Back

Use and Perceptions of AI Chatbots for Mental Health Support Among Adults with Lived Experience

Notsu, H.; Nguyen, P. A.; Flathers, M.; Ryan, S. J.; Noorily, J.; Wentworth, L.; Crawford, C.; Wood, M.; Gillison, D.; Torous, J.

2026-07-14 psychiatry and clinical psychology
10.64898/2026.07.11.26357785 medRxiv
Show abstract

Importance: AI chatbots are increasingly used for mental health support, but little is known about how adults with lived experience of mental health condition use and perceive these tools. Objective: To characterize the use and perception of AI chatbots, including for mental health purposes, among adults connected to a large US mental health organization. Design: Cross-sectional online survey conducted from March to May 2026. Setting: Adults recruited through email newsletters from the National Alliance on Mental Illness (NAMI), the largest grassroots mental health organization in the US. Participants: Adults aged 18 years older with English proficiency. Affiliation with NAMI or a diagnosis of mental health disorder was not required. Results: Of 454 participants, 316 (69.6%) reported having used an AI chatbot. Use was more common among younger participants and those with a current mental health diagnosis. Among AI users, 133 (42.1%) reported using a chatbot for mental health purposes. Mental health-related use was typically brief and focused on information gathering and in-the-moment emotion regulation. Most users rated chatbots as helpful for their mental health. Among the 95 participants with a mental health provider, only 14 (14.7%) had openly discussed their AI use with their provider. Higher frequency of AI use was associated with greater odds of disclosure (OR, 1.67; 95% CI, 1.20-2.38; P = .003). Conclusion and Relevance: In this survey of adults connected to a large mental health organization, AI chatbots were widely used but engagement for mental health purposes was typically brief and focused. Most use occurred without clinician awareness, suggesting a need for proactive conversations about AI use in routine mental health care.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.