From Pattern Recognizers to Personalized Companions: A Survey of Large Language Models in Mental Health

arXiv:2609.25186v1 Announce Type: new Abstract: The rising global prevalence of mental health conditions, together with longstanding barriers in traditional healthcare, such as limited resources, high cost, stigma, and privacy concerns, has created an urgent need for accessible and scalable support. Large Language Models (LLMs) have emerged as a transformative technology with strong potential to democratize mental health support through advanced natural language understanding and generation. However, the rapidly expanding, fragmented body of work in this area lacks a coherent evolutionary narrative, making it difficult to contextualize current progress and identify future directions. This survey addresses this gap by organizing and analyzing the literature around a central thesis: the role of LLMs in mental health is evolving through three distinct, increasingly sophisticated phases. We trace this trajectory from Phase I, in which LLMs act primarily as passive Information Tools and Pattern Recognizers for assessment; through Phase II, where they function as Empathetic Conversationalists for in-the-moment, stateless interactions; to the current frontier, Phase III, which seeks Longitudinal, Personalized Companions implemented as stateful cognitive agents. To support this framework, we systematically review core technologies, agent architectures (Profile, Memory, Reasoning, and Planning), and the critical infrastructure of datasets and benchmarks, highlighting how their evolution underpins this developmental path. Viewing the field through this developmental lens, we provide

LLMs may improve mental health support.

The survey highlights how LLMs can democratize mental health support by overcoming traditional barriers such as cost, stigma, and resource limitations. It identifies a clear progression toward personalized, stateful AI companions that may enhance accessibility and personalize care, which is materially optimistic for societal mental health outcomes. However, risks related to privacy and efficacy remain, moderating the score.

This is a survey paper summarizing and categorizing existing research trends, not reporting new evidence of real-world impact, efficacy, or harm. Its societal implication is inherently speculative and forward-looking, framed positively but without concrete outcome data.

This academic survey outlines the potential for LLMs to transform mental health support, detailing an evolution towards personalized AI companions. It emphasizes increased accessibility and reduced barriers to care as key benefits. While acknowledging existing challenges, the article primarily presents an optimistic future roadmap for AI in this domain, focusing on the positive implications for human flourishing.

SOURCE     arXiv cs.CY (Computers and Society) (tier 1)
   ↓
DOCUMENT   fa46de1e-7c6d-4768-b0fe-867c6b27a700
           https://arxiv.org/abs/2609.25186
   ↓
EVIDENCE   2 extracted excerpts
   ↓
MODEL RUN  4 runs, methodology 1.0.1
   ↓
SCORE      +4  (Mixed or uncertain)
   ↓
CONFIDENCE 75%