Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Applying LLM-Powered Virtual Humans to Child Interviews in Child-Centered Design

Linshi Li, Hanlin Cai

TL;DR

This work establishes design guidelines for LLM-powered virtual humans to conduct child-centered interviews, standardizing multimodal interactions (color, facial features, voice, expressions, head movements, gestures) and prompting strategies. It introduces three human–AI workflows (LLM-Auto, LLM-Interview, LLM-Analyze) and demonstrates, via a within-subject study with 15 children (ages 6–12), that the LLM-Analyze workflow most effectively engages children, eliciting longer responses, higher user experience, and improved engagement. The approach combines structured VH design (Color Design Matrix, Baby Schema, expressive voice, and a Mancini–Pelachaud-inspired nonverbal framework) with a ChatGPT-based conversational interface using a three-part prompt (Background, Command, Interaction) and emotion-aware adjustments based on Plutchik’s wheel. Overall, the paper provides a concrete, scalable framework for child-friendly, AI-assisted interviews that can enhance data quality and ethical engagement in child-centered design.”

Abstract

In child-centered design, directly engaging children is crucial for deeply understanding their experiences. However, current research often prioritizes adult perspectives, as interviewing children involves unique challenges such as environmental sensitivities and the need for trust-building. AI-powered virtual humans (VHs) offer a promising approach to facilitate engaging and multimodal interactions with children. This study establishes key design guidelines for LLM-powered virtual humans tailored to child interviews, standardizing multimodal elements including color schemes, voice characteristics, facial features, expressions, head movements, and gestures. Using ChatGPT-based prompt engineering, we developed three distinct Human-AI workflows (LLM-Auto, LLM-Interview, and LLM-Analyze) and conducted a user study involving 15 children aged 6 to 12. The results indicated that the LLM-Analyze workflow outperformed the others by eliciting longer responses, achieving higher user experience ratings, and promoting more effective child engagement.

Applying LLM-Powered Virtual Humans to Child Interviews in Child-Centered Design

TL;DR

This work establishes design guidelines for LLM-powered virtual humans to conduct child-centered interviews, standardizing multimodal interactions (color, facial features, voice, expressions, head movements, gestures) and prompting strategies. It introduces three human–AI workflows (LLM-Auto, LLM-Interview, LLM-Analyze) and demonstrates, via a within-subject study with 15 children (ages 6–12), that the LLM-Analyze workflow most effectively engages children, eliciting longer responses, higher user experience, and improved engagement. The approach combines structured VH design (Color Design Matrix, Baby Schema, expressive voice, and a Mancini–Pelachaud-inspired nonverbal framework) with a ChatGPT-based conversational interface using a three-part prompt (Background, Command, Interaction) and emotion-aware adjustments based on Plutchik’s wheel. Overall, the paper provides a concrete, scalable framework for child-friendly, AI-assisted interviews that can enhance data quality and ethical engagement in child-centered design.”

Abstract

In child-centered design, directly engaging children is crucial for deeply understanding their experiences. However, current research often prioritizes adult perspectives, as interviewing children involves unique challenges such as environmental sensitivities and the need for trust-building. AI-powered virtual humans (VHs) offer a promising approach to facilitate engaging and multimodal interactions with children. This study establishes key design guidelines for LLM-powered virtual humans tailored to child interviews, standardizing multimodal elements including color schemes, voice characteristics, facial features, expressions, head movements, and gestures. Using ChatGPT-based prompt engineering, we developed three distinct Human-AI workflows (LLM-Auto, LLM-Interview, and LLM-Analyze) and conducted a user study involving 15 children aged 6 to 12. The results indicated that the LLM-Analyze workflow outperformed the others by eliciting longer responses, achieving higher user experience ratings, and promoting more effective child engagement.
Paper Structure (15 sections, 5 figures, 1 table)

This paper contains 15 sections, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Interviewing Children
  4. LLM-Powered Virtual Human Interview
  5. Multimodal Interactive VHs in Child Interviews
  6. Character and Scenario Design
  7. Color Scheme Design
  8. Facial Features Design
  9. Voice Effect Design
  10. Multimodal Emotional Expressions Design
  11. LLM-Powered Conversations for VHs
  12. ChatGPT for Conversations
  13. Human--AI Workflows for Child Interviews
  14. User Study
  15. Conclusion and Future Work

Figures (5)

  • Figure 1: The core design dimensions of LLM-powered virtual human in child interview scenarios.
  • Figure 2: Color Design Matrix and Basic Color Scheme for Virtual Humans in Child Interviews.
  • Figure 3: Virtual Human Prompt Generating Interface.
  • Figure 4: Optimizing the human--AI collaborative process design for each stage of child-centered interview research design.
  • Figure 5: (a) Comparison of Interview Duration and Response Length; (b) User Experience Rating Distribution.