Large language model counseling system learns to improve with each session
PsyEvo: A Personalized Counseling Agent That Self-Evolves at Test Time
Artificial Intelligence
Summary
Many people with mental health issues cannot easily find enough trained counselors. To help, the authors created PsyEvo, a chatbot that talks to users like a counselor and gets better over time. PsyEvo personalizes conversations for each client and learns from feedback during sessions to improve its responses. The authors tested it with simulated clients and found it performed better than versions lacking key parts.
What this means in practice
- •For mental health service providers: Offer scalable, adaptive counseling support that personalizes interventions and improves through client interactions.
- •For conversational ai developers: Develop dialogue agents that self-improve their intervention strategies and response quality during live use from feedback signals.
Authors
Yuting Yan, Shihao Xu, Junhao Yu, Mingcong Zuo, Lu Chen, Nan Xiang, Haiyang Geng, Dongjie Tao, Minghao Wang
Abstract
Mental health disorders affect a substantial proportion of the global population, yet a persistent shortage of trained practitioners leaves the majority without adequate care. Large language model (LLM)-based counselors present a promising direction for delivering scalable conversational psychological support. Offline model training alone leaves limited room to adapt to individual clients or to learn from ongoing therapeutic interaction at test time. We introduce PsyEvo, an LLM-based counseling framework that enables both client-specific personalization and response-policy improvement at test time through three components: Hierarchical Bayesian Skill Policy (HBSP) personalizes what intervention to apply by maintaining a per-client skill posterior updated from session feedback; Inter-session Listwise Preference Optimization (LiPO) improves how the selected skill is expressed by updating a shared response adapter from cross-client preference evidence; and State-conditioned Ordinal Credit Assignment (SOCA) supplies candidate preferences and trajectory credit to the two components through consistency-checked comparisons and ordinal projection. In simulated-client evaluation with shared online cohort adaptation, PsyEvo obtains 7.684 Overall on PsychEval and exceeds every component variant in each of three matched runs. Removing individual components lowers mean overall score by 0.138--0.171 under the shared configuration, supporting conditional contributions within the complete scaffold. Our code is available at https://github.com/Lingxi-mental-health/PsyEvo