Why it’s still open
Heavy users of companions are lonelier and do worse; a one-year study finds the route runs through fewer in-person contacts. But lonely people may simply turn to AI more. Trials so far last a month at most, use low doses, and do not test real commercial companion apps.
What would count as an answer
Longer randomized studies with real companion products, measuring actual contact with people rather than self-report alone.
Papers that tackle it
- Zhang et al 2026 - Living with AI companions — Sustained use predicts fewer in-person interactions.
- Fang et al 2025 - Psychosocial effects of chatbot use — More voluntary use, more loneliness and dependence.
- De Freitas et al 2025 - AI companions reduce loneliness — Relief in the moment.
- Zhang et al 2025 - The rise of AI companions — Companionship use and lower well-being.
- Common Sense Media 2026 - AI use by tweens and teens — Daily young users are lonelier.
Methods in play
- Cross-sectional surveys
- Longitudinal panel surveys
- Platform log analysis
- Randomized controlled trials
Who’s closest
Placed by the editor, v1 — corrections welcome.
Related questions
Know a paper that belongs here, or think this is framed wrongly? Suggest a paper or a correction (forms on the Atlas home).