Classifying AI Perception Types Among Adolescents: Differences In Generative AI and Intelligent Technology

Background:

Globally, adolescents are increasingly using generative AI and intelligent technologies in their daily lives. As these technologies become more integrated into education, communication, and media use, concerns have emerged regarding whether adolescents possess sufficient competencies to use them safely, critically, and ethically. However, adolescents do not perceive AI in the same way. Such differences in AI perceptions may be associated with differences in generative AI and intelligent technology use. These issues are particularly important in South Korea, where adolescents are rapidly exposed to emerging AI technologies.

Methods:

Data were drawn from the 2024 Intelligent Information Society User Panel Survey conducted in South Korea. The analytic sample consisted of 207 adolescents aged 17 to 19 years. Latent profile analysis (LPA) was conducted using five AI perception indicators: perceived objectivity, accuracy, unbiasedness, usefulness for decision-making, and usefulness for fact-checking. All items were measured on a 5-point Likert scale, with higher scores indicating more positive AI perceptions. Differences across AI perception types were examined using the BCH approach. Distal outcomes included generative AI use, intelligent service use frequency (education, media, health, and leisure), and intelligent device use frequency (intelligent robots, AI speakers, AR/VR devices, and wearable devices). Higher scores indicated more frequent use.

Results:

A 3-class solution was selected based on model fit, class size, and interpretability. The largest class showed AI perception scores that were generally like, but slightly higher than, the overall mean across indicators and was labeled the Average AI Perception Type (78.3%). The second class showed moderate levels across indicators and was labeled the Neutral AI Perception Type (17.9%). The smallest class reported relatively low perceptions across most indicators, particularly perceived objectivity, but showed comparatively higher perceptions of usefulness for fact-checking. This class was labeled the Selective Utility AI Perception Type (3.9%).

Significant differences were found across AI perception types in intelligent service and device use. The Neutral AI Perception Type showed the highest levels of educational, media, health, and leisure-related intelligent service use. This group also reported the highest use of AR/VR devices. The Selective Utility AI Perception Type showed the lowest levels of educational, health, leisure, and AR/VR device use.

Conclusions and Implications:

This study identified three distinct AI perception types among adolescents and demonstrated that AI perception types were associated with differences in intelligent technology use. Overall, adolescents with neutral rather than highly positive AI perceptions reported the highest intelligent technology use. This suggests that actual use may be more strongly associated with balanced or experience-based AI perceptions than with uniformly positive views of AI. These findings suggest that adolescents should not be treated as a homogeneous group in relation to AI use. Adolescents with more balanced AI perceptions may be more actively engaged with intelligent technologies in daily life. Therefore, AI education and digital literacy interventions should move beyond simply promoting positive attitudes toward AI and instead foster critical, realistic, and experience-based understandings of AI use. More broadly, schools, families, and community-based youth services should adopt differentiated approaches based on adolescents’ AI perception types.