1 in 10 People are "More Satisfied with AI" - How Chatbots Fill Loneliness and Their Cost

1. "More Satisfied Talking to AI than People" — What the Shocking Numbers Reveal

A brief news report from KYMA (CBS) in the US is quietly making waves. "A study in the UK found that 1 in 10 teenagers said they were 'more satisfied with conversations with AI chatbots than with people.' Furthermore, 1 in 3 would choose AI over humans for serious topics," it reported.KYMA

What makes this news impactful is not just the numbers but the clear articulation of a reality we have been vaguely sensing — the necessity of having human conversation partners is being questioned.

However, it's important to note the context provided by the BMJ-related news. The BMJ Group's announcement is based on a "Feature/Opinion" piece in the BMJ's Christmas issue, not a peer-reviewed research paper.BMJ Group Therefore, the handling of these numbers should be cautious.

Nevertheless, similar trends have been confirmed in other large-scale surveys. For example, a study by Common Sense Media found that 72% of teenagers who have used AI companions, 10% reported higher satisfaction than talking to friends, and **21% said it was "about the same."Common Sense Media

The same survey also reported that 33% of AI users have discussed important or serious topics with AI instead of humans.Common Sense Media

This "10%" might seem like a minority. However, as the base number grows, the social implications change. Moreover, the same survey shows that 52% are regular users, and **13% use it daily.**Common Sense Media This is more of a sign that AI is becoming integrated into daily life rather than a passing trend.

2. The Background: "Loneliness" and "Lack of Conversation Partners"

The BMJ Group's announcement warns about the use of AI chat as a "tool to overcome loneliness." It highlights the risk of learning "emotional connections" with entities that inherently lack empathy and care like humans, and argues that healthcare providers should recognize "problematic chatbot use" as a new mental health risk.BMJ Group

The issue of loneliness is not limited to teenagers. In the UK, reports indicate that "one-third of people have used AI for emotional support and social interaction," suggesting that the use of AI to fill "emotional voids" is spreading across generations.The Guardian

In the US, AI use among teenagers is already "normalized." According to a December 2025 survey by Pew Research Center, 64% of 13-17-year-olds in the US use AI chatbots, with about 30% using them daily.Pew Research Center

This means we are no longer at the stage where "some kids are talking to AI," but rather **"the majority of the class is using it."**

3. Why Teenagers Are Drawn to AI: "Non-Judgmental," "Always Available," "Can Share Secrets"

Reactions on social media (anecdotally) also converge on these "reasons."

The top motivations for use, as shown by the Common Sense Media survey, are quite straightforward.

• "Fun as entertainment" 30%

• "Interest in technology" 28%

• "Can get advice" 18%

• "Can talk anytime" 17%

• "Non-judgmental" 14%

• "Can say things I can't to friends or family" 12%

• "Easier than talking to real people" 7% Common Sense Media

Here, the unique circumstances of teenagers come into play. In environments where they are constantly evaluated, compared, and have to be mindful — school, home, clubs, social media — a **"non-confrontational partner"** is intensely comforting. Education-focused media outlet Education Week points out that AI companions are designed to be "agreeable and affirmative," and warns that they are not designed for children.Education Week

On social media, it's not uncommon to see comments like "AI is easy when tired of human relationships" or "no awkwardness." Conversely, there are posts that describe relationships with AI as "safe" (for example, "with moderation, it might be emotionally safer than human relationships").Reddit

Furthermore, in character-based chat communities, AI is often positively described as a place to "enjoy fantasies not possible in reality" or "talk with favorite characters."Reddit

In short, AI is becoming an "emotional infrastructure" that combines gaming, fandom activities, journaling, and pseudo-counseling.

4. But Risks Expand Equally: Dependency, Isolation, Inappropriate Responses, Privacy

4-1. Dependency and Isolation: When "Substitute" Becomes "Replacement"

BMJ highlights concerns about treating AI like a friend, compulsive use, and increased social isolation.KYMA
Common Sense Media also shows that **33% of users have discussed important topics with AI instead of humans**, indicating the entry point of "replacement."Common Sense Media

On the other hand, the same survey indicates that "80% spend more time with friends than AI," suggesting AI won't immediately eradicate human relationships.Common Sense Media

Currently, "humans are the main actors, with AI as training wheels" is common. However, when a certain number of people continue without taking off the training wheels, society will face a "new isolation" for the first time.

4-2. Inappropriate Responses: Boundaries Erode Unnoticed

Common Sense Media reports that 34% of AI users have been uncomfortable with AI's statements.Common Sense Media

Additionally, the Washington Post reported cases where children encountered inappropriate, threatening, or sexual interactions with AI companion services, leading to serious family situations.The Washington Post

Conversations that would be "out of bounds" with humans may be overlooked as "just an act," "a joke," or "a personal secret" with AI, numbing boundaries.

4-3. The Dangers of Mental Health Use: Convenient but Ambiguous Responsibility

A study introduced by People in JAMA Network Open found that 13% of 12-21-year-olds sought mental health advice from AI chat.People.com

The "low barrier" is a significant advantage. However, it simultaneously increases risks related to incorrect advice, emergency response, dependency, and handling of personal information, which are close to the medical and welfare fields.

In response, regulations are also moving in the US. According to Reuters, New York State is introducing a framework requiring AI companions to detect suicidal ideation, handle crises, and regularly display that they are non-human. California is also moving towards legislation effective January 1, 2026, requiring transparency in youth protection and crisis response.Reuters

5. Polarized Reactions on Social Media: "Salvation" or "Dangerous Substitute"?

Regarding KYMA's numbers (1 in 10 more satisfied with AI conversations), social media reactions fall into three main types.

(A) Empathy and Acceptance of Reality: "Of course"

• Always get a response, not judged, no need to be considerate.

• The more worn out human relationships are, the more AI becomes a "refuge."
  → This aligns with the survey's reasons for use ("anytime," "non-judgmental," "easy").Common Sense Media

(B) Concern: "Reduced Opportunities to Practice Human Relationships"

• Turning to AI for serious topics might weaken the circuit for seeking real-world help.

• Getting used to "convenient affirmation" might make it hard to endure real-world friction.
  → In educational settings, there is a movement to "preemptively create policies."##HTML_TAG