A Large Language Model-Integrated Robotic Storyteller to Improve College Student’s Mental Health
Loading...
Mentor Information
Dr. Zhao Han
Description
Since the COVID-19 pandemic, mental health issues in college students have surged, presenting various challenges to their academic success and overall well-being. Robots with storytelling capabilities have emerged as promising tools to tackle this issue, because, besides eliciting empathy, engaging in storytelling activities can temporarily shift people’s focus away from worries and tension, allowing for mental and emotional rejuvenation. However, most of the robots in those research that were published before the popularity of large language models (LLMs) usually relied on pre-programmed responses and scripts based on a finite set of scenarios or situations, which might limit the robot’s capacity to have a deep conversation and personalize to user’s response and emotion. In this work, we explore an LLM-integrated robot that is tailored to act as a storyteller who can both generate stories and listen to users’ narratives. We plan to deploy the robot in participating students’ dormitories in 5 days and engage them in story-based conversations with robot Misty in both first-person narrative voice (1PNV) and third-person narrative voice (3PNV). We anticipate a notable improvement in the mental well-being of participants upon completion of the study. Moreover, with LLM integration, the robot with 1PNV will likely yield superior results compared to the one with 3PNV. Our work offers valuable insights into the effectiveness of LLM application in storytelling robots to address mental health issues among college communities, paving the way to the integration of LLM-integrated robotic storytellers into campus mental health services as a complementary resource.
A Large Language Model-Integrated Robotic Storyteller to Improve College Student’s Mental Health
Since the COVID-19 pandemic, mental health issues in college students have surged, presenting various challenges to their academic success and overall well-being. Robots with storytelling capabilities have emerged as promising tools to tackle this issue, because, besides eliciting empathy, engaging in storytelling activities can temporarily shift people’s focus away from worries and tension, allowing for mental and emotional rejuvenation. However, most of the robots in those research that were published before the popularity of large language models (LLMs) usually relied on pre-programmed responses and scripts based on a finite set of scenarios or situations, which might limit the robot’s capacity to have a deep conversation and personalize to user’s response and emotion. In this work, we explore an LLM-integrated robot that is tailored to act as a storyteller who can both generate stories and listen to users’ narratives. We plan to deploy the robot in participating students’ dormitories in 5 days and engage them in story-based conversations with robot Misty in both first-person narrative voice (1PNV) and third-person narrative voice (3PNV). We anticipate a notable improvement in the mental well-being of participants upon completion of the study. Moreover, with LLM integration, the robot with 1PNV will likely yield superior results compared to the one with 3PNV. Our work offers valuable insights into the effectiveness of LLM application in storytelling robots to address mental health issues among college communities, paving the way to the integration of LLM-integrated robotic storytellers into campus mental health services as a complementary resource.
