Relationships & loneliness·Overview
A Longitudinal Randomized Control Study of Companion Chatbot Use
arXiv, 2025
The more the bot is someone, the more of yourself you give it.
A result contrary to the rest of the group, kept in the library on purpose: a 21-day randomised study of companion chatbot use found no overall effect on social health or loneliness. A library that drops inconvenient findings is advocacy, not research.
More interesting than the null result is what the researchers found beneath it: effects on human relationships were mediated by anthropomorphisation. The more a user experienced the bot as someone — a person, not a program — the greater the transfer: what they gave the bot was subtracted from relationships with people. The effect depended not on use itself but on who the bot became in the user’s eyes.
For personal use the lesson is subtler than “chatbots harm”: three weeks of moderate contact need not change anything, but the moment the bot becomes someone is the moment it starts to cost. Caveat: a short horizon; the entries with one- and two-year follow-up show what happens next.
- Year
- 2025
- Original language
- EN
- Source
- arXiv, 2025
An overview in the EPIQ team’s own words, not a translation. The full text remains with the publisher — link below. Quotations are short, verbatim and attributed, within the right of quotation.