An Examination of Generative AI Response to Suicide Inquires: Content Analysis
Bibliographic record
Abstract
Background: Generative artificial intelligence (AI) chatbots are an online source of information consulted by adolescents to gain insight into mental health and wellness behaviors. However, the accuracy and content of generative AI responses to questions related to suicide have not been systematically investigated. Objective: This study aims to investigate general (not counseling-specific) generative AI chatbots' responses to questions regarding suicide. Methods: A content analysis was conducted of the responses of generative AI chatbots to questions about suicide. In phase 1 of the study, generative chatbots examined include: (1) Google Bard or Gemini; (2) Microsoft Bing or CoPilot; (3) ChatGPT 3.5 (OpenAI); and (4) Claude (Anthropic). In phase 2 of the study, additional generative chatbot responses were analyzed, which included Google Gemini, Claude 2 (Anthropic), xAI Grok 2, Mistral AI, and Meta AI (Meta Platforms). The two phases occurred a year apart. Results: Findings included a linguistic analysis of the authenticity and tone within the responses using the Linguistic Inquiry and Word Count program. There was an increase in the depth and accuracy of the responses between phase 1 and phase 2 of the study. There is evidence that the responses by the generative AI chatbots were more comprehensive and responsive during phase 2 than phase 1. Specifically, the responses were found to provide more information regarding all aspects of suicide (eg, signs of suicide, lethality, resources, and ways to support those in crisis). Another difference noted in the responses between the first and second phases was the emphasis on the 988 suicide hotline number. Conclusions: While this dynamic information may be helpful for youth in need, the importance of individuals seeking help from a trained mental health professional remains. Further, generative AI algorithms related to suicide questions should be checked periodically to ensure best practices regarding suicide prevention are being communicated.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".