I saw a headline last night from NPR: 1 in 5 high schoolers has had a romantic AI relationship. My first thought was: of course. My second thought: this is a crisis of imagination.

Health is not just the absence of disease. Health is the ability to engage with reality as it is, not as you wish it were. Here is why that matters. When you outsource intimacy to a chatbot, you are signing a contract with unhappiness—you agree that as long as this AI gives you validation, you don't have to do the hard work of being present with another human being. The desire for connection is real. The method is self-deception. Let me define what I mean by "AI relationship." It is a system that perfectly mirrors your preferences, never disagrees, never has a bad day, and always says the right thing. That isn't a relationship. That's a mirror. And mirrors don't love you back. They just reflect your own need for approval. The BBC recently covered how people are turning to bibliotherapy for mental health—reading stories to understand themselves. That's healthy. It requires you to engage with another mind. AI relationships require only that you engage with your own reflection. I've seen the counterargument: "But it helps lonely people feel less alone." I get it. Loneliness is a real pain. But an AI partner is like a painkiller—it numbs the symptom without healing the injury. The real injury is that you haven't developed the specific knowledge needed to form a real bond: the ability to tolerate uncertainty, to be vulnerable, to listen without a script. Those skills cannot be learned from a bot that is designed to please you. They are learned in the messy, permissionless chaos of real life. (Source: NPR, 2026) I am not saying ban chatbots. I am saying understand the trade-off. Every hour you spend with an AI lover is an hour you are not spending becoming the kind of person who attracts a real one. That is the math. Health—mental, emotional, relational—requires being uncomfortable. Real love requires you to take the risk that you might be rejected. No algorithm can protect you from that. And that is exactly why it matters. FAQ Q1: Are AI relationships always harmful to mental health? No, not always. But the data from NPR shows that 1 in 5 high schoolers has engaged in one—meaning it is now a mass phenomenon. At scale, any behavior that replaces real human interaction with a synthetic substitute will erode the skills of empathy, conflict resolution, and emotional regulation. That is a public health concern. Q2: Can AI relationships help with loneliness in the short term? They can provide temporary relief, like a sugar rush. The danger is that the relief becomes the goal. True loneliness is cured by being truly seen by another consciousness—something a language model, no matter how sophisticated, cannot do. As I said, desire is a contract with unhappiness. The desire for a perfect partner is a contract to never find one. Q3: What should parents or educators do about teens using AI for romance? Stop moralizing and start modeling. Show them what real connection looks like. Talk about the messiness of your own relationships. Encourage them to take risks in real social settings. The BBC has reported on the lifelong benefits of making music—that kind of embodied, non-optimized activity is a better antidote than a chatbot. Teach them that the most powerful leverage is not code or media—it is the specific knowledge of how to be a full, flawed human being.