Abstract / Summary
Generative AI – especially ChatGPT – is increasingly used for mental health purposes, but users’ subjective experiences of these tools remain unclear. We characterized how individuals utilize ChatGPT for mental health support through a web-based survey completed by 308 adults (Mage=31.5, SD=5.3; 54% men) who reported prior use of this tool. Around half reported use of ChatGPT for mental health support more than once a week, 80% have used it for more than 3 months, and a third reported increased use over time. Depression (55%), anxiety (74%), stress (66%), and relationship conflict (49%) were the problems most frequently discussed with ChatGPT, with the severity of these problems typically rated as mildly or moderately severe. Endorsed reasons for use were its on-demand availability (71%), free cost (69%), and anonymous nature (54%), while common functions the platform served included obtaining advice (71%), providing a safe space for expression (63%), and to learn coping strategies (61%). Experiences were largely positive, with 8 in 10 agreeing that ChatGPT helped them feel understood, better process their emotions, and improve their mental health literacy. Concerns were also evident, with around half expressing worry about data privacy, and a quarter feeling overly reliant on the platform and reducing their motivation to seek help. Few indicated that ChatGPT had replaced professional help (7%) or discouraged help-seeking (12%). Findings shed light on how users are engaging with generative AI for mental health support and highlight the importance of rigorous research designed to evaluate its safety and long-term efficacy.