Michael Zwaagstra: Students Can’t Use AI to Cheat on Standardized Tests

Commentary

As the school year winds down, many students across Canada will hand in their final assignments and write their final exams. Cutting corners and outright cheating in school is easier than ever. If you need to write an essay, just plug in the assignment instructions and let artificial intelligence (AI) write it for you.

A recent New York Magazine article provided numerous examples of college students using AI to write formal essays, generate programming code, and even draft personalized notes. Whether you need help creating an outline, finding relevant sources or writing an introduction, AI can do all these things and more.

Many K–12 students also use AI for their assignments. Anyone who is worried about being caught just needs to tell ChatGPT (or whichever AI program they use) to make it look like the essay was written by a high school student.

Catching cheaters is nearly impossible—and it’s getting harder as AI gets increasingly sophisticated. Even so-called AI detectors like Turnitin, which scan essays for patterns that indicate the use of AI, are far from perfect. In other words, there’s no easy or low-cost way to prevent students from using AI on their homework assignments.

Obviously, this is a significant problem. If students use AI to do most of their homework, they aren’t going to learn important academic skills. This does not bode well for their future or the general productivity of our labour force.

Fortunately, there’s one academic measurement tool available that AI cannot interfere with—in-person standardized tests, which are administered to all students in a particular grade at the same time and are assessed by outside evaluators using consistent criteria. They can be grade-level tests or exams that are required for graduation.

For example, Grade 12 students in Alberta must write diploma exams in core subjects such as English language arts, mathematics, social studies and science. These exams are created by the provincial Ministry of Education and are marked centrally by a group of teachers. They count for 30 percent of a student’s final grade, with the remaining 70 percent coming from the school-awarded mark.

Because all students must write the same exam and are evaluated according to the same standard, it’s possible to objectively determine whether students have met the appropriate academic outcomes. Importantly, students cannot use AI when writing these exams since all diploma exams are strictly supervised.

Thus, even if some students had, for example, used AI to write their English essays at home, their diploma exam marks will reveal the true level of their writing ability. If there are significant discrepancies between the diploma exam mark and the school-awarded mark, this can indicate where changes need to be made.

Unfortunately, many provinces do not have diploma exams, and this leaves their schools more susceptible to cheating with AI. For example, while British Columbia requires all Grade 12 students to write (but not pass) a literacy assessment, this assessment does not count toward a student’s final grade. Even worse, the assessment is “not based on a particular subject matter or course.” Thus, the B.C. literacy assessment has little value in combating the problem of AI cheating. This puts the burden of catching cheaters entirely on teachers and principals.

To make matters worse, standardized testing is on the decline across the country. Over the last decade in most provinces, standardized tests have been administered at fewer grade levels, given less value by provincial governments, and turned into non-content specific assessments. This is exactly the wrong direction.

If provincial education ministries are serious about maintaining academic standards, they must ensure that students write standardized tests at multiple grade levels and in a variety of subjects. Students need to know that their performance on these tests will impact their final marks and that they only hurt themselves academically if they get AI to do their work for them.

When it comes to AI, we cannot put our heads in the sand. Since AI isn’t going away, it’s important that we assess students with measurement tools where students cannot use AI to cheat.

Instead of moving away from standardized testing, every province should embrace and enhance this important measurement tool. It’s the best way to ensure all students meet basic academic standards.

Michael Zwaagstra is a public high school teacher and a senior fellow with the Fraser Institute.

Views expressed in this article are the opinions of the author and do not necessarily reflect the views of The Epoch Times.