Team:

Neo Jie Xiang is a 3rd year undergraduate at NUS with a major in Computer Science and second major in Physics. He was aided by Professor Gerard Chung Siew Keong from the NUS Department of Social Work, who helped to steer the direction of the project with his Social Work background.

​

Project Overview:

The project aims to design a tool to enhance online text-based counseling/engagement effectiveness by integrating AI to suggest motivational interviewing (MI) strategies and responses, addressing gaps in counseling training and support for social work students, social workers, psychologists, counsellors, and para-counsellors.

 

Why the team applied for CCSGP Fellowship:

We wanted to develop an AI tool to help support social work students in their engagement skills training since there is a lack of diverse, repetitive, and on-demand training experiences for students. Thus, by developing a tool to better train the next generation of social workers, there is a great opportunity to do good and contribute back to society, which aligns well with CCSGP’s mission.

 

About the Project:

We developed the SWAT:RolePlay application in which users can converse with Large Language Models (LLMs) using custom client profiles. These custom profiles are created using social work expertise to accurately reflect real-world client interactions to provide a more realistic training scenario for social work students. The application also incorporates LLMs to provide evaluations of the student’s chat transcripts using social work frameworks such as Motivational Interviewing (MI). This provides an opportunity for self-reflection and improvement, as well as reinforcing ideas and techniques taught in class.

​

What the team learned from the CCSGP experience:
We have gained a better understanding of the challenges faced when developing such a tool. Namely, the creation of high quality prompts, the pedagogical requirements, the appropriate user interfaces for the end user and instructor, the costs to host the application and support the OpenAI API calls, among many others. We have also undergone some small scale user testing and evaluations to better understand the needs of the user, as well as inform future refinements to the tool.

We have also been exposed to the intersection of two different disciplines, Social Work and Computer Science. This has taught us the importance of interdisciplinary communication and utilizing each discipline's perspectives and experiences to best reach our desired goals.
​

Screenshots:

​