JMIR Publications

ABSTRACT

Background:

Generative artificial intelligence (AI) has become increasingly popular for the general public to use for mental health needs despite the lack of regulation oversight.

Objective:

Our study sought to understand perspectives of individuals who engaged with a generative AI, ChatGPT, for therapeutic purposes using a digital ethnographic approach.

Methods:

We systematically extracted and analyzed all posts on the social media platform Reddit that contained the words “ChatGPT” and “therapy” using thematic analysis to understand why, how, and for what purposes individuals have used ChatGPT as a form of mental health self-help.

Results:

Our findings showed that users utilized ChatGPT to manage mental health problems, seek self-discovery, obtain companionship, and gain mental health literacy. Individuals engaged ChatGPT creatively for therapeutic purposes through many ways such as conducting therapist role-play, coaching and manipulating, re-enacting troubling events, externalizing thoughts, and using it to facilitate real-life therapy. Users are drawn to perceived emotional connection and expertise that are similar in human therapists, and also perceived unique advantages of ChatGPT in accessibility, perceived cognitive capacity, personalization, and lack of negative reactions in responses over human beings. Concerns regarding privacy and emotional depth and realness were raised but minimally.

Conclusions:

The findings highlighted the paradoxical nature of AI mental health support in providing the sense of being “seen and heard” and yet lacking complex emotions or natural ruptures. The results also emphasized the key role of users’ agency in directing therapeutic experiences and highlighted concerns regarding limited awareness of ethical risks.