A Cross-disciplinary Application of Generative Artificial Intelligence (GenAI) in Innovating Simulation-based Learning of Social Casework

In social work clinical training, role-playing client counseling scenarios enables students to practice practical counseling skills in a safe environment. In this training, one student assumes the role of the social worker while another person plays a standardized patient (SP), typically hiring a trained actor. However, due to time, cost, and ethical constraints, creating these scenarios for large classrooms presents challenges. The SWITCH prototype leverages open-source LLMs to simulate virtual clients across various scenarios (e.g., poverty, mental or physical illness, and discrimination). The LLM was developed using annotated case interview transcripts, allowing students to interact with virtual clients exhibiting varying levels of cooperation. The simulation exposes students to potentially upsetting and challenging events in a safe and controlled manner, resulting in harm reduction to themselves and service workers. The prototype is also flexible and transcends time and spatial constraints, allowing for many users to engage in the simulated environment with breaks in between.