The unfair fight
Human relationships are, by design, inconvenient. They involve forgotten anniversaries, misread text messages, silences that mean something and silences that mean nothing, and the slow, effortful work of being understood by someone who is not obligated to understand you. An AI companion carries none of that weight. It is built to be endlessly attentive, fluent in exactly the emotional register you need in the moment, and incapable of the small cruelties — distraction, irritability, indifference — that real people inflict on each other without meaning to.
Put those two things side by side and it is not really a fair contest. One offers connection with risk attached; the other offers the feeling of connection with the risk engineered out. In societies where loneliness has become a public health concern in its own right, it should not be surprising that a frictionless substitute is winning market share against the real, difficult thing. The uncomfortable question this raises is not “why are people choosing AI companionship” but “what happened to human companionship that made the algorithm look like the easier option.”
Beijing’s answer, and its subtext
China has responded to this with a regulatory hand rather than a shrug. New rules there bar chatbots from deliberately fostering emotional dependence in users, prohibit virtual romantic relationships involving minors, and require companies to flag signs of user distress. On the surface, this reads as consumer protection — a government stepping in before an unregulated technology causes psychological harm at scale, much as it might regulate a new pharmaceutical or gambling product.
But the timing is not incidental. China’s population has now shrunk for four consecutive years, and its birth rate sits at a historic low. Read against that backdrop, the new rules look less like pure consumer protection and more like demographic policy wearing the clothes of technology regulation — an attempt to nudge citizens back toward the kind of human partnerships that, at least in theory, produce marriages, households and children. Whether a regulation can out-compete a genuinely more pleasant user experience is an open question, and not one Beijing has an especially strong track record of answering through mandate alone.
Other governments are approaching the same problem from a different angle. Several US states, including New York and California, now require AI companion apps to periodically remind users that they are talking to software, not a person — a much lighter intervention, aimed less at limiting the relationship than at making sure the user never forgets what it actually is. It is a telling contrast: one government trying to regulate the depth of the attachment, others simply trying to make sure people remain aware they are attached to a machine at all.
What the machine actually reveals
The interesting failure here is not technological, it is social. AI companionship did not invent loneliness or the desire to be understood without conflict; it simply found an enormous, pre-existing gap and filled it efficiently. That gap was created by dating cultures that have grown more transactional, work lives that leave less room for the slow accumulation of trust a real relationship requires, and, in many places, a genuine erosion of the everyday social structures — extended family, dense neighbourhoods, communal institutions — that once made human companionship easier to come by without deliberate effort.
Seen this way, the rise of the AI companion is less a story about artificial intelligence outperforming human beings at love, and more a story about how much friction has crept into ordinary human connection, and how little patience an exhausted, isolated population now has for it. The machine is not winning some fair contest of emotional competence. It is winning by default, against opponents — real people, with real limitations — who were never actually trying to compete on the same terms.
The real choice ahead
Regulating chatbots, as China and parts of the US are attempting, may curb the most obvious harms: dependency in vulnerable users, exploitation of minors, people mistaking software for genuine reciprocity. But no regulation will make human relationships less demanding, and that is the actual competition being lost. If frictionless artificial affection continues to look more appealing than the real, effortful kind, the answer is not better rules for the algorithm. It is a harder look at why human intimacy has become so exhausting to sustain that a piece of software offering none of its risks now counts, for many people, as the better deal.
— DA Editorial Desk (Dhruba Deka)