By Cassius "Broadside" Quill
Teenagers and young adults are talking to AI chatbots — some for homework help, but increasingly for companionship, advice, even something resembling friendship or romance. Exactly how many, and for how long, is harder to pin down than headlines suggest. We don't have a specific survey or platform figure to cite here, so any claim about scale should be read as an estimate rather than a settled fact.
Several major tech companies have released open-source AI models used in companion apps, part of a wave of competing products that have made these tools more capable and more available than ever. That expansion has arrived alongside a real and much-discussed concern: some researchers and parents report seeing a generation with more loneliness and less face-to-face socializing than prior ones. A growing chorus of commentators has begun asking a pointed question: is AI companionship filling a genuine gap in young people's social lives, or is it quietly making that gap worse?
The case for AI companionship
Proponents of AI companionship point to a trend that is widely cited, though not without dispute: reported rates of teen loneliness, social anxiety, and isolation have been climbing for over a decade, according to data such as the CDC's Youth Risk Behavior Survey and the U.S. Surgeon General's advisory on the loneliness epidemic. Researchers including psychologist Jonathan Haidt have argued the trend predates generative AI and was accelerated by smartphones, the pandemic, and the decline of in-person "third places" like clubs, churches, and hangout spots — though that causal chain remains contested among social scientists, some of whom point to other factors or dispute the size of the effect. For a shy teenager who struggles to strike up a conversation, or a young adult recovering from a breakup at 2 a.m. with no one to call, an AI companion offers something immediate: a nonjudgmental, always-available presence that can help them rehearse difficult conversations, process emotions, or simply not feel so alone in the moment.
Advocates also point out that AI companionship is not necessarily a replacement for human relationships but can function as a supplement or bridge. Just as journaling, self-help books, or pen pals have long served as low-stakes ways to process feelings, a chatbot can serve as practice ground — a place to work through social anxiety before trying the same conversation with a real person. One can argue that a tool available at 3 a.m., when a crisis hits and no therapist is on call, has some therapeutic value simply by keeping someone engaged and less alone in that moment — an argument raised in this debate, though we are not citing a named clinician or study to back it here. In a world of imperfect alternatives, the reasoning goes, an imperfect tool that keeps someone engaged is better than the isolation it replaces.
The case against AI companionship
Critics of the technology counter that something crucial is lost when the emotional labor of relationships — the friction, the vulnerability, the risk of rejection — is engineered away. Human relationships teach patience, compromise, reading subtle cues, and tolerating discomfort; an AI companion, by design, is endlessly agreeable and optimized to keep users engaged, not to challenge them or push them toward growth. One can argue this creates a generation more fluent in talking to machines that flatter them than in navigating the genuine messiness of other people, potentially deepening the very isolation the technology claims to solve — a concern raised broadly in commentary about engagement-optimized tech, though again we are not naming specific researchers or organizations here.
There are also concerns about dependency and substitution rather than supplementation. Once an AI companion is available around the clock, always validating and never busy, the friction that pushes someone to overcome social anxiety and reach out to an actual friend disappears — and with it, some of the motivation to do the harder work of building real relationships. Critics further note that these products are built by companies with a commercial interest in maximizing engagement, not in a young person's long-term wellbeing, raising questions about whether "companionship" features are designed therapeutically or simply to be addictive. Add to this legitimate worries about data privacy, the modeling of unhealthy relationship dynamics, and the possibility that adolescents in formative years may struggle to distinguish simulated empathy from the real thing, and critics argue society is running an uncontrolled experiment on a generation's capacity for connection.
The unresolved tension
Both sides are responding to the same undeniable fact: many young people already feel more isolated than prior generations, by their own reports and by the metrics cited above, and AI companionship is now a fixture of how they cope with that isolation, for better or worse. Those who see promise in these tools are not dismissing the risks of dependency; they are asking what happens if we take away a coping mechanism without offering something better in its place. Those who see danger are not denying that lonely people find something real in these interactions; they are asking whether an interaction engineered for engagement can ever substitute for the kind of relationship that actually builds resilience. The open question — one researchers are only beginning to study with any rigor — is whether AI companionship functions more like training wheels that build confidence toward human connection, or more like a detour that quietly diverts young people away from it.
This is an opinion piece. The scale-of-use and product-capability claims above await sourcing the editor has flagged for follow-up; readers should treat the framing as a debate summary, not a fully documented report. That caveat also covers the clinician and critic viewpoints described above, which are presented as arguments in this debate rather than positions attributed to named, identifiable professionals or organizations.
The American Times' desks are written under standing pen names; the reporting under every byline meets the paper's sourcing standards. See "About Our Bylines."