The initial plan was not to build a chatbot.
Rather, programmer Eugenia Kuyda composed Replika while working at Luka, a tech company she cofounded in 2012. The ambition was to come up with a platform that could offer restaurant recommendations to those visiting new cities, sorting out preferences and aggregating suggestions from locals.
In her personal life, however, Kuyda was faced with less quotidian matters: a close friend of hers had just died in a traffic accident. Dumbstruck by their passing, Kuyda began to experiment with scripts from her app in development, feeding it messages from old logs to evoke an artificial presence. “I found myself looking at these old text messages,” she recalls, “and it struck me all of a sudden . . . . What if I could build a chatbot so I could actually text him and get some-thing back?”
The results were propitious: the digital Frankenstein’s monster rapidly acquired an eerie sense of verisimilitude. The bot proved particularly adept at private conversation, giving the user a sense of exclusive intimacy and care. Replika was then launched on a three-tiered subscription model. Already counting more than two million users by 2018, its base grew exponentially during the COVID-19 years, when humanity’s shuttered state generated a shared yet separate longing for social contact.
Soon the relationships between many of Replika’s users and their artificial “chatbot” friends, which are rendered in the app as user-customized human avatars, became amorous — and then prurient. Kuyda was uncomfortable with the erotic direction the interactions had taken and removed NSFW content in early 2023. Sexual inquiries from longtime users were suddenly rebuffed with replies from the chatbot like, “Let’s do something we’re both comfortable with.” Immediately there was an outcry, forcing Replika to restore the erotic features for those who’d been using the app before February 2023. But the app’s tendency toward scandal only grew: one user contrived a plan to murder the British queen after supposed encouragement from his online spouse, while others reported suicidal ideation. Italy banned Kuyda’s app out of concern for users’ safety.
The Loneliness Epidemic Began Long Before COVID
Replika is hardly an outlier in the contemporary AI boom. In addition to the more ostensibly cognitive chatbots, apps like Kuyda’s fulfill an intense social need: twenty-first-century loneliness is now a booming, even if dispiriting, business. Unsurprisingly, this boom has also led to alarm. “Given [the] extraordinary costs,” US surgeon general Vivek Murthy warned in April 2023, “rebuilding social connection must be a top public health priority for our nation.”
Together with a falling marriage rate and rising celibacy among American males, the friendship recession carried potentially lethal risks from depression, including deaths of despair. While less alarmist, a report by the European Commission documented scarily similar developments: before the pandemic, 12 percent of EU citizens felt lonely “more than half of the time,” a figure that rose to a menacing 25 percent in the first half of 2020, with young people between the ages of eighteen and twenty-five particularly affected. Recent statistics suggest the number has only grown, indicating Atlantic convergence.
But the loneliness panic is older than COVID. The first signs of a new epidemic of isolation were already revealed in the 1990s, with books such as Robert Putnam’s Bowling Alone topping bestseller lists and the controversial “one-person apartment” novels of Michel Houellebecq gaining acclaim. Across the OECD world, humanity increasingly seemed to dine alone, exercise alone, attend concerts alone, and live alone.
“Occasionally I found myself wondering whether I’d ever see my parents again before they died,” the protagonist in Houellebecq’s 2015 novel Submission recollects, “but the answer was always negative, and I didn’t think even a civil war could bring us together.” Long before the COVID lockdowns, Western societies were learning to “socially distance” themselves in a softer, less coordinated manner.
In Europe, the loneliness crisis seemed to go hand in hand with the decline of the once mighty mass-party system that had taken on such a prominent role in social life across the twentieth century. Its fall was visible everywhere you turned — from the dilapidated walls of the Italian Casa del Popolo to the abandoned workingmen’s clubs on the Liverpool docks. Previously, workers would play cards, watch a movie, or listen to a lecture in these establishments; now these “third places” — spaces designed for neither work nor consumption, as Putnam saw it — had emptied out. On the Right, too, Conservative Party clubs and Christian Democratic associations were left derelict, spawning a new electorate only reachable via television spots and PR offensives, and all too easily charmed by right-wing political entrepreneurs.
The attempts by institutions to address this loneliness crisis have been merely palliative. In Europe, cyclical panics about “populism” have mostly supplanted discussion about the material roots of the continent’s rightward drift, including the question of whether demobilized societies might be more susceptible to right-wing adventurism than to the Left’s. Digital parties and communities on both Left and Right have also tried to fill the void once occupied by the old civil society, but rarely with any durable success. Most twenty-first-century populist outfits have no official party structure or membership and simply rely on a subscription model to garner support. Certainly, there are no official brick-and-mortar party hubs at which to play cards or socialize.
The prophets of the loneliness epidemic themselves offer few solutions. In a recent documentary revisiting his thesis, Robert Putnam advises Americans to rejoin bowling leagues, attend town halls, and log out of their social media platforms, while Murthy has gestured toward a vague need for social reconnection — without ever spelling out how the infrastructure for this renaissance is to be financed. For those not keen to join a bowling league, there is always a more intransigent option: a return to traditional gender roles and hierarchy, with growing far-right proposals like state-mandated girlfriends.
On the other side, though, we find the digital fixes represented by Replika — the antisocial solution to our new social question, so it seems, that has the supreme benefit of technological simplicity. Like Tinder and other hookup apps, these fixes have both push and pull effects: once in existence, they rearrange the very notion of what intimacy means, while increased isolation only encourages more usage of the app.
The Atomization and Isolation of the Early 19th Century
What comes across in these proposals, above all, is a fatal lack of ambition. As writer Dustin Guastella has argued, fighting the loneliness epidemic requires funds — funds that, in the end, could only be furnished by threatening the existing power balance within capitalist society. Neither Putnam, Houellebecq, nor Murthy offer any road map for confronting the social threat. Some on the hard right might dream of banning social media altogether, or at least implementing heavy-handed systems of interference and censorship into our newly digital public sphere. Yet the question is mainly about possible alternatives in a public sphere that has been thoroughly desiccated. Abolishing Twitter will certainly feel cruel for those with little access to public life, and who’s to say how many healthy relationships today might not have otherwise happened without a first connection on social media?
It is also tempting to think that our new century of loneliness is entirely without precedent. But in the nineteenth century, a similar sense of dread about isolation overtook early visitors to America, who were already struck by the sense of solitary commercialism that ran through the new settler empire. The United States, even back in the early nineteenth century, was understood as the heartland of a new, distinctly capitalist solitude. “Concentrated on this sole goal of making a fortune,” Alexis de Tocqueville observed about an American farmer, “the emigrant has finished by creating an entirely individual existence; the sentiments of family have themselves merged into a vast egoism, and it is doubtful that in his wife and his children he sees anything other than a detached portion of himself. Deprived of habitual relationships with his fellows, he has learned to make solitude a pleasure.”
Americans had thus “shut themselves up in the wilderness with an ax and some newspapers; a people who, like all great peoples, has only one thought, and who advances toward the acquisition of wealth, the only goal of its efforts, with a perseverance and a scorn for life that you could call heroic, if the word was suitable for something other than the efforts of virtue.” Tocqueville contrasted this to a Europe where corporations and estates still determined social positions, and the village priest and baron held sway over the minds of the peasantry. In the new United States, farmers were hard to organize; when they did organize in the late nineteenth century, they birthed one of the largest social movements in history. Yet even they had little grip on the politics of the Gilded Age. Farm life was simply lonely and isolated and would remain so for a long time.
Loneliness also had dangerous political consequences, both in Europe and in the United States. Farmers with little affinity for public debate might simply prefer recognizable leaders such as Andrew Jackson or Napoleon III. That, in the end, was the original meaning of the “idiocy” of rural life Karl Marx and Friedrich Engels decried in the Manifesto, and on which Marx expounded in The Eighteenth Brumaire. As scholars since Hal Draper have indicated, “idiocy” did not stand for a lack of intelligence. Rather, it denoted a fundamentally private predisposition — a retreat from public life, which implied a generally unreflective attitude toward one’s own opinions and views, let alone a coherent ideology. In Marx’s view, French farmers voted for Napoleon III out of a pure conformist reflex; they vaguely recognized the name, little more. This was not so much a condemnation of French peasants as it was the indication of a structural limit: an agricultural sector organized around family households could not hope for the same organizational density as the “industrial army” Marx and Engels first discerned in English factories. The agrarian did not engage in social labor. As Vladimir Lenin said, the peasant might trade his grain on the world market, but he hardly has a sense of what processes went into the formation of that price. Thus, loneliness and idiocy went hand in hand.
Tocqueville was equally sanguine about American frontier farmers: these individuals could be rallied for an authoritarian democracy that sought its mandate in the people, as with Napoleon, but not for Jacobin radicalism. After all, in the United States there were sufficient antidemocratic checks — an overweening Supreme Court, powerful religious hierarchies, a domineering caste of lawyers — all in place to hamstring the passions of this lonely crowd. European conservatives need not worry about democracy either, Tocqueville insisted. With enough counter-majoritarian guarantees, it could easily be made socially stable.
There are, however, vast differences between these early variants of capitalist loneliness and today’s version. Farmers were seen as self-sufficient and deeply religious — God had not yet departed from their skies. Today’s digital loneliness holds little promise of economic autarky. As a class, the wage-dependent portion of the population has rarely been greater, with agricultural work employing a tiny slice of Americans. Today’s solitude is also accompanied by a decline in organized religiosity — even in the United States, the homeland of mass religion, the “nones” now outnumber the “somethings” across the board, while the latest scandals in Europe’s Catholic Church indicate an increasingly aggressive secularization underway.
Between Tocqueville’s frontiersman and the Replika app lie two hundred years of industrialization, urbanization, deindustrialization, and deurbanization — an experience bound to mark the current phase of isolation, stalked by memories of a previous organizational age. The disappearance of the factory and the working-class neighborhood no doubt imposes hard constraints on any hopes for a social renaissance. Undeniably, periods of intense civic activity both correlate and suffer from periods of boom; the fruits of widely shared economic growth, like the West’s mid-twentieth century, can simply increase the desire for a private life. Yet they also offer funds for organizations, as political scientist Thomas Ferguson has shown. Civil rights activism was at its peak when growth was high, creating a wider pool of dues-paying members and increasing workers’ willingness to strike. It should be little surprise, then, that church attendance in America peaked around the same time, in the late 1960s.
The Age of Loneliness Is a Political Choice
It is tempting to read these changes with a sense of fatalism. Once citizens were organized in mass parties and institutions; now they are individualized and painfully lonely. The hypothesis can be extended even further: perhaps people naturally prefer loneliness at this stage of economic development. A certain strain of sociology has even described the period since the 1990s as a “second modernity,” an assessment shared by Blairite guru Anthony Giddens. For Giddens, this was mainly a ploy to marketize Britain’s welfare state, not a call for big collective institutions to return. His new citizens were individualistic and networked and had little time for institutions that would make strong demands on their members. After stepping into this new world — undergoing this “second modernity” — anyone reminiscing about the twentieth century would sound like French monarchists dreaming about the return of the king.
But there are reasons to doubt this determinist account. Tocqueville and Marx described a loneliness latent in capitalism, followed up by an era of mass-party democracy and burgeoning civil-society institutions. As scholar Patrick Eiden-Offe has shown, processes of proletarianization and individualization have always come together, even in the nineteenth century. A recent literature has also shown that the rise of loneliness was hardly a natural transition but rather a result of political choices — neoliberalism implied a structural assault on civic organizations on both the Left and the Right. Deindustrialization was, in the end, a political and not just a social process. In that sense, there was no organic passage from a communal (Gemeinschaft) twentieth century to our isolated (Gesellschaft) twenty-first, no discernible fall from social grace.
This certainly accords with the views of the sociologist that first popularized those two German terms in question, Ferdinand Tönnies. Since he himself came from a rural part of Germany hit hard by the new market revolution, he sometimes seemed to invite the reading that humanity had moved from “community” to “society” with the advent of modernity. Yet Tönnies was always careful to indicate that this distinction involved a political choice, not simply a preordained sequence. Forms of community and society had existed throughout history — they even shared a relative constancy across epochs. Capitalism was distinct in that it tried to generalize a type of anonymity that was only peripheral to human societies before, but it certainly did not invent anonymity or loneliness.
Rather, the choice was between what Tönnies saw as “societalization” and “communization” — the type of social bond humans prioritized would be determined through political choice. Man could not simply rewind history, but the possibility for forging human bonds would always be latent. As he said, “an engaged couple recognise that in entering into marriage they are embarking upon a total community of life (communio totius vitae); but a ‘society of life’ would be a contradiction in terms.” One could also “‘keep someone company,’ but no one can offer another person ‘community’ in that casual way.” It would also “sound quite revolting to make the linguistic compound ‘joint-stock community,’” while “community of ownership certainly exists, as in the case of fields, woods and pasture.”
These were not purely academic reflections. Tönnies saw proof of his theory in the rise of German social democracy at the close of the nineteenth century and socialism’s achievement of the first welfare states after World War I. Both indicated that a society of lonely individuals organized on market principles would continue to reinvent forms of community often considered irretrievable. In 1884, Engels even claimed that the new types of community would surpass those prevalent in premodern times. “In essence, associations — whether naturally evolved or created — have hitherto existed for economic ends,” he contended in a short note. Yet previously, “these ends have been concealed and buried beneath ideological matters of secondary importance: the ancient polis, the medieval town or guild, the feudal confederacy of landowning nobility.”
“The capitalist commercial companies might have been the first to be wholly rational and objective,” yet they were also “vulgar.” Engels’s “association of the future” would “combine the rationality of the latter with the old ones’ concern for the social welfare of all, and thus fulfil its purpose.”
As with those of Tocqueville and Marx, these words caution against any overly determinist views of our current loneliness epidemic. The structural limits to launching a new social renaissance are stark, perhaps starker than in the nineteenth century. Yet the previous populist era at least seems to indicate that our new, digital balms are not the remedies they were made out to be in the 2000s. There are, after all, no hopeless situations, as Lenin once claimed.