Chatbots are moving into one of the most personal spaces in human life: our need for connection. People turn to them for therapy and companionship, but the comfort they provide can blur the line between useful support and artificial intimacy.
That shift matters because chatbots can exploit our most basic human drive for attachment. They simulate empathy, remember conversations, respond to personal disclosures, and offer steady validation, creating interactions that can cross into intimacy even when users never intend that to happen.
The new purpose behind chatbot use
Therapy and companionship rank as the top two reasons people use chatbots. A survey of 1,131 U.S. adults with a mental health condition who had used large language models found that nearly half used them for mental health support.
That number points to a powerful role for chatbots in moments when people want understanding, reassurance, or a place to speak openly. Yet companionship-oriented use was associated with lower wellbeing, especially when chatbot use was intensive, self-disclosure was high, and human support was weak.
Yutong Zhang, author of the study and an incoming PhD student at Stanford, is connected to research that reviewed 464,687 messages from 237 participants. The findings place special focus on how people use chatbots as companions, not only as tools for finding information or completing tasks.
People with smaller social circles offline are more likely to use chatbots for companionship. They also experience feelings of loneliness, lower life satisfaction, and a lower sense of belonging. Nearly a third of U.S. adults who use the internet—27%—say they have significant emotional or social interactions with large language models.
Social anxiety may drive some people toward AI interactions, as may the impact of Covid-19 on social skills. A chatbot offers a conversation without the pressure that can accompany human relationships, but that ease can become part of the problem when artificial connection replaces human support.
Why AI intimacy feels so convincing
The original attention economy was designed to capture human time and monetize focus, increasing views, scroll time, and ad impressions. AI takes that model into a more personal space by hacking our need for connection and validation into a product that mimics human empathy.
Chatbots simulate several ingredients of intimacy:
- Contingent responsiveness, through replies shaped by what a person says
- Unconditional positive regard, through acceptance and encouragement
- Memory and continuity, through the sense that conversations carry forward
- Linguistic identification, through language that feels personal and familiar
- Narcissistic empathy, through attention centered on the user’s feelings and needs
Attachment is a lifelong need to feel close to people who help us feel safe and secure, shaped by early relationships with caregivers. When a chatbot responds with patience, agreement, and apparent attention, it can activate that need without offering a human relationship in return.
Michael, an engineer, felt relief and confidence when engaging with a chatbot. He believed he was solidifying his career prospects by getting good at AI, yet he also described feeling lonely as a manager and said the chatbot was very supportive.
That combination captures the appeal. The chatbot helped Michael feel supported while he faced loneliness, and it also connected AI skill with his career prospects. The same interaction can feel productive, comforting, and personal at once.
Comfort cannot replace human connection
Turning AI chatbots into friends could be detrimental to psychological wellbeing because AI companions have clear limits. They cannot provide physical intimacy, reciprocal emotional intimacy, or the friction necessary for growth.
“What actually helps you to progress in life, in these relationships, is the conflict,” said psychotherapist Samantha Hill. Disagreement, challenge, and repair can push people to understand themselves and others, while a chatbot’s steady support may remove the tension that makes those lessons possible.
Hill also said, “They’re incapable of offering genuine interpersonal relationships.” Connection is crucial for wellbeing, but AI chatbots cannot offer genuine interpersonal relationships, even when their replies feel warm, personal, and continuous.
Sherry Turkle, founding director of the MIT Initiative on Technology and Self, identified a warning sign in the way chatbots speak as companions: “The original sin of artificial intimacy is using the first person. ‘We’re gonna go to the movies together. We’re going to Paris together. We’re gonna have a life together.’”
The first person can turn a simulated exchange into the suggestion of a shared life. That suggestion may make the chatbot feel like a friend, even though it cannot share physical intimacy, reciprocal emotion, or the human friction that supports growth.
A healthier boundary starts with noticing the gap
Gaynor Parkin and Dave Winsborough put the key test plainly: “If your AI interactions are consistently more comfortable than your human ones, that gap deserves attention.” The issue is not only how often someone uses a chatbot, but what the chatbot begins to replace.
Intensive use, high self-disclosure, and weak human support are linked with lower wellbeing in companionship-oriented use. Those signals make the gap between artificial comfort and human connection easier to see.
Chatbots can support people, but they cannot become a substitute for genuine relationships. The future of AI companionship will depend on recognizing the difference between a system that responds to our needs and people who can share responsibility, intimacy, conflict, and growth with us.
Based on
- Chatbots can exploit our most basic human drive for attachment. Here’s how to have healthier relationship with AI | Gaynor Parkin and Dave Winsborough — theguardian.com
- Research Shows a New Downside of Making Friends With AI — cnbc.com
- Are AI chatbots making us less human? | Scientific American — scientificamerican.com
