Flourish, a generative AI app for well-being support, helped some measures move in a better direction among college students. The study also found clear limits. Gains were modest and short-term. There were no significant changes in depression, anxiety, or stress.

What happened

Cachia and colleagues recruited 486 undergraduates from three US post-secondary institutions. Researchers randomly assigned students to Flourish access or their existing support resources. The app offered AI coaching, emotion check-ins, personalized well-being activities, and summaries of mood patterns. Students in the app group were asked to use it at least twice each week.

What the study measured

Students completed validated questionnaires at week 0, week 2, week 4, and week 6. The measures covered positive affect, resilience, loneliness, belonging, mindfulness, and flourishing. Social well-being here means reported feelings such as belonging and loneliness. It does not count a student’s friends or real-world conversations.

Of the 486 participants, 320 completed all four surveys. Students assigned to Flourish reported 3.49 active app days each week. That number describes use, not a guaranteed health benefit.

What changed

Across the six-week period, the Flourish group had more favorable trajectories for resilience, loneliness, belonging, mindfulness, and flourishing. At week six, positive affect was higher in the Flourish group. The scores were 19.38 versus 17.21. Mindfulness scores were 14.99 versus 13.67. The effect sizes were modest: 0.33 for positive affect and 0.22 for mindfulness.

The picture became less decisive at the final comparison. Week-six differences for resilience, loneliness, and flourishing were not statistically significant. Changes in depression, anxiety, and stress were also not significant. The study therefore does not show that the app treated a mental illness or eliminated loneliness.

Why the result matters

Many young adults face barriers to professional support. A personalized app could offer a low-barrier way to notice emotions and practice small well-being activities. This study is useful because it compared randomly assigned groups over repeated surveys. It was not only a satisfaction poll from people who chose the app.

Still, a better questionnaire score is not the same as stronger relationships. The study cannot show whether students made more friends, talked to people more often, or kept the gains after the app period. It also cannot show whether the AI itself caused every change. Other experiences during six weeks may have mattered.

What is known—and what is not

The evidence comes from a randomized study with 486 US undergraduates and repeated, validated measures. It suggests modest benefits in positive affect and mindfulness, with several social well-being measures moving favorably over time.

The limits are equally important. The follow-up lasted six weeks. Attrition meant that only 320 students completed every survey. The study had no active comparison app, and participants were not selected for clinical symptoms. Results may not apply to older adults, younger children, or people seeking treatment for serious distress.

What to watch next

Future studies should follow people for longer and include more varied ages and mental health needs. They should compare AI apps with human support, other apps, or both. Researchers should also measure real-world social behavior, not only feelings reported on a questionnaire. For now, Flourish looks like a promising support tool, not a proven replacement for mental health care.

Sources: 2 Minute Medicine and NEJM AI