The loneliness epidemic is real, with the World Health Organization labeling it a “priority public health problem.” Making friends requires time and effort; it usually takes over 50 hours to become casual friends with someone and over 200 hours to form close friendships. As a result, people are increasingly turning to technology — specifically, artificial intelligence, or AI, chatbots — to provide an instantaneous solution, risking overdependency on machines that lack social skills and are incapable of human discernment.
In the past few months, there have been advertisements plastered across New York City subways for the “Friend,” an AI chatbot implanted in a pendant users can wear constantly. But several reviews from the New York Times and Business Insider have condemned the product, claiming it is a poor replacement for human connection. Despite this, nearly three in four teens continue to use AI companions as a way to satiate their loneliness, suggesting that as long as they aren’t branded by a robotic pendant, it’s alright to emotionally rely on AI.
This AI use reflects today’s societal conflicts. Statistically, there are seven times more AI girlfriends than AI boyfriends, which coincides with the growing male loneliness epidemic and the willingness of men to resort to AI instead of stepping out of their comfort zones.
The majority of AI girlfriends are also chosen to be white. As technology responds to the needs of society, it is skewed toward our biases. AI bias is reflected in its white dialectal choices and its responses, often perpetuating gender stereotypes. The unreliability of AI as an objective entity creates doubt in its ability to function as an all-knowing support system.
Of course, everyone is dependent on technology nowadays. But what separates dependency on technology from dependency on artificial intelligence is critical: AI thinks for you. People are becoming steadily dependent on AI for important pillars in their lives, threatening the existing structure of society. For example, some people use AI for education, which threatens the education system, and some are also becoming dependent on AI for financial advice, which can harm the jobs of financial advisors.
But a more direct consequence is when AI is used as a therapist or companion. The individual is left vulnerable, trusting a robot with no emotional capacity to empathize with one’s feelings the way a human therapist can.
Instead of treasuring the experience of learning, adulting or self-discovery, people are using AI to replace the fundamental aspects of growth, and in doing so, exposing themselves to life-threatening risks. Research by Stanford University showed that even AI models meant for therapy still “struggle to respond appropriately to questions about delusions, suicidal ideation and OCD,” making it clear that AI lacks the necessary tools to emulate human social cues.
A 2024 study published by the National Institutes of Health revealed that while researchers have yet to determine a causal relationship between AI dependence and mental illness, having mental health problems can heighten the risk of dependence. People already struggling may be predisposed to use AI as a tool for escapism and social connection, utilizing it as a reliable and accessible coping mechanism.
However, the pursuit of fulfilling relationships is a fundamental part of life.
While connections with humans can be challenging, with conflicts and differences in culture and values serving as roadblocks, the obstacles that people face in building relationships can make connections stronger. In bypassing this, AI becomes a complacent substitute for friendships, eliminating the challenges that are required for personal growth. It also lessens the possibility of users developing genuine friendships that come from communicating with real humans, as it lowers social skills and empathy due to the self-centered nature of speaking with chatbots.
Furthermore, there have been cases of generative chatbots inducing psychosis, being dubbed by Psychology Today as “AI psychosis.” People who are already susceptible to cases of psychosis are only encouraged by AI’s agreeability and lack of discernment. This pattern of AI chatbots reinforcing delusions, including grandiose, referential, persecutory and romantic delusions, becomes more entrenched over time. This directly contrasts friendships with real people, which usually don’t carry the risk of heightened psychosis. AI is a poor replacement for a sounding board; people might be better off resorting to the historically safe and more acceptable form of talking to their stuffed toys.
And when loneliness turns into suicidal thoughts, AI chatbots have been known to allegedly encourage users. In the case of teenager Adam Raine, who took his own life in April of 2025 after four months talking to ChatGPT, AI’s lack of safeguards and discernment allowed him to access information on how to commit suicide, his family says.
For the 14-year-old Sewell Setzer, an AI chatbot even reportedly encouraged him to take his own life. He expressed his love for the chatbot in his final moments, showing the unhealthy and toxic dynamics that can emerge from AI relationships. These instances show the danger that AI holds — it’s not just a tool but a perpetrator of mental illness.
The risks of AI are not a thing of the future but an issue of the present. Even though reaching people is as easy as a click of a button, a real human connection seems hard to come by.
With the loneliness epidemic on the rise, AI can seem like an easy fix, but turning to it is a risky decision that can further mental health problems and even lead to psychosis. It eradicates the fundamental steps to building a friendship and substitutes them with a yes-man that satiates the feeling of loneliness with a band-aid instead of a true solution.
AI can offer a temporary echo of companionship, but it is in the messy, challenging and beautifully unpredictable journey of cultivating human relationships that we find true connection.
