AI or Human Support for Suicide Prevention? Examining Help-Seeking Intention in Suicidal Crisis
Abstract
This preprint has now been published in the peer-reviewed journal Suicide Policy Research. Please read and cite the published version:
Nakamura, D., Sueki, H., & Ito, J. (2026). AI or human support for suicide prevention? Examining help-seeking intention in suicidal crisis. Suicide Policy Research, 5(1), 1–12.
The published article is freely available at: https://jscp.or.jp/assets/img/SPR_5-1_1_Nakamura.pdf
Background: Chat-based crisis support is a common strategy for suicide prevention, and artificial intelligence (AI) chatbots have the potential to improve the low response rates to such support. Although AI chatbots are technically feasible for crisis support, it remains unclear whether individuals experiencing suicidal ideation intend to use them.
Aims: To examine the intentions of adults with depressive symptoms experiencing suicidal ideation to use chat-based crisis support (human vs AI) and explore how these intentions relate to their background characteristics.
Methods: A cross-sectional web-based survey was performed, recruiting participants aged 18–69 with a Kessler-6 score ≥13 from an internet research company’s database in Japan. Participants completed a questionnaire assessing their intention to use chat-based crisis support when feeling suicidal, sociodemographic factors, and suicide-related characteristics. Subsequently, participants were categorized into four groups: (1) those intending to use both human and AI-based chats, (2) AI chat only, (3) human chat only, and (4) neither. We calculated the proportions of each group and used the chi-square, Kruskal–Wallis, and Fisher’s exact tests to compare group differences.
Results: Overall, 1,024 participants completed the survey, and we determined that 19.5% intended to use AI chat support and 21.3% intended to use human chat. The proportions of participants who intended to use both human- and AI-based chat, AI chat only, human chat only, and neither option were 15.6 %, 3.9 %, 5.7 %, and 74.8 %, respectively. Significant differences were observed in gender (p = 0.028), history of crisis support use (p <0.001), and current issues (financial problems [p = 0.002)] and sexuality-related concerns (p = 0.006) across the groups.
Conclusion: In this web-based survey, approximately 20% of individuals with depressive symptoms feeling suicidal were willing to seek help, even if it was provided by AI.
Related
- “We need all the support we can get”: a rapid qualitative analysis of cross-national LGBTQ+ perspectives on social support, mental health, and suicide prevention
- Bayesian, network, and cross-validated modeling of suicidal ideation in people with suicide attempt histories
- Non-suicidal self-injury and subsequent suicide attempts and death: A protocol for a systematic review of temporally ordered evidence
- Characterizing Use of the Internet and Social Media to Access Suicide-Related Content among Adolescents with Suicidal Ideation
- Psychometric Evaluation of Sinhala and Tamil Measures of Suicidal Behavior and Interpersonal Suicide Risk Constructs in Sri Lanka.
BibTeX
@misc{nakamura2026human,
title = {{AI or Human Support for Suicide Prevention? Examining Help-Seeking Intention in Suicidal Crisis}},
author = {Daisuke Nakamura and Hajime Sueki and Jiro Ito},
howpublished = {PsyArXiv},
year = {2026},
month = oct,
doi = {10.31234/osf.io/z5wrv_v2},
url = {https://osf.io/preprints/psyarxiv/z5wrv_v2/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - AI or Human Support for Suicide Prevention? Examining Help-Seeking Intention in Suicidal Crisis
AU - Daisuke Nakamura
AU - Hajime Sueki
AU - Jiro Ito
PY - 2026
DA - 2026/10/01
PB - PsyArXiv
DO - 10.31234/osf.io/z5wrv_v2
UR - https://osf.io/preprints/psyarxiv/z5wrv_v2/
AB - This preprint has now been published in the peer-reviewed journal Suicide Policy Research. Please read and cite the published version: Nakamura, D., Sueki, H., & Ito, J. (2026). AI or human support for suicide prevention? Examining help-seeking intention in suicidal crisis. Suicide Policy Research, 5(1), 1–12. The published article is freely available at: https://jscp.or.jp/assets/img/SPR_5-1_1_Nakamura.pdf Background: Chat-based crisis support is a common strategy for suicide prevention, and artificial intelligence (AI) chatbots have the potential to improve the low response rates to such support. Although AI chatbots are technically feasible for crisis support, it remains unclear whether individuals experiencing suicidal ideation intend to use them. Aims: To examine the intentions of adults with depressive symptoms experiencing suicidal ideation to use chat-based crisis support (human vs AI) and explore how these intentions relate to their background characteristics. Methods: A cross-sectional web-based survey was performed, recruiting participants aged 18–69 with a Kessler-6 score ≥13 from an internet research company’s database in Japan. Participants completed a questionnaire assessing their intention to use chat-based crisis support when feeling suicidal, sociodemographic factors, and suicide-related characteristics. Subsequently, participants were categorized into four groups: (1) those intending to use both human and AI-based chats, (2) AI chat only, (3) human chat only, and (4) neither. We calculated the proportions of each group and used the chi-square, Kruskal–Wallis, and Fisher’s exact tests to compare group differences. Results: Overall, 1,024 participants completed the survey, and we determined that 19.5% intended to use AI chat support and 21.3% intended to use human chat. The proportions of participants who intended to use both human- and AI-based chat, AI chat only, human chat only, and neither option were 15.6 %, 3.9 %, 5.7 %, and 74.8 %, respectively. Significant differences were observed in gender (p = 0.028), history of crisis support use (p <0.001), and current issues (financial problems [p = 0.002)] and sexuality-related concerns (p = 0.006) across the groups. Conclusion: In this web-based survey, approximately 20% of individuals with depressive symptoms feeling suicidal were willing to seek help, even if it was provided by AI.
ER -