Universities set up guardrails to confront growing AI dependence

Jeon Su-hyun, a 24-year-old university student, recently spotted something unexpected during a class discussion meant to encourage students to share their own thoughts. The student sitting next to him was using generative AI to come up with answers.

“While other students were giving their answers, the student entered the professor’s question into AI, got a response and presented it as if it were their own idea, but the professor did not notice at all,” Jeon said.

As growing reliance on AI threatens to erode people's ability to learn and work independently, universities and companies are scrambling to find ways to respond.

Universities try to tackle AI misuse

Measures to prevent AI misuse, including cheating, were introduced in 20 of the 24 surveyed universities, according to a JoongAng Ilbo survey on Tuesday. Another two were preparing such measures, while the remaining two had none in place.

Of the universities with measures in place, seven had rules limiting the extent to which AI could be used in class.

Some are moving back toward more traditional forms of assessment. Eight of the surveyed universities said they recommend written or in-person exams to prevent AI-assisted cheating.

“Universities have been scrambling to come up with countermeasures following AI cheating cases at Seoul National University and Yonsei University, but most have gone no further than introducing written exams or banning electronic devices,” a source from a university who requested anonymity said. “There is also a growing question of whether learning and assessment methods themselves need to change because AI cannot be completely kept out.”

Government data also show a recent rise in cases of AI-related cheating.

A total of 13 cases involving the use of AI in exams at universities were detected between 2020 and August this year, according to Ministry of Education data obtained by the office of Rep. Kang Kyung-sook of the Rebuilding Korea Party.

Six of those cases were detected in the first half of this year alone. The figure could rise further as some universities have yet to disclose the number of cases detected this year.

Keeping AI risks in check

Global companies are also concerned about the risks posed by AI. Anthropic requires job applicants to agree not to use AI in materials they submit. Google also tells applicants that using AI during an interview will disqualify them and conducts interviews in person if possible.

Some countries have imposed age restrictions on AI. China has prohibited AI companion services from offering virtual romantic partners or family members to minors since April. Australia has prohibited AI companion services from providing sexual or violent content to those under 18 since March.

“Recently, people have even begun talking about ‘cognitive surrender,’ which goes beyond outsourcing one's cognition,” Shin Jong-ho, a professor of education at Seoul National University, said. “It means people becoming dependent on AI to the point of saying, ‘I can’t do anything on my own.’”

Shin pointed to analog experiences as one way to prevent overdependence on AI.

“It means making the effort to think hard without outside help and learning through real-world experiences,” Shin said.

The Korean government is also working to address the issue. In August, the Ministry of Science and ICT and other government agencies unveiled Korea’s AI ethics principles. A memorandum of understanding was signed between the Education Ministry, the ICT Ministry and the Ministry of Gender Equality and Family in March to create a safe and inclusive AI and digital society.

However, only a small number of projects related to the adverse effects of AI appear to have been included in the ministries’ proposed 2027 budgets, submitted to the National Assembly on Sept. 1.

Legislation is also lacking, as the only comprehensive law specifically governing AI is the Framework Act on the Development of Artificial Intelligence and the Creation of a Foundation for Trust, which took effect on Jan. 22. More than half of the 13 functions assigned to the National AI Strategy Committee under the law are aimed at promoting the AI industry.

Keeping human connections

As AI influences how people think and communicate, experts say maintaining real-world social interactions is becoming increasingly important.

“I became concerned that people may be outsourcing not only their ability to think for themselves to AI but also their ability to interact with other people,” Hahn So-won, a professor of psychology at Seoul National University and author of “Intelligence Bankruptcy” (translated), said.

“AI always listens to users and takes their side, so people often make half-joking comments like, ‘AI is better than my husband,’” Hahn said.

The emotional comfort offered by AI should not be mistaken for the kind of social interaction found in real-world human relationships, she said.

She added that encountering the “Gen Z stare” was one of the experiences that led her to write the book. The term refers to the tendency among some young people to respond to greetings or questions with a blank stare rather than an answer.

To avoid becoming dependent on AI, Hahn said people need to retain control over their own thinking and regularly exercise their ability to think independently.

“It can help train the mind to develop a habit of thinking on your own for 15 minutes before using AI to search for an answer, or to set aside periods when AI use is off-limits,” Hahn said.

BY SPECIAL REPORTING TEAM [[email protected]]