Research map / Large language models / AI and society

Position: Beyond Assistance – Reimagining LLMs as Ethical and Adaptive Co-Creators in Mental Health Care

ICML 2025 · Large language models › AI and society · approach: Critiques of LLM claims and judging · 2 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Position paper arguing LLMs in mental health should be co-creators rather than assistants, proposing the SAFE-i implementation guidelines and HAAS-e multidimensional evaluation framework with four metrics (CES, CSI, PAS, ASA).

Problem

LLM adoption in mental health care is limited by concerns about bias, evaluation gaps, over-reliance, dehumanization, and regulatory uncertainty, and existing evaluation focuses too narrowly on technical accuracy.

Approach

Reframes LLMs as augmentative co-creators and offers two structured pathways: SAFE-i guidelines (ethical data foundations, model engineering, real-world integration) and HAAS-e, a human-centered evaluation framework with conceptual metrics for empathy, cultural sensitivity, personalization, and actionability.

Key results

• Illustrative appendix case study scores a single hypothetical conversation at CES 0.90, CSI 0.90, PAS 0.60, and ASA 0.80; no empirical experiments are run.
• Position paper; proposes SAFE-i guidelines with three pillars (ethical data foundations, model engineering, real-world integration) and HAAS-e with four evaluation dimensions.

Limitations

• Metrics are conceptual and illustrated only on a single hypothetical case with no empirical validation.
• Frameworks may need revision as LLM technology and mental health understanding evolve.
• Cannot fully capture human empathy and therapeutic relationship, and cannot eliminate user over-reliance.

Experimental setup

Metrics: CES CSI PAS ASA

Builds on: chain-of-thought tree of thoughts retrieval-augmented generation EPITOME

Closest papers by meaning

• TherapyGym: Evaluating and Aligning Clinical Fidelity and Safety in Therapy Chatbots (NeurIPS 2026 · similarity 0.87)
• Bridging the Knowledge Gap: Understanding User Expectations for Trustworthy LLM Standards (AAAI 2025 · similarity 0.85)
• MindGuard: Guardrail Classifiers for Multi-Turn Mental Health Support (NeurIPS 2026 · similarity 0.84)
• Apathetic or Empathetic? Evaluating LLMs' Emotional Alignments with Humans (NeurIPS 2024 · similarity 0.84)
• CARES: Comprehensive Evaluation of Safety and Adversarial Robustness in Medical LLMs (NeurIPS 2025 · similarity 0.84)
• Evaluating LLMs Across Multi-Cognitive Levels: From Medical Knowledge Mastery to Scenario-Based Problem Solving (ICML 2025 · similarity 0.84)