Former Brandon University instructor alleges more than half his students used AI to cheat - Action News
Home WebMail Sunday, December 29, 2024, 09:15 AM | Calgary | -8.9°C | Regions Advertise Login | Our platform is in maintenance mode. Some URLs may not be available. |
Manitoba

Former Brandon University instructor alleges more than half his students used AI to cheat

A former sessional instructor at Brandon University alleges students used artificial intelligence to complete assignments, and is criticizing the university's response to those allegations.

Provost defends southwestern Manitoba university's 'robust process' to deal with cheating allegations

A man with a beard stands with his arms crossed.
Former Brandon University sessional instructor Matt MacDonald alleges the use of artificial intelligence to complete assignments was rampant in his class. (Chelsea Kemp/CBC)

A former Brandon University instructor says the school is failing to address the use of artificial intelligence to cheat something he believes at least half his students did on assignments.

Matt MacDonald taught an introduction to English literature course as a sessional instructor from January to mid-April at the southwestern Manitoba university. By March, he had suspicions students were using AI to complete assignments and began an investigation. He also notified administration, including the dean of arts.

MacDonald says henoticed assignments that "sounded very, very similar, that used similar words and phrases," and some that "came in with really bizarre errors and mistakes, oversights."

Some students turned in assignments on literature that wasn't discussed in class, or referenced sources that did not exist, he said.

MacDonald developed a systemto check for cheating by looking for a list of words and phrases, combined with language and sentence structures that repeatedly appeared in what he believed to be AI-generated assignments.

"It turned into ... a cipher" as he went through assignments looking for evidence of AI use, MacDonald said.

A building and teepee at Brandon University.
A statement on Brandon University's website defended its response to the cheating allegations, saying it has a 'robust process' to determine the validity of such claims. (Chelsea Kemp/CBC)

At least 15 of his 71 students cheated and all but two admitted to it, he said, but based on his analysis, he thinks there could be another 20 to 25 who cheated but were not caught.

MacDonald could not conclude how many cheated because he stepped away from the university's investigationthree weeks after his contract ended, asking to be let go from itbecause "they kept asking me to conclude [it] as quickly and hastily as possible," he said.

University defends response to allegations

Ultimately, some students were penalized following the investigation, while others were cleared, he said, but hefeelsthe dean essentially ignored much of the proof of cheating he presented to the university.

"Most of the students ... [who] admitted to using AI to compose their assignments when I met with them have been all but cleared," he said.

MacDonaldalso alleges international and male students received harsher penalties, likezero grades onassignments, and more scrutiny for cheating than domestic and female students.

As an example, he said onemale student admitted to using AI with the understanding there would be leniency, only to receive a grade of zerofrom the dean and having to drop out of the program.

MacDonald estimates that when he recused himself from the investigation, there were about200 assignments from his studentsthat he felt stillneeded to be graded and tested for AI usage.

In an emailed statement, a Brandon University spokesperson told CBC News the school will not comment on MacDonald's allegations, but said in general, each instructor can choose what level of AI use is acceptable in a course.

A statement on the university's websitedefended its response to what it termed "allegationsof departures from academic integrity," saying it has a"robust process" to determine the validityof such allegations.

Provost and vice-president Kofi Campbell said in the statement that in this case, "the dean personally met with each student before rendering their decision," and that if either a student or instructor disagrees with a decision, the process allows for appeals before theallegations are turned over to a 10-member senate committee, which makes a final decision.

"Just because one member of the process disagrees with the final decision, does not in any way mean that the process is flawed," Campbell said.

He said he has received a legal opinion "which affirms our position in regards both to the press [coverage], and the instructor's public allegations," and the university is "currently considering our legal options."

AI use 'difficult to detect'

AI can be a powerful tool to help polish language skills for students who speak English as an additional language,said David Gerhard, the head of computer science at the University of Manitoba.

But there can be a temptation to cross over into plagiarism or cheating when AI-generated text is presented as one's own work.

Since everybody has access to the technology, students need to talk with instructors to see how to appropriately use it, Gerhard said.

A man in a beige suit jacket with blue shirt is standing in his office with books and cabinets behind him.
David Gerhard, the head of computer science at the University of Manitoba, says there's no foolproof way to determine if artificial intelligence has been used in an assignment. (Jeff Stapleton/CBC)

But there isn't a good model yet for post-secondary schools to use to detect AI usage, and generally speaking, there's no way to look at writing and say definitively it was written with AI, Gerhard said.

"It is difficult to detect. So you can't just throw it through a machine and say, 'These ones are AI, these ones aren't,'" Gerhard said.

There also can't be a one-size-fits-all rule for using AI, since it can serve many purposes, he said.

Gerhard fears that schools may create knee-jerk rules that make it more difficult for people to use AI in creative ways.

"The biggest problem ... is that there's a lot of people, students and teachers, who are using it without really thinking about the consequences," Gerhard said.

"We're not going to prohibit it, but we're going to say, 'Be very careful about how you use it.'"