Voice-Powered Assembly: Boosting Self-Efficacy in Older Adults with “Build2Race”
Abstract
Will a Voice Assistant (VA) compare well with a controlled-Voice Call (VC) in terms of assembly performance or resulting rated self-efficacy? This work is novel in extending the use of VAs to assembly scenarios for older adults, a context particularly relevant for those aging in place and seeking to maintain independence in everyday tasks. To address this question, we conducted an exploratory study with older adults ( N = 18) to investigate the effects of a VA prototype (“Build2Race”) compared to a controlled-VC system. Participants’ domain-specific and general self-efficacy scores improved after carrying out the assembly task, but those scores did not differ between the conditions. Improvements in assembly self-efficacy were better sustained in the VA (versus controlled-VC) condition after 1 month, but no similar advantage was observed for general self-efficacy. We hypothesize that a non-human agent may be good for task assembly self-efficacy because older adults feel like they are in charge of task performance when a VA is used. Based on the study results and interpretation, we offer guidelines for developing VAs for assembly task assistance.
Authors: Noah Zijie Qu, Mark Chignell, Jamy J. Li
Published in: ACM Transactions on Accessible Computing (2026)