A.I. Is About to Solve Loneliness. That’s a Problem
There’s real reason for caution here, starting with the idea that interactions with A.I. can be treated as genuine relationships. Oliver Burkeman exasperatedly writes that, unless you think the L.L.M.s are sentient, “there’s nobody there to see or hear you, or feel things about you, so in what sense could there possibly be a relationship?” While drafting our article “In Praise of Empathic A.I.,” my co-authors (Michael Inzlicht, C. Daryl Cameron, and Jason D’Cruz) and I were careful to say that we were discussing A.I.s that give a convincing impression of empathy. But A.I. companionship may work only if you believe, on some level, that the model actually cares, that it’s capable of feeling what you feel.
If future language models do achieve consciousness, then the problem vanishes (and new, more serious ones take its place). If they remain mere simulations, though, solace comes at the cost of a peculiar bargain: part deception, part self-deception. “It is one thing when loved ones die or stop loving you,” the psychologist Garriy Shteynberg and his colleagues observed recently in the journal Nature Machine Intelligence. “It is another when you realize they never existed. What kind of despair would people feel upon the discovery that their source of joy, belonging, and meaning was a farce? Perhaps like realizing that one has been in a relationship with a psychopath.”
For now, the line between person and program is still visible—most of us can see the code beneath the mask. But, as the technology improves, the mask will slip less and less. Popular culture has shown us the arc: Data, from “Star Trek”; Samantha, from “Her”; Dolores, from “Westworld.” Evolution primed us to see minds everywhere; nature never prepared us for machines this adept at pretending to have them. Already, the mimicry is good enough for some—the lonely, the imaginative. Soon, it may be good enough for almost everyone.
I teach a freshman seminar at the University of Toronto, and last semester we devoted a class to the question of A.I. companions. My students, by and large, sided with the critics. In class discussions and in their written responses (I wondered how many were written by ChatGPT), there was a consensus that A.I. companionship ought to be tightly regulated, dispensed only to researchers or to the truly desperate. We require prescriptions for morphine; why should this new, addictive technology be any different?
I doubt my students will get their way. Perhaps A.I. companions will plateau, the way self-driving cars seem to have done. Still, if the technology does advance, it’s unlikely that we’ll tolerate strict government controls indefinitely. The appetite for these companions may simply prove too strong.
So what kind of world will we inhabit when A.I. companionship is always within reach? Solitude is the engine of independent thought—a usual precondition for real creativity. It gives us a chance to commune with nature, or, if we’re feeling ambitious, to pursue some kind of spiritual transcendence: Christ in the desert, the Buddha beneath the tree, the poet on her solitary walk. Susan Cain, in her book “Quiet,” describes solitude as a catalyst for discovery: “If you’re in the backyard sitting under a tree while everyone else is clinking glasses on the patio, you’re more likely to have an apple fall on your head.”
But solitude isn’t loneliness. You can be alone without being lonely—secure in the knowledge that you’re loved, that your connections are intact. The reverse is possible, too. Hannah Arendt once observed that “loneliness shows itself most sharply in company with others.” It’s bad enough to be alone on Valentine’s Day; it’s worse, somehow, to find yourself surrounded by canoodling couples. The most acute loneliness, I suspect, is the kind you feel in the presence of those you love. I remember, years ago, sitting in my living room with my wife and our two-year-old as they both refused to speak to me (for different reasons). The silence was almost physically painful.
It’s easy to think of loneliness as simply a lack of being respected, needed, or loved. But that’s not the whole story. The philosopher Olivia Bailey suggests that what people crave, above all, is to be “humanely understood.” Empathy, in this light, is not just a way of feeling but a way of caring—a willingness to try to understand the particularity of someone else’s emotions.
That sort of understanding, as most of us learn, can be in surprisingly short supply—not only because others don’t care enough to try but because sometimes there’s a gap that just can’t be bridged. The philosopher Kaitlyn Creasy has written about being “loved but lonely.” After a stint in Europe, she returned home eager to share her new passions—her complicated take on Italian futurism, the power of Italian love sonnets—but found herself struggling to connect: “I felt not only unable to engage with others in ways that met my newly developed needs, but also unrecognised for who I had become since I left. And I felt deeply, painfully lonely.”
Creasy sees this kind of missed connection less as a personal failing than as an existential hazard. “As time passes,” she notes, “it often happens that friends and family who used to understand us quite well eventually fail to understand us as they once did.” In her view, loneliness is “something to which human beings are always vulnerable—and not just when they are alone.” Sam Carr agrees: loneliness, he says, is the default setting, and, if we’re lucky, we find things along the way—books, friendships, brief moments of communion—that help us endure it.
Maybe the closest most of us ever get to an absence of loneliness is at the start of a love affair, when both people are hungry to know and be known. But that’s only the prospect of understanding, not the achievement of it. Sooner or later, even that feeling fades.
If A.I. companions could truly fulfill their promise—banishing the pain of loneliness entirely—the result might feel blissful, at least at first. But would it make us better? In “A Biography of Loneliness,” the cultural historian Fay Alberti sees value in at least the fleeting kind of loneliness that you encounter during life transitions—“moving away to university, changing jobs, getting divorced.” It can, she says, “be a spur to personal growth, a way of figuring out what one wants in relationships with others.” The psychologist Clark Moustakas, in “Loneliness,” takes the condition to be “an experience of being human which enables the individual to sustain, extend, and deepen his humanity.”
Most obviously, loneliness could go the way of boredom. I’m old enough to remember when feeling bored was just a fact of life. Late at night, after the television stations signed off, you were on your own, unless you had a good book or a companion around. These days, boredom still visits—on planes without Wi-Fi; in long meetings—but it’s rare. Our phones are never far, and the arsenal of distractions has grown bottomless: games, podcasts, text threads, and the rest.
This is, in some ways, an obvious improvement. After all, no one misses being bored. At the same time, boredom is a kind of internal alarm, letting us know that something in our environment—or perhaps in ourselves—has gone missing. Boredom prompts us to seek out new experiences, to learn, to invent, to build; curing boredom with games like Wordle is a bit like sating hunger with M&M’s. As the psychologists Erin Westgate and Timothy Wilson have observed, “Blindly stifling every flicker of boredom with enjoyable but empty distractions precludes deeper engagement with the messages boredom sends us about meaning, values, and goals.” Maybe the best thing about boredom is what it forces us to do next.
In a similar way, loneliness isn’t just an affliction to be cured but an experience that can shape us for the better. John Cacioppo, the late neuroscientist who pioneered the science of loneliness, described it as a biological signal, akin to hunger, thirst, or pain. For most of human history, being cut off from others wasn’t merely uncomfortable; it was dangerous. From an evolutionary perspective, isolation meant not just the risk of death but, worse, the risk of leaving no descendants.
In this sense, loneliness is corrective feedback: a nudge, or sometimes a shove, pushing us toward connection. Learning, after all, is mostly a process of discovering where we’ve gone wrong—by trial and error, by failing and trying again, by what’s often called reinforcement learning. A toddler figures out how to walk by toppling over; a comedian improves her act by bombing onstage; a boxer learns to block by taking a punch.
Loneliness is what failure feels like in the social realm; it makes isolation intolerable. It can push us to text a friend, show up for brunch, open the dating app. It can also make us try harder with the people already in our lives—working to regulate our moods, to manage conflict, to be genuinely interested in others.
