Crisis-line workers' perspectives on AI in suicide prevention: a qualitative exploration of risk and opportunity

Crisis support services offer crucial intervention for individuals in acute distress, providing timely access to trained volunteers whose human connection is key to the effectiveness of these services. However, there are significant disparities in who utilises these services. Recent advancements in artificial intelligence (AI) present new possibilities for crisis intervention in the form of AI-powered conversational partners. Yet, there is little guidance on how AI might be used in this context; and the risks and opportunities remain largely unexplored. This study aims to explore the risks and opportunities of integrating artificial volunteers (AVs) into crisis support, focusing on meeting the needs of those at risk of suicide. It collects the perspectives of crisis service volunteers to contemplate an AVs potential effect on the quality and effectiveness of support provided in crisis situations.

A thematic analysis was used to interpret 13 semi-structured, exploratory qualitative interviews of an UK-based crisis service volunteers.

Three concerns were identified regarding perceived inflexibility and inauthenticity of AI, and the potential for dehumanisation of texters, aligning with factors influencing suicidal behaviour, such as perceived rejection and feelings of entrapment. Despite these concerns, the study found potential advantages, including reduced perceived burden on texters, enhanced confidentiality, and consistent, impartial responses. Recommendations for potential implementations suggest a cautious yet open-minded approach to integrating AI into crisis support, emphasising the need for transparency, accountability, and clarity on the role of human oversight.

AI could complement rather than replace human support in the form of an AV, offering a differentiated, accessible avenue for those who might prefer or benefit from non-human interaction.