VCounselor: A Psychological Intervention Chat Agent Based on a Knowledge-Enhanced Large Language Model

Conversational artificial intelligence can already independently engage in brief conversations with clients with psychological problems and provide evidence-based psychological interventions. The main objective of this study is to improve the effectiveness and credibility of the large language model in psychological intervention by creating a specialized agent, the VCounselor, to address the limitations observed in popular large language models such as ChatGPT in domain applications. We achieved this goal by proposing a new affective interaction structure and knowledge-enhancement structure. In order to evaluate VCounselor, this study compared the general large language model, the fine-tuned large language model, and VCounselor's knowledge-enhanced large language model. At the same time, the general large language model and the fine-tuned large language model will also be provided with an avatar to compare them as an agent with VCounselor. The comparison results indicated that the affective interaction structure and knowledge-enhancement structure of VCounselor significantly improved the effectiveness and credibility of the psychological intervention, and VCounselor significantly provided positive tendencies for clients' emotions.

Introduction. Psychological problems have gradually evolved into global health problems (Bullis et al., 2019). Depression, for example, is one of the most common psychological disorders in the world, yet almost two-thirds of those suffering from it do not receive effective and sufficient treatment (WHO, 2017; 2020). The reason for this situation is not only that the number of well-trained staff in psychiatric medical services is far less than the number of patients with psychological disorders, but also the economic burden of offline psychological treatment, such as travel costs. Besides that, psychotherapy is a long-term process, and it is difficult to see progress in a short period, which Artificial Intelligence (AI) provides a path to solve these problems. AI can process large amounts of complex data in a short period, providing therapists with the corresponding information to help them achieve patient management, thus accomplishing almost real-time decision-making (Luxton, 2014).

Discussion / Conclusion. Overall, the experimental results show that the VCounselor achieved better results in comparing groups. This indicates that the affective interaction structure providing empathy and transference and the knowledge-enhancement structure providing professional knowledge and guidance are factors that must be attended to when applying the LLM to the field of psychological intervention. However, there are still some issues that need to be discussed in the comparison group. The questionnaire analysis results indicated better results with avatars than without, and the knowledge-enhanced model was superior to the fine-tuned model and the fine-tuned model was superior to the general model. But in detail, in the fine-tuned model, the group with avatars scored slightly lower than the group without avatars on both Trustworthiness and CSS. This may be related to the characteristics of the fine-tuned model we observed in cases.