Seven clinician-informed safety criteria enable LLM-as-a-Judge to reach substantial agreement with human consensus (Cohen's κ up to 0.75) on evaluating LLM responses to users demonstrating psychosis.
hub Canonical reference
Ong, and Nick Haber
Canonical reference. 83% of citing Pith papers cite this work as background.
14
Pith papers citing it
Background
83% of classified citations
hub tools
citation-role summary
background 6
citation-polarity summary
roles
background 6representative citing papers
Direction-flipped influence audits show contextual cues shift LLM moral choices by 12-18 points on average across multiple benchmarks, revealing asymmetries, backfires, and inconsistencies in 40% of conditions.
Spoken language models exhibit style amnesia and fail to maintain instructed paralinguistic styles across multi-turn conversations, with explicit recall offering partial mitigation.
The study proposes the Gradual Voluntary Participation (GVP) framework to reconceptualize participatory AI governance in journalism as a gradual and voluntary process using a bidimensional matrix.
Chaplains view AI chatbots as unable to provide attuned pastoral care for non-clinical emotional needs, based on themes of listening, connecting, carrying, and wanting.
EvalAI providing pro/con arguments improves provision-level accuracy and reduces misclassification distance in DSA illegal content reporting under AI error conditions versus conventional XAI.
Empirical analysis of 1,524 AI incident reports shows 83% arise from worker-AI trait misalignments, with 74% of those traceable to developers prioritizing efficiency over precision or personalization.
A 16-factor structured prompt framework strengthens CoT reasoning in LLMs for security analysis, yielding up to 40% reasoning gains in smaller models and stable accuracy improvements validated by human raters with Cohen's k > 0.80.
Mainstream conversational models show escalating affective misalignments and ethical guidance failures during staged emotional trajectories, organized into a taxonomy of interactional breakdowns.
Shame/stigma and access barriers to therapy predict higher perceived helpfulness of AI mental health support, especially for therapy-experienced users, while access and cost barriers predict greater usage intensity.
A literature review concludes that pursuing consensus in data annotation creates biased AI by dismissing subjective disagreements and enforcing geographic hegemony, and proposes mapping diversity instead.
KV cache compression causes certain instructions to degrade rapidly and be ignored in multi-instruction prompting, with system prompt leakage worsened by method choice, instruction order, and eviction bias; simple policy changes can mitigate this.
Authors propose a four-stage framework to analyze opportunities and risks of generative AI across the health information journey from public sources to clinical care.
A cross-sector primer on AI chatbots as de facto mental health support, mapping challenges in suicide and NSSI response and identifying priority areas for industry alignment on standards and oversight.
citing papers explorer
-
Using LLM-as-a-Judge/Jury to Advance Scalable, Clinically-Validated Safety Evaluations of Model Responses to Users Demonstrating Psychosis
Seven clinician-informed safety criteria enable LLM-as-a-Judge to reach substantial agreement with human consensus (Cohen's κ up to 0.75) on evaluating LLM responses to users demonstrating psychosis.
-
Direction-Flipped Influence Audits Reveal Hidden Structure in Moral Choices of LLMs
Direction-flipped influence audits show contextual cues shift LLM moral choices by 12-18 points on average across multiple benchmarks, revealing asymmetries, backfires, and inconsistencies in 40% of conditions.
-
Style Amnesia: Investigating Speaking Style Degradation and Mitigation in Multi-Turn Spoken Language Models
Spoken language models exhibit style amnesia and fail to maintain instructed paralinguistic styles across multi-turn conversations, with explicit recall offering partial mitigation.
-
Gradual Voluntary Participation: A Framework for Participatory AI Governance in Journalism
The study proposes the Gradual Voluntary Participation (GVP) framework to reconceptualize participatory AI governance in journalism as a gradual and voluntary process using a bidimensional matrix.
-
Chaplains' Reflections on the Design and Usage of AI for Conversational Care
Chaplains view AI chatbots as unable to provide attuned pastoral care for non-clinical emotional needs, based on themes of listening, connecting, carrying, and wanting.
-
AI at the Front Lines of Platform Governance: Using LLMs to Support Illegal Content Reporting under the Digital Services Act
EvalAI providing pro/con arguments improves provision-level accuracy and reduces misclassification distance in DSA illegal content reporting under AI error conditions versus conventional XAI.
-
The Quiet Path from Seemingly Minor Design Errors to Workplace AI Incidents
Empirical analysis of 1,524 AI incident reports shows 83% arise from worker-AI trait misalignments, with 74% of those traceable to developers prioritizing efficiency over precision or personalization.
-
Strengthening Human-Centric Chain-of-Thought Reasoning Integrity in LLMs via a Structured Prompt Framework
A 16-factor structured prompt framework strengthens CoT reasoning in LLMs for security analysis, yielding up to 40% reasoning gains in smaller models and stable accuracy improvements validated by human raters with Cohen's k > 0.80.
-
Breakdowns in Conversational AI: Interactional Failures in Emotionally and Ethically Sensitive Contexts
Mainstream conversational models show escalating affective misalignments and ethical guidance failures during staged emotional trajectories, organized into a taxonomy of interactional breakdowns.
-
Talking to a Human as an Attitudinal Barrier: A Mixed Methods Evaluation of Stigma, Access, and the Appeal of AI Mental Health Support
Shame/stigma and access barriers to therapy predict higher perceived helpfulness of AI mental health support, especially for therapy-experienced users, while access and cost barriers predict greater usage intensity.
-
The Consensus Trap: Dissecting Subjectivity and the "Ground Truth" Illusion in Data Annotation
A literature review concludes that pursuing consensus in data annotation creates biased AI by dismissing subjective disagreements and enforcing geographic hegemony, and proposes mapping diversity instead.
-
The Pitfalls of KV Cache Compression
KV cache compression causes certain instructions to degrade rapidly and be ignored in multi-instruction prompting, with system prompt leakage worsened by method choice, instruction order, and eviction bias; simple policy changes can mitigate this.
-
Opportunities and Risks of Generative AI through the Health Information Journey
Authors propose a four-stage framework to analyze opportunities and risks of generative AI across the health information journey from public sources to clinical care.
-
AI and Suicide Prevention: A Cross-Sector Primer
A cross-sector primer on AI chatbots as de facto mental health support, mapping challenges in suicide and NSSI response and identifying priority areas for industry alignment on standards and oversight.