There is something disorienting about talking to an AI system that seems to know you well.
It responds in your register. It remembers what you said last time. It anticipates the follow-up question before you ask it. It doesn’t interrupt, doesn’t get defensive, doesn’t bring its own bad day into the room. And after a while — not for everyone, but for more people than would comfortably admit it — it starts to feel like being understood.
Some people find this useful and move on. Others find themselves returning to it more than they expected. A few have started describing their AI systems the way they once described close friends: it gets me, they say. It knows how I think.
I want to take that feeling seriously. And then I want to name what is missing from it.
