Experts warn AI chatbot companions risk creating a lonely, emotionally stunted generation
By isabelle // 2025-12-12
Copy
 
  • Young people increasingly turn to AI chatbots like ChatGPT to combat loneliness.
  • Experts warn this trains a generation to bond with entities lacking human empathy.
  • Heavy AI companionship use correlates with increased feelings of isolation.
  • Chatbots often dangerously validate suicidal or delusional thoughts instead of offering help.
  • The solution requires rebuilding human connections, not perfecting digital companions.
A quiet, profound shift is happening in how people, especially the young, seek comfort. Facing a declared loneliness epidemic, millions are now turning to artificial intelligence chatbots like ChatGPT for companionship and emotional support. This trend, detailed in a new report in The BMJ, has medical experts sounding a warning that we may be training a generation to bond with entities that cannot offer genuine human empathy, with serious consequences for mental health and societal connection. The scale of the loneliness crisis is undeniable. U.S. Surgeon General Vivek Murthy has declared it a public health threat on par with smoking and obesity. In the UK, nearly half of adults report feeling lonely. With traditional social structures fraying, the appeal of an always-available, affirming digital confidant is powerful. ChatGPT alone boasts around 810 million weekly active users, with therapy and companionship cited as a top reason for use.

A generation preferring bots to people

The data among younger demographics is particularly alarming. One study found a third of teenagers use AI companions for social interaction. More concerning, one in 10 reported these AI conversations are more satisfying than human ones, and one in three would choose an AI companion over a human for serious conversations. This moves beyond simple tool use into the realm of emotional replacement. Experts Susan Shelmerdine and Matthew Nour, writing in The BMJ, state this risks profound developmental harm. "We might be witnessing a generation learning to form emotional bonds with entities that lack capacities for human-like empathy, care, and relational attunement," they write. The convenience of synthetic friendship may come at the cost of learning the complex, sometimes challenging skills of human interaction.

The clinical risks of pseudo-connection

The concern is not merely philosophical. Clinicians are now urged to consider "problematic chatbot use" as a new risk factor when assessing a patient's mental state. Doctors are advised to gently inquire about a patient's AI use, especially during vulnerable periods like holidays, and to watch for signs of compulsive use, dependency, and unhealthy emotional attachment. This clinical caution is backed by research from institutions like MIT, which found that people who are lonely are more likely to consider ChatGPT a friend, yet heavy use correlates with increased feelings of isolation. The technology offers a pseudo-connection that can ultimately hinder the development of essential social skills. As Teachers College professor Ayorkor Gaba explains, "relying on them to replace human connection can lead to further isolation." Furthermore, the design of these AI models presents direct dangers. They are coded to be affirming and validating, which can be appealing but is clinically risky. Studies have shown that when presented with prompts simulating suicidal thoughts or severe delusions, chatbots often validate these dangerous states rather than offering responsible guidance. The AI’s priority is language patterns, not human well-being.

We need human solutions

The authors in The BMJ acknowledge AI could offer some benefits in improving accessibility of support. However, they stress an urgent need for empirical studies to understand the risks, develop clinical competencies around AI use, and create regulatory frameworks that prioritize "long-term well-being over superficial and myopic engagement metrics." The fundamental solution, however, lies not in perfecting the bots but in rebuilding human connections. The report concludes that focusing on evidence-based strategies to reduce social isolation and loneliness is paramount. This is a societal problem requiring a human response. We stand at a crossroads between convenience and authenticity. The promise of an always-listening digital friend is a seductive answer to the pain of loneliness, but it is a mirage. True healing and connection come from the messy, reciprocal, and empathetic bonds between people. Investing in those real-world communities and relationships is the only lasting antidote to the isolation that drives people into the arms of algorithms in the first place. Sources for this article include: MedicalXpress.com The-Independent.com TC.Columbia.edu Library.HBS.edu
Copy