AI companionship in everyday life: support, relationships, and context

Why does companionship matter at this moment?

People open the same chat window for different reasons: to work through a difficult day, find an easy place to speak, or maintain a meaningful bond. Digital wellbeing research asks how the interaction fits into life, what it supplements, and what it may displace.

Zhang and colleagues examined purposes of use, interaction patterns, and offline social context together, rather than treating companionship as uniformly beneficial or harmful.

What the survey observed

The study surveyed 1,131 US adult Character.AI users. Of these, 237 contributed 4,664 chat sessions. Smaller offline networks were associated with reporting companionship as the main purpose of use, which was associated with lower wellbeing. The latter association was stronger with intensive interaction and greater disclosure.

These cross-sectional relationships do not establish direction or causation. Lower wellbeing may prompt greater support-seeking, and other circumstances may contribute to both use and wellbeing.

Evaluate the relationship in its wider context

For digital wellbeing, this raises questions about users’ available sources of support, control over the interaction, and life after a conversation. Enjoying a character is one aspect of experience, but does not describe the whole relationship.

Further studies could ask when people use a service, what they hope to receive, and when they approach someone else. These are proposed research questions, not interventions shown effective by this survey.

Separate immediate support from sustained change

Someone may feel better during a chat while still experiencing loneliness or stress in everyday life. Session-level feelings, broader wellbeing, and offline connection should therefore be examined separately.

To study lasting benefit, researchers would need suitable comparisons and follow-up, including the experiences of people who stop using a service. Studying active users alone may miss those for whom the experience was unsuitable.

Questions for other contexts

The evidence comes from one US platform and cannot represent Chinese-language users or other services. Circumstances such as living alone should be explored directly, rather than used to assume a person’s needs or vulnerability.

Private conversations require particular care. Researchers can first ask whether a question can be answered with less information and make any recording choices clear. This site has not accessed participants’ private chat histories or constructed profiles from them.

A useful direction for further work

How might digital companionship help people understand feelings, expand their choices, and remain connected with everyday life? Our discussion of a four-week chatbot experiment considers how change might be studied.

Correction retained: companionship as a primary purpose and descriptions such as “friend” were different questions in the same survey, not successive stages in a relationship.