The ChatGPT Effect: Investigating Shifting Discourse Patterns, Sentiment, and Benefit-Challenge Framing in AI Mental Health Support.

Résumé

AI has the potential to enhance mental health by scaling support. However, its implementation brings uncertainties and challenges that require careful review to ensure safety. This study examined evolving public views on AI mental health support by analyzing relevant Reddit posts ( = 517). Following the release of ChatGPT in 2022, discussions about AI in the context of mental health surged, with a noticeable shift in preference toward large language models (LLMs) over conventional therapy chatbots. Users appreciated AI for its emotional support, companionship, and accessibility, while also expressing concerns about adverse effects and lack of conversational depth and emotional connection. Distinct patterns in how benefits and challenges were discussed emerged between experienced and non-experienced AI users, as well as between AI-focused and mental health-focused communities. AI-experienced users acknowledged both the benefits and limitations, whereas AI communities emphasized the positives and mental health communities highlighted the lack of conversational depth. These findings underscore the need for tailored communication strategies to set realistic expectations about the utility of AI in mental healthcare among different stakeholders. This research provides insights into developing ethical AI systems that complement traditional care while addressing current limitations.