Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

GenAI for Social Work Field Education: Client Simulation with Real-Time Feedback

James Sungarda, Hongkai Liu, Zilong Zhou, Tien-Hsuan Wu, Johnson Chun-Sing Cheung, Ben Kao

TL;DR

This work introduces SWITCH, a GenAI-based trainer for social work field education that combines realistic client simulation, real-time counseling-skill classification, and a Motivational Interviewing (MI) progression system. A cognitive client model with static and dynamic fields governs agent behavior, while a skill-classifier feeds into an MI controller to modulate session stages; the system supports end-to-end training with objective feedback. The authors evaluate two classification approaches—in-context learning with retrieval and a fine-tuned BERT model—showing substantial performance gains over baselines ($Accuracy$ for ICL ≈ $0.92$–$0.94$ and for BERT ≈ $0.98$–$0.99$) and demonstrating SWITCH’s scalability and utility for supervisors. Deployed in a university course, SWITCH promises low-cost, consistent training that augments traditional field education and can be extended to richer modalities and feedback mechanisms in future work.

Abstract

Field education is the signature pedagogy of social work, yet providing timely and objective feedback during training is constrained by the availability of instructors and counseling clients. In this paper, we present SWITCH, the Social Work Interactive Training Chatbot. SWITCH integrates realistic client simulation, real-time counseling skill classification, and a Motivational Interviewing (MI) progression system into the training workflow. To model a client, SWITCH uses a cognitively grounded profile comprising static fields (e.g., background, beliefs) and dynamic fields (e.g., emotions, automatic thoughts, openness), allowing the agent's behavior to evolve throughout a session realistically. The skill classification module identifies the counseling skills from the user utterances, and feeds the result to the MI controller that regulates the MI stage transitions. To enhance classification accuracy, we study in-context learning with retrieval over annotated transcripts, and a fine-tuned BERT multi-label classifier. In the experiments, we demonstrated that both BERT-based approach and in-context learning outperforms the baseline with big margin. SWITCH thereby offers a scalable, low-cost, and consistent training workflow that complements field education, and allows supervisors to focus on higher-level mentorship.

GenAI for Social Work Field Education: Client Simulation with Real-Time Feedback

TL;DR

This work introduces SWITCH, a GenAI-based trainer for social work field education that combines realistic client simulation, real-time counseling-skill classification, and a Motivational Interviewing (MI) progression system. A cognitive client model with static and dynamic fields governs agent behavior, while a skill-classifier feeds into an MI controller to modulate session stages; the system supports end-to-end training with objective feedback. The authors evaluate two classification approaches—in-context learning with retrieval and a fine-tuned BERT model—showing substantial performance gains over baselines ( for ICL ≈ – and for BERT ≈ –) and demonstrating SWITCH’s scalability and utility for supervisors. Deployed in a university course, SWITCH promises low-cost, consistent training that augments traditional field education and can be extended to richer modalities and feedback mechanisms in future work.

Abstract

Field education is the signature pedagogy of social work, yet providing timely and objective feedback during training is constrained by the availability of instructors and counseling clients. In this paper, we present SWITCH, the Social Work Interactive Training Chatbot. SWITCH integrates realistic client simulation, real-time counseling skill classification, and a Motivational Interviewing (MI) progression system into the training workflow. To model a client, SWITCH uses a cognitively grounded profile comprising static fields (e.g., background, beliefs) and dynamic fields (e.g., emotions, automatic thoughts, openness), allowing the agent's behavior to evolve throughout a session realistically. The skill classification module identifies the counseling skills from the user utterances, and feeds the result to the MI controller that regulates the MI stage transitions. To enhance classification accuracy, we study in-context learning with retrieval over annotated transcripts, and a fine-tuned BERT multi-label classifier. In the experiments, we demonstrated that both BERT-based approach and in-context learning outperforms the baseline with big margin. SWITCH thereby offers a scalable, low-cost, and consistent training workflow that complements field education, and allows supervisors to focus on higher-level mentorship.
Paper Structure (18 sections, 5 equations, 2 figures, 7 tables)

This paper contains 18 sections, 5 equations, 2 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. AI-Powered Simulated Clients for Training
  4. In-Context Learning
  5. Counseling Skills Classification
  6. System Design and Architecture Work
  7. Motivational Interviewing (MI)
  8. Counseling Skill Classification Module
  9. Client Profile Modeling and Response Generation
  10. Enhancing Classification Accuracy
  11. Dataset Preparation
  12. In-Context Learning
  13. Finetuning BERT
  14. Experiments
  15. Dataset and Experiment Setup
  16. ...and 3 more sections

Figures (2)

  • Figure 1: Workflow of SWITCH
  • Figure 2: Distribution of Skill Counts per Utterance