A book asks whether AI can love and heal: the empathy it simulates is not the bond that cures

By CIONoticias · September 8, 2026.

"Can artificial intelligence love, desire and heal?" has been published, an essay that starts from an uncomfortable question: when someone tells a chatbot about their worries, asks it for advice about a relationship or seeks companionship in a moment of loneliness, are they receiving something resembling a therapeutic bond or merely a response that sounds empathetic? The source does not detail the book's content beyond that question, nor does it identify its author.

It is a distinction worth separating out from the start, because public debate tends to blur it. One thing is a system's ability to generate language that mimics emotional understanding —something current models do with ease— and quite another is the existence of a therapeutic relationship in the clinical sense: continuity, professional responsibility, the capacity to intervene when there is real risk. It is a question best raised from psychology, not from product marketing, which is precisely the terrain where rigor is most lacking.

As sector context, the question does not arise in a vacuum. AI-driven companionship and emotional wellbeing apps have been growing for several years, and quite a few users —teenagers and young adults especially— use them daily to talk about anxiety, a breakup or simply to avoid feeling alone. Regulators and professional bodies in various countries have already begun examining what happens when that use replaces, rather than complements, care from a human professional in a serious case. The original article provides no figures or studies of its own on this, so we leave it as framing, not as verified data.

Our reading is that a very specific version of the clash between short and long term that runs through the entire AI conversation is being played out here. In the short term, the risk is real: a simulation of empathy with no clinical judgment behind it can generate parasocial attachment, delay the search for real help or give a false sense of therapeutic progress to someone who needs more than well-calibrated words. Ignoring that would be irresponsible, and it is precisely what a book like this tries to put on the table before the market settles it by default.

But the long horizon is different, and that is where it connects with the underlying thesis we hold at Zendoric: the abundance AI can bring is not only material. An assistant that never tires, available at any hour and with no marginal cost, can broaden access to a first level of emotional support in places where today there are neither enough therapists nor the budget for them. The key is for that support to be designed to complement the human bond —referring, alerting, accompanying— and not to simulate it to the point of passing as a substitute. The technology already knows how to imitate the form of empathy; the outstanding work, both clinical and regulatory, is making sure that is not mistaken for the substance.