Empowerment of Large Language Models in Psychological Counseling through Prompt Engineering
Shanshan Huang, Fuxiang Fu, Ke Qiang Yang, Ke Zhang, Fan Yang · 2024
The rapid advancement of artificial intelligence (AI) has led to significant strides in the development of large language models, which are increasingly adept at comprehending and processing natural language. These models possess substantial memory capacity and can exhibit logical reasoning. However, their efficacy in addressing the emotional and empathetic aspects of human interaction remains a challenge. Current models may provide superficial advice that fails to penetrate the depth of an individual's emotional state.We introduce a novel approach to enhance the emotional intelligence of large language models, aiming to fostering a more empathetic and emotionally attuned interaction with users. We conducted experiments with several prominent models, including ChatGLM and ERNIE Bot, em-ploying a variety of promptings. These ranged from presenting examples without explicit directives to providing a limited set of examples and extending narratives from a given context. We simulated therapeutic conversations to evaluate the models' performance in emotionally charged scenarios. Our findings indicate that by refining the guidance mechanisms for these models, it is possible substantially to improve their capacity for emotional engagement and understanding.