
Generative AI chat services — seen primarily as study or work tools — may serve as emotional companions and even informal “counselors” for many young people in South Korea.
A new survey by local market research firm Embrain, conducted from Feb. 7-12 with a sample of 1,000 Koreans aged 13 to 59, showed that about 40 percent of respondents in their teens and 20s believe they can have meaningful emotional exchanges with generative AI.
Specifically, 38 percent of teenagers and 42 percent of people in their 20s expressed this view. This is notably higher than respondents in their 30s (29.5 percent), 40s (32.5 percent) and 50s (31.0 percent).
Whether it’s venting about a bad day, sharing personal thoughts, or simply engaging in casual back-and-forth conversations, these AI chat services might be filling a role that, until recently, was for human interaction.
Seeking advice: AI as a source of guidance
Beyond casual conversations, some younger users go a step further, using AI chat services for mental and emotional guidance. The survey found that 19.8 percent of teenagers and 17 percent of people in their 20s said they would turn to AI for mental health support or personal advice.
This number was lower for those in their 30s (14.8 percent), 40s (14.3 percent) and 50s (14.8 percent).
While AI chat services are not trained therapists, their ability to provide immediate responses, non-judgmental interactions and seemingly personalized advice may explain why some young users find them helpful when dealing with stress, loneliness or personal dilemmas.
Another reason young people engage more deeply with AI could simply be their higher adoption rates. The survey found that 95 percent of teenagers and 84 percent of people in their 20s have used ChatGPT — exceeding the overall average of 71.3 percent.

Can AI chats improve mental health?
There is some scientific evidence suggesting that AI chats can have mental health benefits. A January 2025 study by Ulsan National Institute of Science & Technology and South Korean AI startup ScatterLabs found that regular interaction with the startup's AI chatbot Luda Lee 2.0, introduced in 2022, reduced users’ loneliness scores by 15 percent on average and social anxiety scores by 18 percent.
Despite their increasing use, generative AI chat services still face trust and reliability issues. The survey found that 69.6 percent of respondents said they do not fully trust AI-generated information. Additionally, 83 percent of participants emphasized the importance of fact-checking AI outputs.
There is also growing anxiety over AI’s broader societal impact. Some 81.2 percent of respondents expressed concern about deepfake risks, while 79 percent worried about copyright issues involving AI-generated content.
mjh@heraldcorp.com