JMIR Publications

ABSTRACT

Background:

Artificial intelligence (AI) has become increasingly utilized in mental health care for applications such as diagnosis, monitoring, and treatment support. These include tools like clinician support systems, large language models, and conversational agents used to augment psychotherapy and clinical decision making. While prior research suggests potential benefits of and concerns with artificial intelligence, little is known within the domain of obsessive-compulsive disorder (OCD). Given the expanding role of AI in psychiatry, understanding these perspectives is essential to ensuring AI implementation aligns with patient priorities and values.

Objective:

To explore the perspectives of individuals with OCD on the use of artificial intelligence in health care, including perceived benefits, risks, and its role in relation to human clinicians.

Methods:

We conducted semi-structured interviews with 24 adults self-reporting OCD, recruited through online communities and advocacy networks. Eligible individuals (≥18 years with self-reported OCD) completed screening, provided informed consent, and participated in remote HIPAA-compliant Zoom interviews (May–December 2024). Transcripts were de-identified, open-coded, and used to develop a codebook. Focused codes were applied using a thematic analysis framework in Dedoose (v9.2.22). Each transcript was independently coded by two reviewers, with discrepancies resolved through consensus. Themes were developed through iterative interpretive analysis of code clusters.

Results:

Participants’ perspectives encompassed concerns and benefits of AI in mental healthcare. Participants expressed concerns about the accuracy and efficacy of information provided by AI as well as a limited ability for clinical judgement in psychiatric care. Additionally, participants emphasized the importance of human connection, particularly therapeutic alliance, empathy, and reassurance provided by clinicians, which they felt AI could not replicate. Concerns about data privacy, security, and downstream use of information were also highlighted. Despite concerns, many endorsed the use of AI as an adjunct rather than a replacement to clinicians, noting potential benefits in symptom monitoring, preliminary information gathering, and support for administrative tasks, provided that human oversight is maintained.

Conclusions:

Individuals with OCD expressed nuanced views on AI in mental health care, balancing cautious optimism with several concerns. While AI may improve efficiency, standardization, and symptom monitoring, participants highlighted risks related to de-individualization, accuracy, and erosion of human connection. These findings underscore the importance of patient-centered, ethically guided AI integration that preserves the therapeutic alliance while leveraging technological benefits.