Anticipated Acceptance of AI-Proactive Smart Sofas: A Value–Reactance Dual-Pathway Model

Proactive AI can make domestic products more responsive while raising concerns about unwanted intervention, control, and privacy. This study examines anticipated acceptance of AI-proactive smart sofas through a value–reactance framework. A theory-guided exploratory sequential mixed-methods design combinedinterviews with 10 prospective users and a scenario-based survey of 401 valid cases. PLS-SEM showed that perceived AI proactivity was positively associated with perceived usefulness and perceived intrusiveness, although the usefulness-related coefficient was substantially larger. Perceived usefulness was positively associated with AI trust and behavioral intention, whereas perceived intrusiveness was positively associated with psychological reactance, which was negatively associated with AI trust and behavioral intention. Privacy sensitivity positively moderated the intrusiveness–reactance association, although the interaction was small. Because participants evaluated a hypothetical product rather than interacting with an actual sofa, behavioral intention is interpreted as anticipated acceptance rather than observed adoption. The findings highlight user control, transparency, and privacy as design considerations.