Conversational AI for Social Worker Training

In collaboration with the University of Kentucky’s Center for Innovation in Population Health (IPH) and CreativeLink for Digital Health, the Center for Applied Artificial Intelligence (CAAI) is exploring the use of Large Language Models (LLMs) in social work training scenarios. LLMs show promising potential for generating synthetic personas to support teaching and simulation across various disciplines. This experimental project specifically focuses on enhancing the remote training experience for Child and Adolescent Needs and Strengths (CANS) certification.

This interdisciplinary initiative aims to boost engagement in recurring training by using templated modules that generate synthetic personas and conversational AI agents. By leveraging CAAI’s existing modules, the system simulates realistic training scenarios with dynamic, AI-powered interactions.

The system generates realistic client personas based on standardized training vignettes. a novel adapter training approach enables the model to effectively demonstrate the persona of a child or youth in care. Instead of passively reading a vignette and writing a response, trainees engage in real-time, conversational interactions with synthetic personas designed to mirror the types of real-world situations and assessments social workers may encounter in the field.