For as long as tests have been administered in schools, students have found ways to cheat, from peering over classmates’ shoulders to scribbling notes on the palms of their hands or crib sheets.
But as technology evolves and the pressure to achieve top grades increases, students are turning to AI-powered smart glasses to gain an edge. And in East Asia’s test-obsessed societies, where a single test can impact a student’s future career and social status, educators are scrambling to get ahead of the problem.
Twice in the past month, people in South Korea have been caught using smart glasses while taking tests to assess their English skills, the results of which are often used to make hiring decisions.
See how to cheat on exam questions using AI glasses
In Taiwan, a student taking an entrance exam for a top medical school was found wearing smart glasses, and an invigilator noticed the student staring strangely at the exam, and upon inspection, the frame was found to be emitting heat.
Cheating using smart glasses is nothing new. But as AI-enabled wearable devices become more common, affordable and sophisticated, traditional aspects of education, from teaching to assessment, are under immense pressure to evolve. More broadly, this technology also reignites the debate about how to balance learning efficiency with the risk of cheating.
Countries have already stepped up testing for test takers.
Earlier this month, authorities made it mandatory for all glasses to be tested during China’s tough university entrance exam, which is taken by more than 10 million applicants each year. In the UK, the head of the country’s exam watchdog warned earlier this month that smart devices such as AI glasses and earphones could exacerbate cheating in exams.
The two incidents in South Korea were the first cases of fraud using AI glasses reported in the country.
In response, South Korean university entrance exam administrators told CNN that they are discussing measures with the Ministry of Education and local education offices to prevent cheating using AI glasses, which are already banned from entrance exam halls like all other electronic devices.
In Taiwan, a university where a student was found to have cheated is reviewing its rules and standard operating procedures for using AI glasses during exams.
But experts worry these individual cases are indicative of a broader problem.
“If there are a small number of reported cases, there are far more unreported cases,” said Thomas Corbyn, a lecturer at Australia’s Deakin University. He conducts research on the use of AI-enabled glasses and other smart devices in academic assessment.
However, with the rapid development of AI technology, smart glasses are becoming slimmer and less obtrusive, integrating AI models that can operate independently with connectivity, raising concerns not only about exam integrity but also broader privacy risks.
These wearable devices are quickly moving from novelty to mainstream. US tech giant Meta launched its first AI-enabled glasses with Ray-Ban in late 2023 and has released several new versions since then. Last year alone, more than 7 million pairs were sold.
“I don’t think there’s any real way to ensure that exams go forward as much as ChatGPT is to essays in 2022 and wearable AI is to exams,” Corbin said.
A year ago, Hong Kong University of Science and Technology (HKUST) assistant professor Meng Zili noticed the fancy glasses worn by students during an exam she was supervising. As an AI eyewear researcher, this frame caught his eye.
Glasses have become commonplace. However, this incident led him to test commercially available AI glasses on his undergraduate-level exam for his electrical engineering course.
The results were surprising and showed that AI glasses are a “viable technology” to handle the exam, Meng said. Just look at your exam paper, and the glasses send questions to a connected AI large language model that generates answers and displays them on the lenses.
The scores generated by this device were in the top five in a class of over 100 students and significantly higher than the average score of 72.
“After conducting the experiment, given the current capabilities of AI, the question really arose as to how much knowledge students actually need to memorize for the exam, and whether they should be allowed to use AI during assessment,” Meng said.
Zhang Jun, a professor of electrical engineering at the Hong Kong University of Technology who co-led the project with Meng, said the pace of advancement in technology and AI is making it difficult for on-the-ground education to keep up. “Every teacher is feeling it,” he said.
“The real question is how quickly we can rethink and adapt our education system: how we change the way we teach and how we assess students,” he says.
But despite the great disruption it brings to education, the power of AI should not be a reason to dissuade its use, said Kong Siu Chan, professor and director of the Center for AI and Digital Competence Education at the Education University of Hong Kong.
Rather, he said, education systems should focus on developing students’ thinking skills and metacognition to prevent over-reliance on AI.
“We should use technology. We should use AI. We should not just say avoid it, stop using it…The important thing is not to outsource the thinking capacity,” he added.
