Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Investigating AI in Peer Support via Multi-Module System-Driven Embodied Conversational Agents

Ruoyu Wen, Xiaoli Wu, Kunal Gupta, Simon Hoermann, Mark Billinghurst, Alaeddin Nassani, Dwain Allan, Thammathip Piumsomboon

TL;DR

The paper addresses the need for accessible, empathetic peer support for youth mental well-being amid a shortage of professionals. It introduces EmoCBT, a multi-module LLM-driven ECAs framework guided by CBT, incorporating detection, emotion recognition, deep CBT dialogue, crisis resources, and memory for context. Through a formative study and a user study with psychology practitioners, it demonstrates potential benefits in availability and structured support while highlighting concerns about empathy, trust, and workflow integration. The work argues for careful design, human oversight, and personalization to safely scale peer-based mental well-being support using embodied AI.

Abstract

Young people's mental well-being is a global concern, with peer support playing a key role in daily emotional regulation. Conversational agents are increasingly viewed as promising tools for delivering accessible, personalised peer support, particularly where professional counselling is limited. However, existing systems often suffer from rigid input formats, scripted responses, and limited emotional sensitivity. The emergence of large language models introduces new possibilities for generating flexible, context-aware, and empathetic responses. To explore how individuals with psychological training perceive such systems in peer support contexts, we developed an LLM-based multi-module system to drive embodied conversational agents informed by Cognitive Behavioral Therapy (CBT). In a user study (N=10), we qualitatively examined participants' perceptions, focusing on trust, response quality, workflow integration, and design opportunities for future mental well-being support systems.

Investigating AI in Peer Support via Multi-Module System-Driven Embodied Conversational Agents

TL;DR

The paper addresses the need for accessible, empathetic peer support for youth mental well-being amid a shortage of professionals. It introduces EmoCBT, a multi-module LLM-driven ECAs framework guided by CBT, incorporating detection, emotion recognition, deep CBT dialogue, crisis resources, and memory for context. Through a formative study and a user study with psychology practitioners, it demonstrates potential benefits in availability and structured support while highlighting concerns about empathy, trust, and workflow integration. The work argues for careful design, human oversight, and personalization to safely scale peer-based mental well-being support using embodied AI.

Abstract

Young people's mental well-being is a global concern, with peer support playing a key role in daily emotional regulation. Conversational agents are increasingly viewed as promising tools for delivering accessible, personalised peer support, particularly where professional counselling is limited. However, existing systems often suffer from rigid input formats, scripted responses, and limited emotional sensitivity. The emergence of large language models introduces new possibilities for generating flexible, context-aware, and empathetic responses. To explore how individuals with psychological training perceive such systems in peer support contexts, we developed an LLM-based multi-module system to drive embodied conversational agents informed by Cognitive Behavioral Therapy (CBT). In a user study (N=10), we qualitatively examined participants' perceptions, focusing on trust, response quality, workflow integration, and design opportunities for future mental well-being support systems.

Paper Structure

This paper contains 35 sections, 2 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Negative Emotions and Regulation in Young Adulthood
  4. Conversational Agents in Mental Well-being
  5. Large Language Model Empower Mental Well-being Conversational Agents
  6. Formative Study
  7. System Design
  8. Detection Module
  9. Empathetic Response Module
  10. Emotion Recognition Module
  11. Deep Conversation Module
  12. Crisis Intervention and Resource Provision
  13. Memory Module
  14. An Example User Case
  15. User Study
  16. ...and 20 more sections

Figures (2)

  • Figure 1: User Interface of the dialogue system: (A) Real-time monitoring screen of the human participant (blurred for anonymity), (B) Real-time monitoring screen of the Virtual Human, (C) Summary of the current conversation, (D) Recommended response, (E) Default response, (F) Chat history of the current conversation, (G) User input interface.
  • Figure 2: User Case: an example of a stressed first-year student talking with EmoCBT about her negative feelings in school. The figure shows the workflow of EmoCBT to support the user's emotional regulation.