Establish Causality Between Chatbot Companionship Use and Mental Health

Establish the causal relationship between companionship-oriented use of AI companion chatbots and users' psychological well-being by determining whether companionship-motivated chatbot engagement causally affects well-being or whether lower well-being leads to increased companionship use, accounting for residual confounding and reverse causation.

Background

The paper analyzes cross-sectional survey data and donated chat histories from Character.AI users to examine associations between companionship-oriented chatbot use and well-being. Because the data are observational and collected at a single point in time, residual confounding and reverse causation cannot be ruled out.

The authors explicitly note that they cannot establish causality and propose that longitudinal or experimental designs are needed to determine the direction of effects between companionship-focused chatbot engagement and psychological well-being.

References

First, the cross-sectional nature of our data, as well as the possibility of residual confounding or reverse causation means we cannot establish causality between chatbot companions use and mental health.

— The Rise of AI Companions: How Human-Chatbot Relationships Influence Well-Being (2506.12605 - Zhang et al., 14 Jun 2025) in Subsection "Limitations" (under Discussion)