Structure Matters: Evaluating Multi-Agents Orchestration in Generative Therapeutic Chatbots
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866918429510860800 |
|---|---|
| author | Elahimanesh, Sina Mohammadkhani, Mohammadali Movahed, Sara Zahedi Abootorabi, Mohammad Mahdi Salehi, Shayan Edalat, Abbas |
| author_facet | Elahimanesh, Sina Mohammadkhani, Mohammadali Movahed, Sara Zahedi Abootorabi, Mohammad Mahdi Salehi, Shayan Edalat, Abbas |
| contents | While large language models (LLMs) excel at open-ended dialogue, effective psychotherapy requires structured progression and adherence to clinical protocols, making the design of psychotherapist chatbots challenging. We investigate how different LLM-based designs shape perceived therapeutic dialogue in a chatbot grounded in the Self-Attachment Technique (SAT), a novel self-administered psychotherapy rooted in attachment theory. We compare three architectural variants: (1) a multi-agent system utilizing finite state machine aligned with therapeutic stages and a shared long-term memory, (2) a single-agent using identical knowledge-base and the same prompts, and (3) an unguided LLM. In an eight-day randomized controlled trial (RCT) with N=66 Farsi-speaking participants, balanced across the three chatbots, the multi-agent system is perceived as significantly more natural and human-like than the other variants and achieves higher ratings across most other metrics. These findings demonstrate that for therapeutic AI, architectural orchestration is as critical as prompt engineering in fostering natural, engaging dialogue. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_00774 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Structure Matters: Evaluating Multi-Agents Orchestration in Generative Therapeutic Chatbots Elahimanesh, Sina Mohammadkhani, Mohammadali Movahed, Sara Zahedi Abootorabi, Mohammad Mahdi Salehi, Shayan Edalat, Abbas Human-Computer Interaction While large language models (LLMs) excel at open-ended dialogue, effective psychotherapy requires structured progression and adherence to clinical protocols, making the design of psychotherapist chatbots challenging. We investigate how different LLM-based designs shape perceived therapeutic dialogue in a chatbot grounded in the Self-Attachment Technique (SAT), a novel self-administered psychotherapy rooted in attachment theory. We compare three architectural variants: (1) a multi-agent system utilizing finite state machine aligned with therapeutic stages and a shared long-term memory, (2) a single-agent using identical knowledge-base and the same prompts, and (3) an unguided LLM. In an eight-day randomized controlled trial (RCT) with N=66 Farsi-speaking participants, balanced across the three chatbots, the multi-agent system is perceived as significantly more natural and human-like than the other variants and achieves higher ratings across most other metrics. These findings demonstrate that for therapeutic AI, architectural orchestration is as critical as prompt engineering in fostering natural, engaging dialogue. |
| title | Structure Matters: Evaluating Multi-Agents Orchestration in Generative Therapeutic Chatbots |
| topic | Human-Computer Interaction |
| url | https://arxiv.org/abs/2603.00774 |