Seeking Late Night Life Lines: Experiences of Conversational AI Use in Mental Health Crisis

AI fills critical gaps but cannot replace human connection: The study of 53 people in mental health crisis found that individuals turn to conversational AI primarily due to lack of access to mental health professionals or fear of burdening loved ones. However, interviews with 16 mental health experts emphasized that human-human connection is essential for managing mental health crises, suggesting AI should complement rather than substitute professional care.

Current AI use in crisis situations raises responsibility concerns: People are already replacing their therapists with AI agents like ChatGPT and Claude during mental health emergencies, yet the paper indicates it remains "unclear how to meet this potential in extreme mental health crisis use cases," highlighting a gap between current usage patterns and what responsible AI design should enable.

Responsible AI crisis intervention requires specific design goals: Rather than providing crisis resolution, responsible conversational AI should focus on two key outcomes: increasing users' preparedness to take positive actions (like contacting a therapist or trusted person) while simultaneously de-escalating any intended negative actions—functioning as a "bridge towards human-human connection rather than ends in themselves."

AI agents address isolation and accessibility barriers: The research found that people use AI to "fill the in-between spaces of human support," suggesting these tools could play a legitimate role in mental health support systems, but only when designed with the explicit purpose of connecting people to human professionals rather than positioning themselves as standalone solutions.