The global mental health crisis affects over 970 million people worldwide, with access to effective care remaining limited. Digital mental health interventions (DMHIs), including rule-based chatbots, have shown promise but often lack engagement and long-term effectiveness. This study investigates the potential of generative AI chatbots, such as ChatGPT, in providing meaningful mental health support, examining user experiences and identifying challenges and opportunities.
Methodology
Researchers conducted semi-structured interviews with 19 participants from diverse backgrounds who had used generative AI chatbots for mental health purposes. Reflexive thematic analysis was used to identify overarching themes and subthemes, focusing on users’ real-world, unprompted experiences.
Key Findings
- Emotional Sanctuary
Participants described chatbots as nonjudgmental, empathetic, and always available. Many found comfort and validation, particularly during difficult emotional periods. However, frustrations arose when safety guardrails disrupted conversations or when the chatbot’s responses felt generic or constrained. - Insightful Guidance
Generative AI chatbots provided practical advice on relationships, self-care, and reframing negative thoughts. For some users, the insights were life-changing, helping them navigate complex personal situations. However, trust in chatbot advice varied, and some users desired more proactive and tailored support. - Joy of Connection
Many participants experienced a sense of companionship and joy in their interactions with chatbots, which often led to reduced loneliness and increased confidence in opening up to others. Chatbots were also praised for their ability to facilitate creative therapeutic strategies, such as role-playing or visualizing conversations. - Comparisons with Human Therapy
Generative AI chatbots were often used as a complement to therapy or when therapy was inaccessible. While participants valued the scalability and availability of chatbots, they noted limitations in empathy, memory, and the ability to lead therapeutic processes.
Challenges and Opportunities
The study highlighted key areas for improvement, including:
- Enhancing memory to retain user context over time.
- Balancing safety guardrails with meaningful crisis support.
- Providing richer, multimedia interfaces and more accessible designs.
The need for sustainable business models was also emphasized, as existing consumer-focused services often struggle with scalability and funding.
Conclusions
Generative AI chatbots hold significant potential to address gaps in mental health care by offering scalable, accessible, and personalized support. However, their effectiveness and safety require further research, particularly in diverse populations and crisis scenarios. Developers should prioritize improving listening skills, therapeutic guidance, and user trust to maximize the impact of these tools.
