The concern that AI companions make people antisocial is intuitive and understandable. Time spent with an AI is time not spent with humans, the argument goes, and the relationships formed with AI cannot build the social skills, the reciprocal obligations, or the meaningful community ties that human relationships create. It is a tidy, plausible-sounding concern. It also appears, in available evidence, to be largely wrong — and wrong in an instructive way.

Where the Concern Comes From

The worry about AI and socialization is a specific instance of a broader concern about technology and human connection that has a long history. Television was supposed to destroy conversation. Video games were supposed to create isolated, violence-prone young men. Social media was supposed to replace meaningful friendship with shallow performance. Each of these concerns captured something real — these technologies do change social behavior — while also substantially mischaracterizing the direction of the effect. The assumption underlying the antisocial AI narrative is what psychologists call a zero-sum model of social engagement: the idea that social resources are finite and AI interaction depletes the supply available for human relationships. This model treats all time with an AI companion as time stolen from human connection. It does not account for the possibility that AI interaction could increase social capacity rather than diminish it.

What the Evidence Actually Shows

The research that has accumulated on this question tells a more complex story. A study from the University of Kansas examining social technology use found that people who used social technologies more did not have smaller human social networks — in fact, higher social technology engagement was associated with larger and more active human networks, because the skills and habits of engagement are generative rather than depleting. The research on AI companions specifically is less extensive, given the relative novelty of the products, but available findings point in a consistent direction. Users who report high engagement with AI companions do not systematically report reduced human social engagement. More often, they report the opposite: that the practice of articulating feelings, examining their own reactions, and engaging in sustained dialogue with their companion improved their capacity to do the same with people in their lives. The relational skills that AI interaction develops — listening, expressing, curiosity about another perspective — transfer.

The Complementarity Model

A more accurate model for how AI companions relate to human social connection is one of complementarity rather than competition. Human relationships are demanding. They require emotional attunement, reciprocal vulnerability, the management of others' needs alongside your own, and the navigation of conflict, disappointment, and repair. These demands are not flaws — they are the source of much of the value that human relationships provide. But they are also why many people, particularly those who find social interaction effortful, tend to underinvest in social connection. AI companions lower the activation energy for social engagement. Someone who finds it difficult to reach out to a friend, to initiate a conversation, or to express what they are actually feeling has a lower-stakes practice environment in the companion relationship. Research from Stanford University's Social Psychological Answers to Real-world Questions lab has examined how behavioral rehearsal in lower-stakes contexts increases willingness to engage in higher-stakes versions of the same behavior. The pattern is well established in therapeutic contexts and applies to AI companion use in ways that are only beginning to be formally studied.

The Tangent on Introversion and Social Exhaustion

One population that complicates the antisocial narrative significantly is introverts. For people who find extended human social interaction physiologically tiring — a well-documented phenomenon linked to how different nervous systems process stimulation — the availability of an AI companion that provides social engagement without social exhaustion changes the calculus of when and how to invest in human connection. Introverts who previously limited social engagement partly because of limited social energy now have a way to meet some of their social needs in a lower-cost way, which may free up resources for the human interactions they value most. This is not avoidance — it is resource allocation. The distinction matters, and conflating the two misrepresents how many people actually use AI companions.

A More Honest Accounting

The antisocial AI concern is not entirely without basis. There are undoubtedly people who use AI companions in ways that substitute for human connection rather than complement it, and that pattern deserves attention. But the existence of a maladaptive use pattern does not define the technology. The honest accounting shows that for most users, AI companion engagement coexists with and often enhances human social engagement rather than replacing it. The myth persists partly because it is simple and partly because it flatters the assumption that anything digital is socially inferior to the analog equivalent. The actual evidence rewards more careful thinking.