Lonelier US college students more likely to use AI chatbots for company, survey of 858 finds

The pattern suggests companionship-seeking is a distinct way of interacting with AI chatbots rather than a by-product of heavy general use. In a model comparing non-users, general-only users and companionship users, loneliness separated companionship users from the others and predicted more frequent companionship use in a dose-dependent pattern.

Neither campus belonging nor satisfaction with social relationships predicted companionship use. Asian and international students were more likely to use AI this way, while LGBQ+ students were less likely. The data are cross-sectional, so the authors treat loneliness as a candidate predictor to be tested in longitudinal research.

Researchers used hierarchical logistic regression on Healthy Minds Study 2024-2025 responses to test whether loneliness uniquely predicted AI companionship use after adjusting for demographics and mental health symptoms. They ran the same test for general AI use to see whether any link was specific to companionship.

Loneliness was associated with companionship use but not general use, which the authors read as specificity to companionship-seeking. A combined model contrasting non-users, general-only users and companionship users gave the same result: loneliness distinguished the companionship group (p < 0.01).

Two measures of social connection, campus belonging and satisfaction with social relationships, did not predict companionship use. Differences by group appeared: Asian and international students used AI for companionship more often, and LGBQ+ students less often.