Keep an Eye on Whether They Touch Their Glasses? Universities Helpless as "AI Cheating" Evolves

As artificial intelligence (AI) glasses newly emerge as a means of cheating on exams, the educational community is failing to come up with clear countermeasures.

While the latest models are increasingly difficult to distinguish from regular glasses due to their thin temples and weight of around 50 grams, measures remain at a makeshift level of simply trying to spot AI glasses with the naked eye.

According to media reports on August 1, the Ministry of Education recently sent official notices to regional offices of education instructing them to include AI glasses on the list of prohibited items and to closely watch for unnatural behavior.

This follows a series of incidents starting with three test-takers caught wearing AI glasses at the Electrical Engineer and Fire Equipment Engineer exams in May, followed by two more test-takers caught consecutively during regular TOEIC examinations.

However, universities are still falling short on preventive measures against cheating ahead of the upcoming College Scholastic Ability Test (CSAT) and early admissions essay screenings in the second half of this year.

In-school cheating, in particular, remains in a blind spot.

Unlike the CSAT and other national tests, cheating on campus is not subject to criminal punishment, leading to criticism that universities are stepping back and responding passively.

Even after universities suffered consecutive blows last year from cheating cases using generative AI, their responses to cheating are failing to keep pace with the speed of AI technology development.

An official from Seoul National University explained, "As far as I know, there have been no discussions yet regarding responses to cheating using smart glasses on campus."

Korea University, Ewha Womans University, and Sungkyunkwan University stated that while they do not have separate guidelines yet, they are discussing potential measures.

Evaluations indicate that even where internal campus guidelines have been established, their practical effectiveness has clear limitations.

The College of Medicine at Chung-Ang University prohibited AI glasses from test venues early on in its student evaluation guidelines established in September of last year.

The detection guidelines also included methods such as "checking whether the glasses temples are abnormally thick" or "observing repetitive touching or manipulation of the glasses temples."

This reflected the forms of AI glasses available on the market at the time.

This was because they were thicker than regular plastic frames and could be operated with noticeable hand gestures.

However, as technology has rapidly advanced in just one year, such detection methods have become difficult to apply.

Taking Samsung's AI glasses unveiled on July 22 as an example, they are voice-driven models without a display, and with thin temples and a weight of just 50g (110 lbs), it has become difficult to distinguish them from regular thick-rimmed glasses.

Experts unanimously agree that extraordinary measures tailored to the speed of technological advancement are necessary.

Park Nam-gi, professor emeritus of the Department of Education at Gwangju National University of Education, suggested, "Since technological advancement has brought an era where they are indistinguishable from regular glasses, we ultimately have no choice but to introduce detectors," adding, "We must also consider methods such as recording test-taking sessions."

There are also many predictions that new cheating methods will continuously emerge as various AI devices develop rapidly.

Calculators equipped with generative AI and built-in cameras have already gone on sale overseas, and small AI devices such as "smart contact lenses" are currently in the development stage.

This is why arguments are being raised that a fundamental shift in examination and evaluation methods is needed in line with changing times.

Moon Hyung-nam, president of the Korea AI Education Association, emphasized, "We must change the testing system to evaluate thinking skills, problem-solving abilities, and creativity under the premise that AI can be utilized," adding, "Just as mathematics education changed with the advent of calculators, innovation in evaluation methods is needed in the AI era."

Moon pointed out that conventional testing methods, which filter out correct answers from a limited set of choices and where memorization determines success or failure, have reached their limits in the face of the AI era.

Professor Emeritus Park Nam-gi also pointed out, "Guidelines that simply prohibit the use of AI will ultimately allow people to continue cheating on exams to whatever extent they want."

He further suggested, "We must change the paradigm of evaluation by having students organize and submit every step of the problem-solving process and attach research notes, providing them with opportunities to think for themselves even when using AI."

(Photo: Yonhap News)
※ Please note: This article was translated by AI and may contain errors.