🗂️ 본문: 개방형 그룹 채팅 환경에서의 CBT 치료 챗봇 설계
While Cognitive Behavioural Therapy (CBT) has proven effective in both individual and group settings, current AI-based therapeutic interventions primarily focus on one-to-one interactions, neglecting group CBT benefits including mutual support and social learning. This project presents a novel AI chatbot framework for facilitating CBT interventions within group chat environments. We developed an agent-based system that autonomously monitors group conversations and delivers targeted CBT interventions through three components: (1) conversation context management via dynamic summarisation, (2) intelligent intervention decision-making based on CBT principles, and (3) strategic action selection from four therapeutic actions—conversation summarisation, key thought identification, coping strategy suggestions, and encouragement provision. The system employs the Transtheoretical Model of Behaviour Change to personalise interventions according to users' motivational stages.
Demonstrating with realistic group chat scenarios demonstrated the system's ability to accurately identify cognitive distortions, provide contextually appropriate therapeutic responses, and maintain group cohesion while delivering personalised interventions. The chatbot successfully distinguished between utterances requiring therapeutic intervention and casual exchanges, ensuring non-intrusive operation. This research demonstrates the feasibility of implementing group CBT principles in AI-mediated environments, offering a scalable approach that bridges traditional group therapy advantages with AI technology's consistency and availability, presenting new possibilities for digital mental health interventions.