This is not another argument about artificial intelligence. Nor is it another argument about mental health.

 

They have become the same conversation.

 

For much of modern history, identity and belonging were organised through institutions: families, neighbourhoods, schools, religious communities, unions, workplaces and public services. These institutions were often imperfect, exclusionary and coercive. But they gave human life structure. They distributed meaning, recognition, obligation and care.

 

Many of them have weakened. The need they answered has not.

 

Into that space came the platform.

 

Commercial technologies no longer merely facilitate communication. They organise the conditions under which people encounter one another, understand themselves and decide what deserves attention. Friendship is filtered through visibility. Status is quantified. Memory is outsourced. Recognition arrives as data.

 

What appears to be social life is also infrastructure.

 

Artificial intelligence extends this process into more intimate territory. Systems are being designed to listen, advise, reassure, interpret emotion and respond to distress. They are entering spaces once occupied by friends, teachers, clinicians, communities and the private work of reflection.

 

This transition is often described as innovation in access. Sometimes it is. A responsive system may be available when no person or service is. It may offer language, information or temporary relief.

 

But availability is not the same as care.

 

Care involves obligation. It creates responsibility between people and within institutions. A commercial system can simulate attentiveness without assuming responsibility for what follows. It can produce the feeling of being understood while converting the interaction into data.

 

The danger is not simply that AI will provide bad advice. It is that commercial platforms will become the default architecture of emotional life because other forms of support have been allowed to disappear.

 

We will then diagnose the consequences psychologically: loneliness, anxiety, dependency, fragmentation. Yet their origins will be political and economic.

 

A person turns to a machine because no one else is available. The system records engagement. The market calls this demand.

 

This is how structural abandonment becomes a product opportunity.

The central question is therefore not whether AI can imitate empathy convincingly. It is what happens when access to recognition, belonging and emotional support is increasingly organised by institutions whose obligations are to growth, prediction and profit.

 

Mental health cannot be separated from the systems through which people are required to live. And AI ethics cannot be confined to technical performance.

 

The future of the inner life is becoming a question of infrastructure.

 

We should ask who is building it, who owns it and what they are entitled to take in return.