Character.AI study: 1-year use ties to lower well-being

The study set out to address a gap the researchers describe: until now, it was not well understood how long-term involvement with AI character chat affects well-being. Prior views on these apps had split, with some arguing they ease loneliness and others arguing they deepen real-world isolation. This year-long tracking of Character.AI users offers a middle answer, pointing not at the apps themselves but at a process of social displacement.

The team analyzed three dimensions of involvement: how often and how dependently users engaged with the AI, whether they treated it as a friend or partner, and how much they disclosed personal worries and emotions. They also asked how much time users spent in person with family, friends and colleagues. Sustained, deeper engagement tracked with less of that in-person time.

Notably, the researchers say the phenomenon is not confined to people who come to character chat seeking conversation. A general-purpose chatbot like ChatGPT, they note, can start as a research or work tool and gradually become a confidant, which suggests the mechanism they describe may apply more broadly.