LLM Agents for Psychology: A Study on Gamified Assessments

LLM Agents for Psychology: A Study on Gamified Assessments

19 Feb 2024 | Qisen Yang, Zekun Wang, Honghui Chen, Shenzhi Wang, Yifan Pu, Xin Gao, Wenhao Huang, Shiji Song, Gao Huang
The paper introduces PsychoGAT (Psychological Game AgenTs), a novel framework that leverages large language models (LLMs) to transform traditional psychological assessments into interactive fiction games. The main insight is that powerful LLMs can function as both adept psychologists and innovative game designers. By incorporating LLM agents into designated roles and carefully managing their interactions, PsychoGAT can convert standardized scales into personalized and engaging interactive fiction games. The framework consists of three main agents: the game designer, the game controller, and the critic, each with specific responsibilities in creating, generating, and refining the game content. The authors validate the effectiveness of PsychoGAT through psychometric evaluations and human evaluations, demonstrating its statistical significance in reliability, convergent validity, and discriminant validity. Human evaluations also confirm improvements in content coherence, interactivity, interest, immersion, and satisfaction. The experimental results show that PsychoGAT is a reliable and valid method for psychological assessment, offering a more engaging and immersive experience compared to traditional self-report scales and psychologist interviews.The paper introduces PsychoGAT (Psychological Game AgenTs), a novel framework that leverages large language models (LLMs) to transform traditional psychological assessments into interactive fiction games. The main insight is that powerful LLMs can function as both adept psychologists and innovative game designers. By incorporating LLM agents into designated roles and carefully managing their interactions, PsychoGAT can convert standardized scales into personalized and engaging interactive fiction games. The framework consists of three main agents: the game designer, the game controller, and the critic, each with specific responsibilities in creating, generating, and refining the game content. The authors validate the effectiveness of PsychoGAT through psychometric evaluations and human evaluations, demonstrating its statistical significance in reliability, convergent validity, and discriminant validity. Human evaluations also confirm improvements in content coherence, interactivity, interest, immersion, and satisfaction. The experimental results show that PsychoGAT is a reliable and valid method for psychological assessment, offering a more engaging and immersive experience compared to traditional self-report scales and psychologist interviews.
Reach us at info@study.space