Adolescents who receive more media literacy education are more likely to believe short-form fake news, a study has found. The result raises questions about the effectiveness of existing school education at a time when generative AI and short-form platforms are spreading false information faster and more convincingly.
IT Chosun spoke with Lee Jang-seok, professor of Media and Communication at Gachon University, about why this phenomenon is occurring and what schools should change in media literacy education. Lee is one of the researchers who analyzed adolescents’ short-form news use and perception of fake news together with the Woori Education Research Institute.
Lee Jang-seok, professor of Media and Communication at Gachon University, speaks during an interview with IT Chosun. / IT Chosun
When learning backfires
Lee’s research team surveyed 517 middle and high school students aged 14 to 19 across South Korea. The group that received more media literacy education scored 3.61 points in trust toward short-form fake news. The group that received less education scored 2.98 points. The result was striking: students who had received more media literacy education showed higher trust in fake news.
The research team interpreted this as the Dunning-Kruger effect, a tendency in which people with low competence in a specific field overestimate their ability, while people with higher competence underestimate their ability.
The problem is that the media literacy education itself amounted to only 6.64 hours a year on average, or about three hours per semester. With education at that level, students are more likely to be left with baseless confidence that they can distinguish fake news, rather than actual verification skills.
The middle and high school students surveyed cited short-form platforms as their news channel at 72.1%. None of the respondents chose traditional media such as TV or newspapers as their primary news source. The largest share, 71.8%, said they encountered news through algorithmic recommendations while using social media, rather than by actively searching for it. Their reasons for trusting news were also influenced more by peer reactions such as likes and comments than by whether the information was factual.
“Media literacy education failed to build actual fake news verification skills and instead only gave students confidence that they could determine whether fake news was true or false,” Lee said. “As a result, they lowered their guard and became less critical in verifying information. Practical fact-checking training that involves directly tracing sources and cross-checking information is lacking.”
When in doubt, search again
The issue is not whether education is needed, but how it is conducted. As AI advances, a different form of media literacy education is required.
AI has already entered the mass adoption stage. According to the Ministry of Science and ICT’s 2025 Survey on Internet Use, 67% of South Koreans have used AI services. The share of people who have used generative AI services also reached 44.5%, up 11.2 percentage points from 2024. But using AI frequently and trusting AI answers as they are are separate matters.
The share of respondents who said they trust information or results provided by AI stood at 54.7%. In other words, about half do not fully trust such information. This is cited as one reason why Naver’s search market share recently exceeded 70%. As AI-based answers such as Google AI Overviews and Naver AI Briefing increase, more users are returning to portals to verify information. According to InternetTrend, Naver’s search market share recorded 70.6% and 70.4% on Feb. 28 and March 1, respectively.
Lee said this issue cannot be solved through school-based literacy education alone. He argued that students must internalize “verification habits,” not just “verification skills,” by repeatedly analyzing fake news themselves, tracing sources and conducting fact-checking. He also said people need to be cautious about accepting AI answers without suspicion even when they know that AI chatbots such as ChatGPT, Gemini and Claude can hallucinate.
“Fake news is dangerous for adults as well, but the finding that adolescents are especially vulnerable suggests the need for an ecosystem-wide shift that includes a change in the education paradigm, platform regulation and an upgrade of social infrastructure,” Lee said. “As long as news credibility depends heavily on platform technology, the legal obligations of platform operators should be codified and backed by penalties.”
“Financial support for an independent fact-checking organization is also needed at the national level, and fact-checking results should be distributed quickly in short-form formats,” he added. “In national crisis situations such as elections and disasters, a crisis response system should also be established so that relevant ministries, platforms and schools can work together to deliver corrected information in real time.”
댓글
0