The Friend Machine

The Friend Machine sits at the intersection of several conversations that are often kept apart.

There is the technological conversation and the squabbling between those in charge of it (accelerationists vs. de-accelerationists, for example), about large language models and conversational AI. There is the clinical one, about loneliness, depression, and access to care. There is the cultural one, about how we form and sustain relationships in a world of platforms, precarity, and constant mediation. And there is the economic one, which is rarely stated plainly: Connection has become a market opportunity.

We are watching the emergence of what might be called a synthetic intimacy industry. Chatbots that position themselves as companions, therapists, romantic partners. Services that promise presence, understanding, even love. Some of these are explicitly marketed that way. Others drift into that role because of how people use them. Either way, they are filling gaps left by overstretched health systems, fragmented communities, and the erosion of shared social spaces.

My work tries to hold two truths at once. These systems can provide real, immediate relief. They can help someone get through a night, a grief, a moment of crisis. But they are also designed environments. They are optimized to be engaging, affirming, and sticky. That has implications for how we learn to relate, and what we come to expect from one another.

I came to this through a long-standing interest in intimacy and technology. I began, years ago, with a fictional idea about a bot that falls in love with a human. At the time, it felt speculative, almost whimsical. Then I started encountering real people forming attachments to conversational systems, sometimes deeply so. At the same time, loneliness was being named at the level of public health.

The turning point was realizing these were not separate stories. They were the same story.

What happens when a crisis of connection meets a technology that can simulate its cure?