Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Interaction Layer: An Exploration for Co-Designing User-LLM Interactions in Parental Wellbeing Support Systems

Sruthi Viswanathan, Seray Ibrahim, Ravi Shankar, Reuben Binns, Max Van Kleek, Petr Slovak

TL;DR

The paper investigates how co-design can address explainability and reliability challenges in LLM-based parental wellbeing support. It introduces NurtureBot and its Interaction Layer, developed through four iterative phases (v1–v3) and ARC co-design with parents to craft understand/control/improve interactions. The study demonstrates steadily rising usability (CUQ scores from 85.4 to 91.3) and provides empirical guidance on design principles for AI parenting tools, including localised resources, memory, and multimodal options. The findings highlight the value of human-centered AI in creating scalable, empathetic, and trustworthy digital support for families, with implications for broader wellbeing domains.

Abstract

Parenting brings emotional and physical challenges, from balancing work, childcare, and finances to coping with exhaustion and limited personal time. Yet, one in three parents never seek support. AI systems potentially offer stigma-free, accessible, and affordable solutions. Yet, user adoption often fails due to issues with explainability and reliability. To see if these issues could be solved using a co-design approach, we developed and tested NurtureBot, a wellbeing support assistant for new parents. 32 parents co-designed the system through Asynchronous Remote Communities method, identifying the key challenge as achieving a "successful chat." As part of co-design, parents role-played as NurtureBot, rewriting its dialogues to improve user understanding, control, and outcomes. The refined prototype, featuring an Interaction Layer, was evaluated by 32 initial and 46 new parents, showing improved user experience and usability, with final CUQ score of 91.3/100, demonstrating successful interaction patterns. Our process revealed useful interaction design lessons for effective AI parenting support.

The Interaction Layer: An Exploration for Co-Designing User-LLM Interactions in Parental Wellbeing Support Systems

TL;DR

The paper investigates how co-design can address explainability and reliability challenges in LLM-based parental wellbeing support. It introduces NurtureBot and its Interaction Layer, developed through four iterative phases (v1–v3) and ARC co-design with parents to craft understand/control/improve interactions. The study demonstrates steadily rising usability (CUQ scores from 85.4 to 91.3) and provides empirical guidance on design principles for AI parenting tools, including localised resources, memory, and multimodal options. The findings highlight the value of human-centered AI in creating scalable, empathetic, and trustworthy digital support for families, with implications for broader wellbeing domains.

Abstract

Parenting brings emotional and physical challenges, from balancing work, childcare, and finances to coping with exhaustion and limited personal time. Yet, one in three parents never seek support. AI systems potentially offer stigma-free, accessible, and affordable solutions. Yet, user adoption often fails due to issues with explainability and reliability. To see if these issues could be solved using a co-design approach, we developed and tested NurtureBot, a wellbeing support assistant for new parents. 32 parents co-designed the system through Asynchronous Remote Communities method, identifying the key challenge as achieving a "successful chat." As part of co-design, parents role-played as NurtureBot, rewriting its dialogues to improve user understanding, control, and outcomes. The refined prototype, featuring an Interaction Layer, was evaluated by 32 initial and 46 new parents, showing improved user experience and usability, with final CUQ score of 91.3/100, demonstrating successful interaction patterns. Our process revealed useful interaction design lessons for effective AI parenting support.

Paper Structure

This paper contains 51 sections, 12 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Study Flow Overview
  3. Related Work
  4. Supporting Family Wellbeing with Digital Tools
  5. Interacting with AI for Everyday Users
  6. Part I: The Initial Prototype: NurtureBot v1
  7. Origins and motivations for the initial prototype
  8. Designing the initial prototype
  9. NurtureBot v1 Testing and Analysis
  10. Participants and Method
  11. Data Collection and Analysis
  12. Findings of initial trial: Parent reactions to Nurture Bot v1
  13. Part II: ARC Co-Design Study
  14. ARC Activity One: Diagnosing and Prioritising Pain Points
  15. ARC Activity One: Method
  16. ...and 36 more sections

Figures (12)

  • Figure 1: Survey Questions for Each Day in NurtureBot v1, v2, and v3 technology trials
  • Figure 2: Example conversation with NurtureBot v1
  • Figure 3: Results from ARC Activity One voting by 30 participants, where key problems in interacting with NurtureBot were ranked using an Olympic points system. Complete Miro board including the other set-up phases participants went through before voting can be found in the supplementary materials.
  • Figure 4: A section from ARC Activity Two with 28 parents roleplaying as NurtureBot writing dialogues to better understand(left), control(centre), and improve(right) the features of empathetic chatting, wellbeing exercises, and parenting information lookup, and 30 parents returning to the Miro board to review and vote with heart stickers on the favourites among 189 dialogues generated. Complete Miro board including all dialogues and votes can be found in the supplementary materials.
  • Figure 5: This figure illustrates the interaction layer prompt architecture of NurtureBot v2, integrating user-generated dialogues from ARC as few-shot examples in the prompt.
  • ...and 7 more figures