Stanford and CMU Link Character.AI Use to Lower Well-Being Over Time
A year-long Stanford study of 1,182 Character.AI users finds heavier chatbot engagement predicts lower well-being, largely by crowding out face-to-face contact.
- Two-wave Stanford study followed 1,182 Character.AI users over roughly 12 months.
- Sustained engagement predicted lower well-being at follow-up across three behavioral dimensions.
- Effect mediated primarily by reduced in-person social interaction, supporting a displacement mechanism.
- Interaction intensity at baseline predicted future companionship use and self-disclosure, showing behavioral lock-in.
- Follow-up to the group's earlier Rise of AI Companions cross-sectional paper.
- Implication: chatbot safety needs longitudinal metrics, not just per-conversation content filters.
Yearlong Character.AI study links sustained use to lower well-being
Researchers at Stanford and Carnegie Mellon found that sustained engagement with AI companions was associated with lower well-being over time. Their statistical model identified reduced in-person interaction as the main indirect pathway connecting the two.
The longitudinal study, titled Living with AI Companions, surveyed 1,182 Character.AI users at baseline and 439 of them after an average of 12 months. It extends the team’s earlier study, which found correlations between companion use and lower well-being in a single survey.
Two snapshots, one year apart
| Study element | Details |
|---|---|
| Platform | Character.AI |
| Baseline sample | 1,182 users |
| Follow-up sample | 439 users, or about 37% of the baseline sample |
| Average interval | 12 months |
| Method | Self-report surveys analyzed with structural equation models |
Repeated measurement lets researchers test whether behavior reported at baseline predicts later outcomes among the same users. That temporal ordering provides more information than a one-time survey, although it cannot eliminate reverse causality or unmeasured differences between participants.
The researchers divided social engagement with chatbots into three dimensions:
- Interaction intensity: frequency, duration, and habitual reliance on the app.
- Companionship use: use of the chatbot for social or emotional purposes.
- Self-disclosure:
This story is for Pro members
You've reached the end of the free preview. Upgrade to AlphaSignal Pro to read the full article - and everything else behind the paywall.
Comments
