A Human-Centred AI System for Multi-Actor Planning and Collaboration in Family Learning
Si Chen, Jingyi Xie, Yao Li, Ya-Fang Lin, He Zhang, Ge Wang, Gaojian Huang, Rui Yu, Ronald Anthony Metoyer, Ting Hua, Nitesh Chawla
TL;DR
The paper addresses the gap that existing AI tutors primarily support a single learner, neglecting the distributed planning and coordination inherent in home learning. It introduces ParPal, a human-centered, LLM-powered system that decomposes weekly learning goals into subtasks, allocates them across caregivers based on availability and expertise, and supplies caregiver-facing tutoring guidance with visibility into contributions. Through expert evaluation and a one-week field deployment with 11 families, the study reveals both benefits—increased coordination clarity and recognition of caregiving effort—and limitations of current LLMs in pedagogy, collaboration reasoning, and handling embodied constraints, underscoring the need for incorporating social cost and temporal planning into AI for education. The work positions multi-actor family learning as a meaningful testbed for advancing planning, adaptation, and pedagogical structure in next-generation AI systems and informs human-centered AI design for everyday learning contexts.
Abstract
Family learning takes place in everyday routines where children and caregivers read, practice, and develop new skills together. Despite growing interest in AI tutors, most existing systems are designed for single learners or classroom settings and do not address the distributed planning, coordination, and execution demands of learning at home. This paper introduces ParPal, a human-centred, LLM-powered system that supports multi-actor family learning by decomposing learning goals into actionable subtasks, allocating them across caregivers under realistic availability and expertise constraints, and providing caregiver-in-the-loop tutoring support with visibility into individual and collective contributions. Through expert evaluation of generated weekly learning plans and a one-week field deployment with 11 families, we identify systematic failure modes in current LLM-based planning, including misalignment with role expertise, unnecessary or costly collaboration, missing pedagogical learning trajectories, and physically or temporally infeasible tasks. While ParPal improves coordination clarity and recognition of caregiving effort, these findings expose fundamental limitations in how current LLMs operationalize pedagogical knowledge, reason about collaboration, and account for real-world, embodied constraints. We discuss implications for human-centred AI design and AI methodology, positioning multi-actor family learning as a critical testbed for advancing planning, adaptation, and pedagogical structure in next-generation AI systems.