificial intelligence.
“When your paper was uploaded on 10/06/23,” Ellison wrote, “it was checked through Turn It In. The program returned a positive response for AI. I also checked your paper through a third-party app utilized by the Criminal Justice department for verification. This app confirmed the Turn It In AI response.”
“Using AI is cheating,” he added, “and not your work. Therefore, you will receive a grade of Zero for your paper. Any further violations will be sent to the Student Academic Integrity Committee.”2
When Stevens asked Ellison for details, he responded with a second e-mail that explained “According to Turn it in, the entire paper except the last couple of sentences in the conclusion” had been written by an artificial intelligence program.3 In the same message, Ellison also explained that he had “confirmed this through another app used by the CJ department.”4
Soon after receiving this second message, Stevens realized that the “AI” in question had been the free version of Grammarly, a relatively simple application designed to propose corrections for mistakes in grammar, spelling, and punctuation that it catches. (The informal experiments I conducted for the sake of this post lead me to endorse the common characterization of the program as a “glorified spellchecker.”)
Stevens pointed this out to her teacher, adding that the “recommended resources” section on the official website of her university provided a link to Grammarly. Ellison, however, stood his ground. Likewise, the university administrators to whom Stevens appealed declined to acknowledge the absurdity of equating the use of a school-endorsed proofreading program with plagiarism.
So began the academic affair that I have come to call Grammarlygate.5
When first I encountered this collegiate controversy (merci mille fois, Mrs. Muros), it struck me as a simple case of professorial petpeevery, albeit one made worse by the Kafkaesque intransigence of academic apparatchiki. However, upon reflection, and a little additional research, I found myself struck by the long tail of this sad story. That is, it seems that mindless confidence in soulless algorithms owes much to the way that many college courses have been taught for close to a century.
This approach, which I like to call “textbook-centric instruction,” asks students to ransack overpriced primers for factoids, phrases, and terms of art, attend lectures to hear the same materials spoken out loud, and prove their mastery (however temporary it may be) of these blocks of text taking tests in which they fish them out of a sea of distractors.
Reviews posted on Rate My Professor make it clear that Ellison practiced a form of textbook-centrism. In the class in question, he made extensive use of multiple-choice tests that, in the view of most students, dealt entirely with “vocab in the textbook.”6 (One student recalled a few test questions that seemed to have been based upon material presented in lectures.)
In the classroom, Ellison peppered his lectures with tales of his time as a police officer, videos of blood-soaked crime scenes, and seemingly random digressions. Nonetheless, when not sharing anecdotes, he seems to have remained fully within the textbook-centric fold. “He does sometimes get sidetracked on his lectures,” wrote one reviewer, “but everything in the power point is on the book along with answers to the quiz questions also found in the book.”
Mr. Ellison declined to answer my requests for information. Thus, I do not know whether he composed the questions he put on his test or pulled them out of a test-making program of the sort that publishers make available to teachers who use their textbooks. Likewise, I have not been able to find out whether he made his own PowerPoint slides or used publisher-provided presentations.7
The textbook-centric approach to schooling has, for a very long time, turned teachers into agents of alien authorities. In the last two decades or so, the proliferation of “instructor resources,” such as prefabricated slide decks and question banks, has exacerbated this trend, thereby reducing, to an even greater degree, the need for instructors to possess deep expertise in the subjects they purport to teach. The use of artificial intelligence to detect plagiarism, evaluate papers, or compose multiple-choice questions pushes professors further in this direction. It is, however, little more than the continuation of a sad and sorry journey that began a long time ago.
Disclaimer: I used the free version of Grammarly to prepare this post for publication.
For Further Reading:
To Share, Subscribe, or Support:
While listed as “Robert Ellison” in the faculty and staff directory of the Criminal Justice Department at the University of North Georgia, Mr. Ellison signed his e-mail messages to Miss Stevens with the name “Kevin Ellison.”
Screenshot of an e-mail message from Kevin Ellison to Marley Stevens, dated 19 October 2024, reproduced in Jeanette Settembre “College student put on academic probation for using Grammarly: ‘AI violation’” New York Post 21 February 2024
I have not corrected any of the mistakes of spelling, grammar, or style that appear in the many quotations used in this post.
Screenshot of part of an e-mail message from Kevin Ellison to Marley Stevens, reproduced in Jeanette Settembre “College student put on academic probation for using Grammarly: ‘AI violation’” New York Post 21 February 2024
Queries in sundry search engines tell me that no one put this expression on the internet before I typed it at the top of that page. The coinage, however, is so obvious that I find it hard to believe that I am the first to come up with it.
Ellison characterized shorter multiple-choice tests as “quizzes” and longer ones as “exams.” However, the only difference between the two types of tests seems to have been the number of questions asked.
The use of an AI program to generate multiple-choice questions would have added to the irony that abounds in this story. However, the class in question seems to have been up and running in the fall semester of 2022, well before the debut of ChatGPT and other programs of its ilk.
The issue you see within the system from the lowest level to the highest is a comprehensive refusal to teach basic skills combined with a comprehensive refusal to test for the presence of those basic skills. Instead, professors demand that students sit for exams and produce papers that the students are not qualified for. Then, everyone gets an inflated grade, and no one learns much. The subject matter of the courses is knowledge already present in books that students must show that they have temporarily retained.
This also what causes many students stress: they know that they are unskilled, they know that they are "imposters," and their emotional reaction is a rational consequence of these realizations. This receives massive subsidies, so alternative models struggle to compete.
Honestly, if you are going to suggest Grammarly as a program, you should stand by your use of that program.
I use the program when I write, and I'm constantly having to dismiss suggestions that it gives me. For some reason, it wants to change names, phrasing, and almost everything I write. I guess it doesn't like my style.
Just as an example, in the above statement, it tried to get me to change, the phrasing to If you suggest Grammarly as a program, you should stand by it.