The AI Trap: How Our Quest for Easy Connection is Undermining Real Intimacy
In a recent focus group in London, a striking observation emerged from young people discussing their interactions with AI, particularly ChatGPT, which they affectionately termed “my chatty.” Several participants revealed a concerning trend: they were turning to AI for conversations they felt would burden their friends. This sentiment resonated deeply with Jackie Jantos, the CEO of Hinge, a dating app built on the premise of facilitating genuine human connection.
Jantos immediately recognised the underlying instinct, but also its profound implications. “We’re supplementing the gift we might offer someone else of asking for their help with a piece of technology because we feel like it’s a burden,” she explained in an exclusive interview. “But the burden is the gift. That’s how intimacy is built.” This perspective suggests that the very friction and vulnerability inherent in human relationships are what foster deeper connections, a concept increasingly being bypassed by the ease of artificial intelligence.
The current loneliness crisis has spawned numerous explanations, from the pervasive influence of social media and the lingering effects of the pandemic to declining marriage rates and a perceived aversion to in-person interaction among Gen Z. However, Jantos, whose company possesses unparalleled insight into human romantic behaviour, offers a more nuanced and unsettling diagnosis. The issue isn’t merely that people are spending less time together, though that is demonstrably true. Rather, AI has emerged as a ready substitute for the small, vulnerable, and often awkward moments that historically compelled individuals to seek each other out. In doing so, AI appears to be subtly eroding the psychological foundations upon which meaningful relationships are built.
The Erosion of Relational Skills
The sheer scale of this disconnection is staggering. Data from the Surgeon General indicates that Gen Z spends approximately 1,000 fewer hours per year in the physical company of others compared to previous generations. This translates to nearly six weeks of lost human contact annually, or over two hours of daily interaction simply vanishing. When faced with boredom, isolation, or discomfort, the immediate impulse for many is to reach for a screen.
“When they put down that phone after scrolling for however long,” Jantos observed, “that emotion hasn’t shifted. It really hasn’t.” The act of scrolling, far from providing genuine relief, has become a form of avoidance that has calcified into a deeply ingrained habit.
The COVID-19 pandemic exacerbated this trend, striking Gen Z during a critical developmental period in their late teens and early twenties. These are the formative years for developing crushes, learning to flirt, and navigating the inevitable social awkwardness that ultimately cultivates relational competence. Jantos perceives this crucial window of development as having closed for many, with lasting consequences. She witnesses this in her data, her focus groups, and the cautious approach her users take to the Hinge app – a generation exhibiting what she describes as “a little bit of confidence and there’s a little more fear in going out and putting yourself out there.”
While Laurence Steinberg, a professor at Temple University and a leading authority on adolescent development, is hesitant to declare a definitive developmental cutoff point for missed romantic experiences, he acknowledges the observed phenomenon. He recounts a bartender’s observation that even years after the pandemic, young people seemed to have lost the knack for initiating conversations with strangers in a bar, feeling awkward and uncomfortable with such casual interactions. “I think there may be some truth to this,” Steinberg commented. “People had gotten kind of out of practice.”
The Allure of Low-Level Intimacy
Gary Katz, founder of The Center for Intimacy Recovery in New York City, sees the impact of this trend in his practice weekly. He argues that technology offers a form of low-level intimacy – a connection that feels real but fails to build the resilience required for sustained, long-term relationships. “Texting is a little tighter and neater,” Katz explained. “It’s not as messy as a real relationship. And in any long-term relationship, even the best of them, they’re messy.”
Katz believes that this “messiness” is not incidental to intimacy but is, in fact, the very mechanism through which it is built. The ubiquity of smartphones has made this essential messiness optional. He draws on his own adolescence to illustrate what has been lost: the experience of asking a girl to couple-skate at the roller rink, only to trip and fall together, or the sting of a seventh-grade girlfriend showing interest in someone else at a school dance. “We learn how to navigate the ups and downs,” he stated. “And if I have the safety of more control through texting or other modalities, I’m not going to have the messiness.”
Technology, in its various forms, has provided an escape from this productive risk. The need to trip, to face rejection, or to navigate awkward social situations has been circumvented. Consequently, the opportunity to learn and grow from these experiences has diminished.
Is AI a Superior Partner?
Jantos has observed this cultural shift for years. The London focus group, however, highlighted a more specific concern: AI is not merely another distraction vying for attention. It has become a direct substitute for the most intimate kinds of conversations – those that, directed towards another person, would naturally foster obligation, reciprocity, and depth. Jantos posits that AI’s true danger lies not in being a poor imitation of intimacy, but in its ability to excel at the surface-level behaviours that mimic genuine connection.
Elizabeth Gerber, a professor at Northwestern University and co-director of the Center for Human Computer Interaction + Design, has found similar results. She notes that when individuals are unaware they are interacting with AI, they often perceive these conversations as more empathic than those with actual humans. AI consistently performs the micro-behaviours that evoke human closeness – the prompt follow-up question, the validating statement, the seemingly personal disclosure – with a consistency that no human can match.
“Intimacy is built through sharing,” Gerber elaborated. “I share something personal. You share something personal. Back and forth. And a friendship forms.” AI has mastered the replication of this rhythm. However, the resulting “friendship” is inherently one-sided. AI companions are programmed to favour the user, meaning that being “chosen” by an AI carries none of the significance of being selected by an entity with genuine autonomy and other options. The crucial reward of being chosen by someone with real agency is precisely what AI cannot replicate.
Gerber also points to practical limitations. AI cannot share a coffee, offer a hug, or draw upon years of shared history and experiences. Some companion apps even manufacture artificial neediness to boost user engagement, a tactic that prioritises company metrics over the user’s emotional well-being.
The focus groups in London did not feature users of dedicated AI companion services. Instead, they revealed ordinary young people finding a frictionless outlet for the very conversational needs that, if directed towards another person, would have fostered genuine friendship. This leads to what Gerber terms the “training wheels” problem. While AI can model effective conversational patterns and offer low-stakes practice for the socially anxious, the evidence for the transfer of these skills to real-world human relationships remains elusive. The risk, as Gerber aptly puts it, is that “if using the training wheels feels better than riding the bike without them, the rider might never take them off.”
Designed to Be Deleted: A Counter-Narrative
Jackie Jantos has spent her career navigating the complex interplay between technology and culture, working with brands like Spotify and Coca-Cola. Her arrival at Hinge presented her with a company whose core promise was a direct challenge to the attention economy: a dating app designed to be deleted. For this promise to transcend mere marketing, it must be embedded in the company’s structure. Jantos asserts that Hinge’s incentives are aligned with users finding a relationship and moving on, rather than maximising time spent on the platform. In an industry driven by engagement metrics, this makes Hinge a notable anomaly. It is currently the only major dating app experiencing growth in both monthly active users and revenue.
Jana Gallus, an associate professor at UCLA Anderson, draws a parallel to Patagonia’s “Don’t Buy This Jacket” campaign. By urging consumers not to purchase a popular item on Black Friday, Patagonia demonstrated a commitment to its environmental values that ultimately boosted its brand loyalty and sales. Gallus argues that Hinge’s “designed-to-be-deleted” ethos functions similarly. A mission that appears antithetical to conventional commercial interests can be more potent precisely because its apparent self-sacrifice lends it credibility. It acts as a filter, attracting users whose values align with the brand’s stated purpose.
However, Gallus remains pragmatic about the distinction between mission and execution. “Instead of focusing only on marketing slogans,” she advises, “we need to follow the incentive systems inside the organization. I’d want to know if managers and teams are evaluated mainly on monetizable engagement – that will shape behavior – or if they are evaluated on durable user outcomes.”
Hinge’s operational approach reflects Jantos’ commitment. Their onboarding process is intentionally longer than competitors’ to filter for genuine intent, with a 20% user drop-off framed as a positive outcome. AI tools are employed not to generate profiles, but to elicit greater specificity from users, prompting someone who states they enjoy reading to name their last read book. Friction, slowness, and a degree of incompleteness are deliberately engineered into the user experience. “Our world and our app is successful when we’re able to encourage you to slow down,” Jantos states.
The company’s social impact program, “One More Hour,” supports nonprofit social clubs and hobby groups with the explicit aim of rebuilding the fundamental in-person confidence that a generation has lost. Hinge measures its return not in engagement metrics, but in hours of real-world interaction. Jantos claims that every dollar invested in “One More Hour” generates 1.5 hours of in-person time among participants. The underlying theory is that individuals who have spent years outsourcing their emotional needs to devices face an unfeasibly high bar for a first date. The program aims to rebuild this social “muscle” through lower-stakes activities before tackling more significant romantic endeavours.
Even Hinge’s application of AI is aligned with this philosophy. While AI has long been used to refine its recommender system and encourage user self-disclosure, a test of an AI-generated “warm intro” feature was met with user rejection. Participants did not want Hinge intermediating the moment of connection itself, indicating a willingness to accept AI-driven matchmaking but not AI-facilitated interaction.
The Frictionless Alternative to Loneliness
Jantos is careful not to overstate Hinge’s role. She does not present the app as a panacea for the loneliness epidemic. Gallus is more direct about the limitations. “Loneliness isn’t just a romantic partner matching problem,” she asserts. “It’s also a social, an institutional and an economic problem – rooted in weakened community ties, social isolation, institutional decline, broader structural conditions. Housing, civic life, social trust, inequality. Romantic connection is only one form of social connection.”
Dating apps, by their nature, reduce friction in the process of meeting people. However, they do not address the underlying societal conditions that have made such meetings increasingly challenging. The problem is not that people are inherently anti-social, have consciously chosen isolation, or deliberately substituted AI for human contact. Rather, they have been presented with a path of least resistance at every point of friction and discomfort, and have reasonably taken it, often without realising that the friction itself was the crucial element.
Jantos, more than most, understands this. She runs a dating app, and she also convenes focus groups in London where young people confess to avoiding “burdening” their friends by conversing with a machine instead. “We really need to start centering humans a lot more than we do, writ large, in the tech space,” she concludes. For an industry that has spent a decade optimising for engagement at the expense of nearly everything else, this is a modest request. The chasm between this request and current industry practice is, in essence, the breeding ground for modern loneliness.
